The AIR Hurricane Model AIR Atlantic Tropical Cyclone Model V9.0

Submitted in Compliance with the 2006 Standards of the Florida Commission on Hurricane Loss Projection Methodology

February 2007

UNICEDE, CATMAP and CATRADER are registered trademarks of AIR Worldwide Corporation.

AIRWeather, ALERT, CLASIC/2 and CLAS are trademarks of AIR Worldwide Corporation.

Microsoft is a registered trademark of Microsoft Corporation.

If you have any questions regarding this document, contact: AIR Worldwide Corporation 131 Dartmouth Street Boston, Massachusetts 02116-7603 USA Tel: (617) 267-6645 Fax: (617) 267-8284

Model Submission Checklist

1. Please indicate by checking below that the following has been included in your submission to the Florida Commission on Hurricane Loss Projection Methodology.

Yes No Item 9 1. Letter to the Commission 9 a. Refers to the Expert Certification Forms and states that professionals having credentials and/or experience in the areas of meteorology, engineering, actuarial science, statistics, and computer science have reviewed the model for compliance with the Standards 9 b. States model is ready to be reviewed by the Professional Team 9 c. Any caveats to the above statements noted with a complete explanation 9 2. Summary statement of compliance with each individual Standard and the data and analyses required in the Disclosures and Forms 9 3. General description of any trade secrets the modeler intends to present to the Professional Team 9 4. Model Identification 9 5. 20 Bound Copies 9 6. 20 CD ROMs containing: 9 a. Submission text in PDF format 9 b. PDF file bookmarked and highlightable by Standard, Form, and section 9 c. Data file names include abbreviated name of modeler, Standards year, and Form name (when applicable) 9 d. Forms V-2, A-1, A3, A-4, A-5, A-6, A-7 and S-5 (for models submitted by modeling organizations which have not previously provided the Commission with this analysis) in PDF format 9 e. Forms V-2, A-1, A-3, A-4, A-5, A-6 and A-7 in Excel format 9 f. Form S-5 (for models submitted by modeling organizations which have not previously provided the Commission with this analysis) in ASCII format 9 7. Table of Contents 9 8. Materials consecutively numbered from beginning to end starting with the first page (including cover) using a single numbering system 9 9. All tables, graphs, and other non-text items specifically listed in Table of Contents 9 10. All tables, graphs, and other non-text items clearly labeled with abbreviations defined 9 11. Standards, Disclosures, and Forms in italics, modeler responses in non-italics 9 12. Graphs accompanied by legends and labels for all elements 9 13. All units of measurement clearly identified with appropriate units used 9 14. Hard copy of all Forms included except Forms A-1 and S-5

2. Explanation of “No” responses indicated above. (Attach additional pages if needed.)

A completed Form S-5 (formerly Form F) was provided to the Commission in 2003.

Atlantic Tropical Cyclone Model V9.0 February 26, 2007 Model Name Modeler Signature Date

- This page intentionally left blank -

Florida Commission on Hurricane Loss Projection Methodology

Model Identification

Name of Model and Version: Atlantic Tropical Cyclone Model V9.0 Program: CLASIC/2 V9

Name of Modeling Company: AIR Worldwide Corporation

Street Address: 131 Dartmouth Street

City, State, ZIP Code: Boston, MA 02116

Mailing Address, if different from above:

Contact Person: Connie Kang

Phone Number: 617-267-6645 Fax Number: 617-267-8284

E-mail Address: [email protected]

Date: February 26, 2007

- This page intentionally left blank -

Trade Secret Information to be Presented to the Professional Team in Connection with the Acceptability Process

G-1 Computer model code, databases and files relevant to the submission.

G-3 Visual representation of the reasonableness of the population-weighted ZIP Code centroids. ZIP Code validation and update process document.

M-1 Updates to the historical storm set.

M-2 Methods for depicting all modeled hurricane characteristics and goodness-of-fit tests. Comparison of observed and modeled hurricane parameters and wind speeds.

M-3 Documentation of wind speed validation.

M-4 Documentation of methodology for smoothing landfalls and goodness-of-fit tests.

M-5 Maps depicting land friction effects. Comparisons of observed and modeled wind speeds showing how winds are spatially distributed.

M-6 Sensitivity analyses.

V-1 Original client data, samples of vulnerability functions, modifications to the vulnerability functions due to building codes, computer code showing minimum wind speed at which damage occurs.

V-2 Source data and methodology used to reflect impact of mitigation factors; Form V-3.

A-4 Demand Surge Analysis.

A-5 Client Data Processing Procedures.

S-2 The results of sensitivity studies for hurricane model losses and input variables.

S-3 The results of uncertainty studies for hurricane model losses and input variables.

S-4 Graphs assessing the contribution to error in loss costs by county.

Computer Standards – All software, documentation, specifications, and procedures as specified in the audit section of each standard.

7 Table of Contents

Table of Contents

Model Submission Checklist...... 3 Model Identification ...... 5 Trade Secret Information to be Presented to the Professional Team in Connection with the Acceptability Process ...... 7 Computer Standards – All software, documentation, specifications, and procedures as...... 7 specified in the audit section of each standard...... 7 Table of Contents ...... 8 List of Figures...... 10 List of Tables...... 12 2006 General Standards...... 13 G-1 Scope of the Computer Model and Its Implementation...... 13 G-2 Qualifications of Modeler Personnel and Consultants ...... 24 G-3 Risk Location...... 41 G-4 Independence of Model Components...... 43 Form G-1: General Standards Expert Certification...... 44 Form G-2: Meteorological Standards Expert Certification...... 45 Form G-3: Vulnerability Standards Expert Certification ...... 46 Form G-4: Actuarial Standards Expert Certification...... 47 Form G-5: Statistical Standards Expert Certification...... 48 Form G-6: Computer Standards Expert Certification ...... 49 2006 Meteorological Standards...... 50 M-1 Base Hurricane Storm Set ...... 50 M-2 Hurricane Characteristics...... 51 M-3 Landfall Intensity...... 58 M-4 Hurricane Probabilities...... 59 M-5 Land Friction and Weakening ...... 62 M-6 Logical Relationships of Hurricane Characteristics...... 67 Form M-1: Annual Occurrence Rates ...... 68 Form M-2: Maps of Maximum Winds ...... 70 Form M-3: Radius of Maximum Winds...... 72 Official Hurricane Set...... 74 2006 Vulnerability Standards...... 77 V-1 Derivation of Vulnerability Functions...... 77 V-2 Mitigation Measures...... 88 Form V-1: One Hypothetical Event ...... 90 Form V-2: Mitigation Measures – Range of Changes in Damage ...... 93 Form V-3: Mitigation Measures – Mean Damage Ratio Trade Secret List Item...... 99 2006 Actuarial Standards ...... 100 A-1 Modeled Loss Costs ...... 100 A-2 Underwriting Assumptions...... 101 A-3 Loss Cost Projections ...... 104

Table of Contents

A-4 Demand Surge...... 105 A-5 User Inputs...... 107 A-6 Logical Relationship to Risk ...... 110 A-7 Deductibles and Policy Limits...... 119 A-8 Contents...... 123 A-9 Additional Living Expense (ALE) ...... 126 A-10 Output Ranges...... 129 Form A-1: Loss Costs...... 132 Form A-2: Zero Deductible Loss Costs by ZIP Code ...... 135 Form A-3: Base Hurricane Storm Set Average Annual Zero Deductible Statewide Loss Costs ...... 137 Form A-4: Hurricane Andrew Percent of Losses...... 139 Form A-5: Distribution of Hurricanes by Size...... 143 Form A-6: Output Ranges ...... 146 Form A-7: Percentage Change In Output Ranges ...... 185 Form A-8: Percentage Change in Output Ranges by County...... 188 2006 Statistical Standards...... 192 S-1 Modeled Results and Goodness-of-Fit ...... 192 S-2 Sensitivity Analysis for Model Output...... 201 S-3 Uncertainty Analysis for Model Output...... 204 S-4 County Level Aggregation ...... 207 S-5 Replication of Known Hurricane Losses...... 208 S-6 Comparison of Projected Hurricane Loss Costs...... 213 Form S-1: Probability of Florida Landfalling Hurricanes per Year...... 215 Form S-2: Probable Maximum Loss (PML) ...... 216 Form S-3: Five Validation Comparisons...... 217 Form S-4: Average Annual Zero Deductible Statewide Loss Costs—Historical versus Modeled...... 222 2006 Computer Standards ...... 223 C-1 Documentation ...... 223 C-2 Requirements...... 224 C-3 Model Architecture and Component Design...... 225 C-4 Implementation...... 226 C-5 Verification...... 228 C-6 Model Maintenance and Revision...... 230 C-7 Security...... 232 Attachment A: Project Information Assumption Form...... 234 Attachment B: Model Evaluation and Curriculum Vitae of Joseph Minor, Ph.D., P.E...... 244 Attachment C: Model Evaluation and Curriculum Vitae of Narges Pourghasemi, M.S...... 256 Attachment D: U.S. Hurricane Individual Risk Methodology...... 274 Attachment E: Remapped ZIP Codes...... 320

List of Figures

Figure 1: AIR Tropical Cyclone Model Components ...... 14 Figure 2: Storm Track Generation...... 17 Figure 3: Flowchart of the AIR Model...... 18 Figure 4: AIR Hurricane Model Workflow...... 38 Figure 5: Hurricane Frequency by Coastal Segment, 1900-2006 ...... 55 Figure 6: Number of Landfalls Per Year, Modeled vs Actual, 1900-2006 ...... 55 Figure 7: Florida Panhandle Storm: Wind speeds and Friction Factors for Simulated Storm.63 Figure 8: South Florida Storm: Wind Speeds and Friction Factors for Simulated Storm...... 63 Figure 9: Actual and modeled filling rates for 2004 Hurricanes ...... 65 Figure 9a: Observed and modeled wind speeds for Hurricanes Dennis and Charley……...…66 Figure 10: State of Florida and Neighboring States By Region...... 69 Figure 11: ZIP Code Level Maximum Wind Speed  Historical 107-year Period...... 70 Figure 12: ZIP Code Level Maximum Wind Speed  Simulated 100-year Return Wind Speeds...... 71 Figure 13: State of Florida and Neighboring States ...... 73 Figure 14: Derivation and Implementation of AIR Vulnerability Functions...... 79 Figure 15: Actual and Simulated Damage Ratios vs Wind Speed: Coverage A - Single Company, Single Storm...... 81 Figure 16: Actual and Simulated Damage Ratios vs Wind Speed: Coverage C - Single Company, Single Storm...... 81 Figure 17: Actual and Simulated Damage Ratios vs Wind Speed: Coverage D - Single Company, Single Storm...... 82 Figure 18: Actual and Simulated Damage Ratios vs Wind Speed: Mobile Homes - Single Company, Single Storm...... 82 Figure 19: Actual and Simulated Damage Ratios vs Wind Speed: Frame - Single Company, Single Storm ...... 83 Figure 20: Actual and Simulated Damage Ratios vs Wind Speed: Masonry - Single Company, Single Storm ...... 83 Figure 21: Total Loss Percentages by Wind Speed...... 92 Figure 22: Actual and Modeled Damage Ratios vs Wind Speed, Coverage A...... 112 Figure 23: Actual and Modeled Damage Ratios vs Wind Speed, Coverage C ...... 113 Figure 24: Actual and Modeled Damage Ratios vs Wind Speed, Coverage D...... 114 Figure 25: Actual and Modeled Damage Ratios vs Wind Speed, Mobile Home...... 115 Figure 26: Actual and Modeled Damage Ratios vs Wind Speed, Frame...... 116 Figure 27: Actual and Modeled Damage Ratios vs Wind Speed, Masonry...... 117 Figure 28: Probability Distribution around the Mean Damage Ratio ...... 119 Figure 29: Contents Vulnerability Relationship...... 123 Figure 30: Relationship of Cov C to Cov A Loss Cost ...... 124 Figure 31: Actual and Modeled Losses, Coverage C...... 125 Figure 32: Time Element Vulnerability Relationships...... 126 Figure 33: Relationship of Cov D to Cov A...... 127 Figure 34: Actual and Modeled Losses, Coverage D...... 128 Figure 35: Loss Costs by ZIP Code for Owners Wood Frame, Zero Deductible ...... 135

List of Figures

Figure 36: Loss Costs by ZIP Code for Owners Masonry, Zero Deductible ...... 136 Figure 37: Loss Costs by ZIP Code for Mobile Home, Zero Deductible...... 136 Figure 38: Percentage of Losses by ZIP Code for Hurricane Andrew...... 139 Figure 39: Graphical comparison of the current submission Return Times to the prior year’s submission Return Times ...... 144 Figure 40: State of Florida by North/Central/South Regions...... 186 Figure 41: State of Florida by Coastal/Inland Counties ...... 186 Figure 42: Percentage Change in Weighted Average Loss Costs by County – Frame Owners ...... 188 Figure 43: Percentage Change in Weighted Average Loss Costs by County – Masonry Owners...... 189 Figure 44: Percentage Change in Weighted Average Loss Costs by County – Mobile Homes ...... 189 Figure 45: Percentage Change in Weighted Average Loss Costs by County – Frame Renters ...... 190 Figure 46: Percentage Change in Weighted Average Loss Costs by County – Masonry Renters ...... 190 Figure 47: Percentage Change in Weighted Average Loss Costs by County – Frame Condos ...... 191 Figure 48: Percentage Change in Weighted Average Loss Costs by County – Masonry Condos ...... 191 Figure 49: Hurricane Andrew Wind Speed Contour Map...... 195 Figure 50: Validation of Wind Speeds for Hurricane Andrew...... 196 Figure 51: Empirical Annual Landfall and Fitted Distributions ...... 197 Figure 52: Historical and Simulated Hurricanes Landfalling in SE Florida ...... 198 Figure 53: Historical and Simulated Central Pressure Distributions...... 199 Figure 54: Sample Damage Ratio Comparison...... 200 Figure 55: Scatter Plot of Comparison #1 ...... 219 Figure 56: Scatter Plot of Comparison #2 ...... 219 Figure 57: Scatter Plot of Comparison #3 ...... 220 Figure 58: Scatter Plot of Comparison #4 ...... 220 Figure 59: Scatter Plot of Comparison #5 ...... 221

List of Tables

List of Tables Table 1: AIR Modeler Personnel and Independent Experts...... 24 Table 2: Primary Databases...... 61 Table 3: Modeled Annual Occurrence Rates...... 68 Table 4: Range of Rmax by Central Pressure...... 72 Table 5: Residential Construction Types in the AIR Model ...... 86 Table 6: Mitigation Measures in the AIR Model ...... 89 Table 7: Modification Factors to Vulnerability Functions...... 96 Table 8: Sample Calculation for Determining Depreciation and ACV Losses...... 102 Table 9: Sample Calculation for Determining Property Value and Guaranteed Replacement Cost Losses...... 103 Table 10: Comparison between Actual and Modeled Losses for Different Coverages ...... 111 Table 11: Calculating Losses Net of Deductibles ...... 121 Table 12: County Weighted Average Loss Costs for Masonry and Frame...... 130 Table 13: Statewide Impact of Model Updates on Weighted Average Loss Costs...... 130 Table 14: Form A-3 - Monetary Contribution to the Average Annual Personal Residential Zero Deductible Statewide Loss Costs...... 138 Table 15: Hurricane Andrew Percent of Losses...... 141 Table 16: List of Zero-weighted ZIP Codes...... 147 Table 17: Sources of Historical Data...... 195 Table 18: Actual vs Modeled Losses for Eight Storms and Five Companies...... 209 Table 19: Actual vs Modeled Losses by Coverage for Eight Storms and Three Companies 210 Table 20: Actual vs Modeled Losses by Construction Type for Eight Storms and Four Companies ...... 211 Table 21: Actual vs Modeled Losses by County for One Company -- Hurricane Bonnie.....212

Standard G-1: Scope of the Computer Model and Its Implementation

Florida Commission on Hurricane Loss Projection Methodology

2006 General Standards

G-1 Scope of the Computer Model and Its Implementation

The computer model shall project loss costs for personal lines residential property from hurricane events.

The AIR hurricane model projects loss costs for personal lines residential property from hurricane events.

Disclosures

1. Specify the model and program version number.

The current AIR hurricane model being submitted to the Commission for approval is: Atlantic Tropical Cyclone Model V9.0; Program: CLASIC/2™ V9.

2. Provide a concise, technical description of the model including each major component of the model used to produce personal lines residential loss costs in the State of Florida. Describe the theoretical basis of the model and include a description of the methodology, particularly the wind components, the damage components, and the insured loss components used in the model. The description should be complete and not reference unpublished work.

Introduction Standard actuarial techniques rely on data on past losses to project future losses. But the scarcity of historical loss data resulting from the infrequency of these events makes standard actuarial techniques of loss estimation inappropriate for catastrophe losses. Furthermore, the usefulness of the loss data that does exist is limited because of the constantly changing landscape of insured properties. Property values change, along with the costs of repair and replacement. Building materials and designs change, and new structures may be more or less vulnerable to catastrophe events than were the old ones. New properties continue to be built in areas of high hazard. Therefore, the limited loss information that is available is not suitable for directly estimating future losses.

AIR Worldwide Corporation (AIR) was the first company to develop probabilistic catastrophe modeling as an alternative to the standard actuarial or “rule of thumb” approaches on which insurance companies had to rely for the estimation of potential

Standard G-1: Scope of the Computer Model and Its Implementation catastrophe losses. In 1987, AIR introduced to the insurance industry a modeling methodology based on simulation techniques and mathematical approaches long-accepted in a wide variety of scientific disciplines. Since the inception of this new approach, the AIR hurricane model has undergone a comprehensive process of refinement, enhancement, validation, and review.

The figure below illustrates the component parts of the AIR model (gray boxes). Each component represents both the ongoing efforts of the research scientists and engineers who are responsible for its design and the computer processes that occur as the simulations are run.

Figure 1: AIR Tropical Cyclone Model Components

Methodology The AIR research team collects the available scientific data pertaining to the meteorological variables critical to the characterization of hurricanes and therefore to the simulation process. These primary model variables include landfall location, central pressure, radius of maximum winds, forward speed, and track direction. Data sources used in the development of the AIR hurricane model include the most complete databases available from various agencies of the National Weather Service, including the National Hurricane Center. All data is cross-verified. If data from different sources conflict, a detailed analysis and the use of expert judgment is applied to prepare the data for modeling purposes.

After the rigorous data analysis, AIR researchers develop probability distributions for each of the variables, testing them for goodness-of-fit and robustness. The selection and subsequent refinement of these distributions are based not only on the expert application of statistical techniques, but also on well-established scientific principles and an understanding of how hurricanes behave.

These probability distributions are then used to produce a large catalog of simulated events. By sampling from the various probability distributions, the model generates simulated “years” of event activity. A simulated year in this context represents a hypothetical year of hurricane experience that could happen in the current year. The AIR model allows for the possibility of multiple events occurring within a single year. That is, each simulated year may have no, one, or multiple hurricanes, just as might be observed in an actual year. Many thousands of these scenario years are generated to produce the

Standard G-1: Scope of the Computer Model and Its Implementation complete and stable range of potential annual experience of tropical cyclone activity. The pattern and distribution of the simulated years approximates the pattern of historical and future years because their derivation is based on a scientific extrapolation of actual historical data.

Once values for each of the important meteorological characteristics have been stochastically assigned, each simulated storm is propagated along its track. A complete time profile is estimated for each geographical location affected by the storm. Based on the wind profile, damages are estimated at each location for different types of structures. Finally, policy conditions are applied to estimate the insured losses resulting from each event.

As opposed to purely deterministic simulation models, probabilistic simulation models enable the estimation of the complete probability distribution of losses from hurricanes. Once this probability distribution is estimated, hurricane loss costs by county, five digit ZIP Code and rating territory can be derived.

Event Generation Component This first module of the AIR tropical cyclone model is used to create the stochastic storm catalog. More than one hundred years of historical data on the frequency of hurricanes and their meteorological characteristics were used to fit statistical distributions for each parameter. By stochastically drawing from these distributions, the fundamental characteristics of each simulated storm are generated. Again, these parameters include landfall location and subsequent track, and the intensity variables of central pressure, radius of maximum winds, and forward speed. The result is a large, representative catalog of potential events.

Landfall Location. In the AIR model, there are 3,100 potential landfall points, one at each nautical mile of smoothed coastline from Texas to Maine. To estimate the probability of a hurricane occurring at each of these points, a cumulative distribution of landfall locations is developed for the fifty nautical mile lengths of coastline as described below.

The coastline is first smoothed for irregularities such as inlets and bays. The actual number of hurricane occurrences is then tabulated for smoothed 50 nautical mile segments. The actual number of occurrences for each segment is then smoothed by setting it equal to the weighted average of 11 successive data points centered on that segment.

This smoothing technique was selected because it has been utilized in other climatological studies and because it maintains areas of high and low frequency while smoothing out variations due to limitations on completeness in the historical record and the coastline orientation.

Bypassing Storms. The AIR model generates bypassing as well as landfalling storms through a track generation procedure that follows each simulated hurricane

Standard G-1: Scope of the Computer Model and Its Implementation

from the time of its inception until it dissipates. A by-passer is a hurricane that does not make landfall but causes damaging winds over land.

Meteorological Standards. The simulated frequency of bypassing hurricanes is consistent with the frequency observed historically over the period 1900-2006.

Meteorological Characteristics. Probability distributions are estimated for central pressure and forward speed for 31 one hundred nautical mile segments of coastline. Separate distributions are estimated for each of these segments because the likely range and probabilities of values within each range for these variables depend upon geographic location and, in particular, latitude. For example, intense hurricanes are more likely to occur in southern latitudes where the water is warmer. Storms affecting the coast in northern latitudes tend to be larger and faster moving on average. Radius of maximum winds is represented using a regression model that relates the mean radius to central pressure and latitude.

Storm Heading at Landfall. Landfall angle is measured clockwise (+) or counterclockwise (-) with 0 representing due North. Separate distributions for storm heading at landfall are estimated for segments of coastline that are variable in length. The length of each segment is governed by the coastal orientation of that segment. Because certain coastal segments seem to be characterized by bimodal storm headings, the AIR model uses a mixture of normal distributions. The resulting distributions are bounded based on the historical record and the coastline orientation.

Storm Track. The methodology used to generate storm tracks is based on the information available in the HURDAT database for the period 1900-2001. This database provides track information for more than 900 North Atlantic storms at 6- hour time intervals. Time series techniques have been used to determine the dependence structure present in key model variables from one time period to the next. Time series models that describe the dependence in the historical data are used in the generation of simulated tracks. For example, a first-order Markov model with transition probabilities estimated from the historical data is used to generate changes in track direction of simulated storms. An illustration of the procedure is given in Figure 2 below. The storm tracks generated using this approach are realistic and resemble storm tracks that have been observed for historical storms.

Standard G-1: Scope of the Computer Model and Its Implementation

Figure 2: Storm Track Generation

AIR uses the event generation component of the model to simulate 50,000 years of potential hurricane activity.

Wind Field Generation Component Once the model probabilistically generates the hurricane’s meteorological characteristics, it simulates the storm’s movement along its track. A complete time profile of wind speeds is developed for each location affected by the storm, thus capturing the effect of duration of wind on structures as well as peak wind speed. Calculations of local intensity take into account the effects of the asymmetric nature of the hurricane wind field, storm filling over land, surface roughness, and radial variation of wind speeds from the radius of maximum winds.

Damage Calculation Component AIR scientists and engineers have developed damage functions that describe the interaction between buildings, both their structural and nonstructural components, and their contents, and the local intensity to which they are exposed. These functions relate the mean damage level as well as the variability of damage to the measure of intensity at each location. Because different structural types will experience different degrees of damage, the damage functions vary according to construction class and occupancy. The AIR model estimates a complete distribution around the mean level of damage for each local intensity and each structural type. Losses are calculated by applying the appropriate damage function to the replacement value of the insured property.

The AIR damageability relationships incorporate the results of well-documented engineering studies, damage surveys, and analyses of available loss data. As part of the ongoing process of refinement and validation of these functions, AIR engineers have performed post-disaster field surveys for every landfalling hurricane since Hugo in 1989. In addition, actual claims data from recent hurricanes, supplied to AIR by client companies, have been extensively analyzed.

Standard G-1: Scope of the Computer Model and Its Implementation

Insured Loss Calculation Component In this last component of the AIR model, insured losses are calculated by applying the policy conditions to the total damage estimates. Policy conditions may include deductibles by coverage, site-specific or blanket deductibles, coverage limits and sublimits, loss triggers, coinsurance, attachment points and limits for single or multiple location policies, and risk specific reinsurance terms.

3. Provide a flow diagram that illustrates interactions among major model components.

Figure 3: Flowchart of the AIR Model

Begin Simulation

Begin New A Begin New Year Start Hurricane Location or Zip Code

Generate Annual Generate Latitude Calculate Wind Yes Number of and Longitude Speed Hurricane Coordinates of Occurrences Landfall Location

Exposure Database Generate Values for Severity Yes Any Variables (Po, R, T, A) Hurricanes? Begin New Risk

at Location

No Yes Calculate Calculate Maximum Wind Incremental Yes Speed Simulate Damage Factor Another Year? Yes

Begin New Hour More Risks at Location? Yes

No Calculate Total Hurricane No Damage Still Active? Store No Results A

Another Update Hurricane Hurricane This No Location and More Year? Intensity Locations?

Output Summarize Results Losses

Standard G-1: Scope of the Computer Model and Its Implementation

4. Provide a comprehensive list of complete references pertinent to the submission by Standard grouping, according to professional citation standards.

The AIR hurricane simulation model is based on well-founded, widely accepted, and state- of-the-art meteorological knowledge and structural engineering expertise. The AIR wind field model follows the Standard Project Hurricane wind field model and the more recent work of Georgiou and others. The damageability component incorporates the work of engineers at Texas Tech, Texas A&M, and Clemson University, among others. The Monte Carlo simulation method is a widely accepted statistical technique used in many different disciplines including meteorology and engineering. Since the AIR model has been developed and refined over a period of 20 years, literally hundreds of source documents have been reviewed and incorporated. Below are listed the primary representative documents describing the meteorological components, damageability components, and the statistical techniques.

Representative Meteorological References Batts, M. E., Cordes, M. R., Russell, C. R., Shaver, J. R., and Simiu, E. Hurricane Wind speeds in the United States, National Bureau of Standards, Report Number BSS- 124, U.S. Department of Commerce, May, 1980. Cry, George W., Tropical Cyclones of the North Atlantic Ocean: tracks and frequencies of hurricane and tropical storms, 1871-1963, U.S. Department of Commerce, Weather Bureau, 1965. Dendrou, Sergios A., Moore, Charles I., and Taylor, Robert S., Coastal Flooding Hurricane Storm Surge Model, Camp Dresser & McKee for FEMA, June 1988. Friedman, Don G., Computer Simulation in Natural Hazard Assessment. University of Colorado: Program on Technology, Environment and Man. Monograph # NSF-R- A-E 15-002, 1975. ESDU International Ltd, ESDU engineering sciences data, Wind engineering. Wind speeds and turbulence, Vols. 1a, 1b, 1994. Georgiou, P.N. Design Wind Speeds in Tropical Cyclone-Prone Regions. Boundary Layer Wind Tunnel Laboratory, Research Report # BLWT-2- 1985. Ho, Francis P., Schwerdt, Richard W., and Goodyear, Hugo V., Some Climatological Characteristics of Hurricanes and Tropical Storms, Gulf and East Coast of the United States, U.S. Department of Commerce. National Weather Service. NOAA Technical Report HW515. May 1975. Ho, Francis P., Su, James C., Hanevich, Karen L., Smith, Rebecca J., and Richards, Frank P., Hurricane Climatology For the Atlantic and Gulf Coasts of the United States, U.S. Department of Commerce. National Weather Service. NOAA Technical Report EmW-84-E- 1589. April 1987. Jelesnianski, Chester P., SPLASH (Special Program to List Amplitudes of Surges From Hurricanes), Department of Commerce, National Weather Service, NOAA Technical Memorandum NWS TDL-46. April 1972. Kaplan, J., and DeMaria, M., “A Simple Empirical Model for Predicting the Decay of Tropical Cyclone Winds After Landfall.” Journal of Applied Meteorology, 1995.

Standard G-1: Scope of the Computer Model and Its Implementation

Neumann, Charles J., The National Hurricane Center Risk Analysis Program.” U.S. Department of Commerce. National Hurricane Center, NOAA Technical Memorandum NWS NHC 38. Reprinted August 1991. North Atlantic Basin Storm Database (HURDAT), 1851-2006. Powell M.D. and Houston S.H., Hurricane Andrew's Landfall in South Florida. Part II: Surface Wind Fields and Potential Real-Time Applications, Weather and Forecasting, pages 329-349, September 1996. Schwerdt, Richard W., Ho, Francis P. and Watkins, Roger R., Meteorological Criteria for Standard Project Hurricane and Probable Maximum Hurricane Wind fields. Gulf and East Coasts of the United States. U.S. Department of Commerce. National Oceanic and Atmospheric Administration, NOAA Technical Report NWS 23, September 1979. Simiu, E. and Scanlan, R. H. , Wind Effects on Structures – Fundamentals and Applications to Design, Wiley Interscience, 1996. Stull, R. B. Meteorology for Scientists and Engineers, West Publishing Company, 1995. Vickery, P.J., and Twisdale, L.A. “Wind-Field and Filling Models for Hurricane Wind- Speed Predictions” Journal of Structural Engineering, 1995.

Representative References for Damage Estimation ASCE 7-98, “Minimum Design Loads for Buildings and Other Structures”. American Society of Civil Engineers, New York, NY, 2000. Bhinderwala, S., “Insurance Loss Analysis of Single Family Dwellings Damaged in Hurricane Andrew,” Master’s Thesis, Clemson University, 1995. FEMA-1992, “Building Performance: Hurricane Andrew in Florida, Observations, Recommendations and Technical Guidance.” FEMA and Federal Insurance Administration, December 21, 1992. FEMA-1997, “Building Performance Assessment: Hurricane Fran in North Carolina, Observations, Recommendations and Technical Guidance,” FEMA, Mitigation Directorate, March 19, 1997. Cermark. J.E., “Wind Tunnel Testing of Structures”. Journal of Engineering Mechanics Division, ASCE, Vol. 106, No. EM6, December, 1980. Cook, N.J., The Designers Guide to Wind Loading of Building Structures. Building Research Establishment Report. Butterworth, London, England, 1985. Crandell J. H., “Statistical Assessment of construction characteristics and performance of homes in Andrew and Opal,” Journal of Wind Engineering and Industrial Aerodynamics, 77 and 78, 1998, pages 695-701. Cross, Mark, L., and Thorton, John, H., “The Probable Effect of an Extreme Hurricane on Texas Catastrophe Plan Insurers”. The Journal of Risk and Insurance, 1983. FEMA 488 (2005), Mitigation Assessment Team Report, Hurricane Charley in Florida: Observation, Recommendations, and Technical Guidance. Friedman, D.G., “Assessing Hurricane Impact on Human Settlements.” Ministry of Human Settlements and Public Works, Government of Mexico: Symposium on Hurricane- Floods: Their Effect on Human Settlements: La Pax, Baja California Sur, Mexico, November 24-27, 1981. Friedman. D.G. “Natural Hazard Risk Assessment for an Insurance Program”, The Geneva Papers on Risk and Insurance, International Association for the Study of Insurance

Standard G-1: Scope of the Computer Model and Its Implementation

Economics (Geneva Association), Geneva, Switzerland, Volume I, Number 30, January, pages 57-128, 1984. Friedman, Don G., Computer Simulation in Natural Hazard Assessment. University of Colorado: Program on Technology, Environment and Man. Monograph # NSF-R- A-E 15-002, 1975. “Hurricanes of 1992”, Proceedings of Hurricanes of 1992, Andrew and Iniki One Year Later, ASCE, Miami, Fl, 1993. “Hurricane Camille - August 1969: A Survey of Structural Damage along the Mississippi Gulf Coast,” National Bureau of Standards report, December 4, 1970. Khanduri, A. C., “Catastrophe Modeling and Windstorm Loss Mitigation”, Proc. Eleventh International Conference on Wind Engineering, Lubbock, Texas, June 2-5, 2003. Kareem, Ahsan, Hurricane Alicia: One Year Later, American Society of Civil Engineers, 1984. Liu, Henry, Wind Engineering- A Handbook for Structural Engineers,Prentice Hall, 1991. Marshall, R.D., Fastest Wind Speeds in Hurricane Alicia. Technical Note 1197, NTIS # PB84220771, 1984. Mehta, K.C., Minor, J.E., and Reinhold, T.A., “Wind Speed-Damage Correlation in Hurricane Frederick”. Journal of the Structural Division, ASCE, Vol. 109, Number 1, Paper No. 17626, January, 1983. Mehta, K.C, Cheshire, R.H., and McDonald, J.R., “Wind Resistance Categorization of Buildings for Insurance.” Journal of Wind Engineering and Industrial Aerodynamics, 1992. Mehta, K.C., Smith, D.A., and Cheshire, R.H., “Knowledge-based System for wind damage to buildings.” Proceedings of Hurricanes of 1992, Andrew and Iniki One Year Later, ASCE, Miami, Fl, 1993 Minor, J.E. and Beason, W.L. “Window Glass Failure in Windstorms”. Journal of the Structural Division, ASCE, Volume 101, No. ST1, Proceedings Paper 11834, January, 1976. Minor, J.E. and Mehta, K.C. and McDonald, J.R. “Wind Damage Observations and Implications”. Journal of the Structural Division, ASCE, Vol. 105, No ST11, November, 1979. Minor, J.E., Mehta, K.C., and McDonald, J.R., “Failures of Structures Due to Extreme Winds”, Journal of the Structural Division, ASCE, Vol. 98, No. ST11, 1972. Rosen H. J, Construction Materials for Architecture, John Wiley and Sons, 1985. Savage, Ruldoph P., Baker, Jay, Golden, H., Kareem, Ahsan, and Manning, Billy R., Hurricane Alicia, Galveston Houston, Texas August 17-18, 1983. U.S. Department of Commerce, NTIS# PB84-237056, 1984. Sill, B.L., Reinhold, T.A. and Kozlowski, R.T., “Analysis of Storm Damage Factors for Low Rise Structures,” Journal of Performance of Constructed Facilities, Vol. 11, No. 4, 1997 Simiu, E., and Scanlan, R.H., Wind Effects on Structures. Wiley-Interscience, 1996. Smith T.L. and McDonald J.R., “Preliminary Design Guidelines for Wind-Resistant Roofs on Essential Facilities,” Proceedings of the 7th U.S. National Conference on Wind Engineering, Vol II, Baltimore, MD, 1997. Sparks, P.R., “The Response of Low Rise Non-Engineered Structures to Extreme Wind Conditions.” Journal of Wind Engineering and Industrial Aerodynamics, 1986.

Standard G-1: Scope of the Computer Model and Its Implementation

Sparks, P.R., “Wind Conditions in Hurricane Hugo and Their Effect on Buildings in Coastal South Carolina.” Journal of Coastal Research, 1991. Sparks, P.R., Hessig, M.L., Murden, J.A., and Sill, B.L. “On the Failure of Single-Story Wood-Framed Houses in Severe Storms.” Journal of Wind Engineering and Industrial Aerodynamics, 1988. Sparks, P.R., Schuff, S.D., and Reinhold, T.A., “Wind Damage to Envelopes of Houses and Consequent Insurance Losses.” Journal of Wind Engineering and Industrial Aerodynamics, 1994. Stubbs, N., and Perry D.C., “Damage Simulation Model for Building and Contents in Hurricane Environment,” Proceedings ASCE Structures Congress XIV, 989-996, 1996. Sutt E., Muralidhar K., and Reinhold T., “Roof Sheathing Uplift Resistance for Hurricanes,” Clemson University Report, 1998. Tryggvason, B.V., Surry, D., and Davenport, A.G., “Predicting Wind Induced Response in Hurricane Zones”. Journal of the Structural Division, ASCE, Volume 102, No ST12, December, 1976. Unanwa C.O., McDonald J.R., Mehta K.C. and Smith D.A, “Development of Wind Damage bands for Buildings,” Journal of Wind Engineering and Industrial Aerodynamics, Vol. 84, January 2000. Vickery, P.J., Skerlj, P.F. and Twisdale, L.A. (2000): “Simulation of Hurricane Risk in the U.S. Using Empirical Track Model”, Journal of Structural Engineering, October 2000, pp. 1222-1237. Watford S. W., “A Statistical Analysis of Wind Damages to Single Family Dwellings Due to Hurricane Hugo,” Master’s Thesis, Clemson University, 1991. Wiggins, J.H., Natural Hazards Tornado, Hurricane, Severe Wind Loss Models, U.S. Department of Commerce. PB-294 594, 1976.

Representative References on Simulation Methodology Law, Averill M. and Kelton, David, Simulation Modeling and Analysis, New York: McGraw-Hill. 1982. Naylor, Thomas H., Balintfy, Joseph L., Burdick, Donald S. and Chu, Kong, Computer Simulation Techniques, New York: John Wiley & Sons. 1968. Rubenstein, Reuben Y., Simulation and the Monte Carlo Method. New York: John Wiley & Sons. 1981.

5. Provide a detailed description of all changes in the model from the prior year’s submission.

AIR continually reviews model assumptions in light of new meteorological and other information and periodically modifies the hurricane model. During the last year, the following refinements were made to the model:

The ZIP Code database is updated each year. For each new ZIP Code centroid, the following data needs to be re-estimated: distance from coastline, elevation and surface

Standard G-1: Scope of the Computer Model and Its Implementation roughness. This is a technical update. This has resulted in minor changes to population- weighted centroids and its impact on losses is a 0.1% decrease.

The historical catalog has been updated through 2006 reflecting two additional years of hurricane experience in which there were a total of seven hurricane landfalls of which three landfalls were in Florida. The annual frequency distribution, the probability distribution for landfall location and the probability distribution for hurricane intensity have been updated to reflect the new data. The resulting stochastic catalog updates leads to small increase of 1.4% in overall loss costs.

Analysis of claims data from recent hurricanes is an on-going process at AIR. As a result of this analysis the relationship of contents damage to building damage for single-family homes and mobile homes has been updated. This change decreases losses for single family homes and mobile homes, contributing a 3.6% decrease to overall loss costs.

FCHLPM is requiring inclusion of demand surge in all modeled loss costs this year. AIR’s demand surge function accounts for increase in losses due to temporary inflation of costs and delays in repair caused in hurricanes. The impact of including demand surge in the overall loss cost is a 12.6% increase.

The overall change in the statewide industry residential average loss cost is 10.3%.

Standard G-2: Qualifications of Modeler Personnel and Independent Experts

G-2 Qualifications of Modeler Personnel and Consultants

A. Model construction, testing, and evaluation shall be performed by modeler personnel or consultants who possess the necessary skills, formal education, or experience to develop the relevant components for hurricane loss projection methodologies.

AIR employs a large, full-time professional staff in actuarial science, computer science, insurance and reinsurance, mathematics, meteorology, and other physical sciences, software engineering, statistics and structural engineering. Most have advanced degrees and more than twenty hold Ph.Ds. These are the academic disciplines required to properly develop, test and evaluate hurricane loss projection methodologies.

B. The model or any modifications to an accepted model shall be reviewed by either modeler personnel or consultants in the following professional disciplines: structural/wind engineering (licensed Professional Engineer), statistics (advanced degree), actuarial science (Associate or Fellow of Casualty Actuarial Society), meteorology (advanced degree), and computer/information science (advanced degree). These individuals shall be signatories on Forms G- 1 through G-6 as applicable and shall abide by the standards of professional conduct if adopted by their profession.

All modifications to AIR’s currently accepted model have been reviewed by modeler personnel or independent experts as indicated below. All AIR staff and independent experts abide by the standards of professional conduct adopted for their respective professions.

Table 1: AIR Modeler Personnel and Independent Experts Years of Professional Professional Name Discipline Experience Education Level Qualification Minor, Joseph* Structural/Wind 32 Ph.D., Texas Tech Licensed P.E. Engineering University Ljung, Greta Statistics 30 Ph.D., University of Advanced degree Wisconsin-Madison Lalonde, David Actuarial Science 23 BMath., University of FCAS Waterloo, FCAS, FCIA, MAAA Dailey, Peter Meteorology 16 Ph.D., UCLA Advanced degree Narges Computer 11 M.S., Computer Advanced degree Pourghasemi* Science/Electrical Science, University of Engineering New Hampshire Guin, Jayanta Structural Engineer 12 Ph.D., State University Advanced Degree of New York

G-2: Qualifications of Modeler Personnel and Independent Experts

*Independent expert. The vulnerability components of the AIR hurricane model were reviewed by Dr. Joseph Minor, P.E., Consulting Engineer and Professor at the University of Missouri-Rolla. The software engineering components of the AIR model were reviewed by Ms. Narges Pourghasemi, independent software engineer.

Disclosures

1. Organization Background

A. Describe the ownership structure of the modeling organization. Describe affiliations with other companies and the nature of the relationship, if any. Indicate if your organizationhas changed its name and explain the circumstances.

AIR Worldwide Corporation (AIR) is a privately held corporation and a wholly-owned subsidiary of Insurance Services Office, Inc. AIR is the sole owner of Applied Insurance Research Limited and AIR Information Technology Private Limited.

B. If the model is developed by an entity other than a modeling company, describe its organizational structure and indicate how proprietary rights and control over the model and its critical components is exercised. If more than one entity is involved in the development of the model, describe all involved.

The AIR hurricane model is developed, maintained and enhanced by full time professional staff employed by AIR.

C. If the model is developed by an entity other than a modeling company, describe the funding source for the model.

The AIR hurricane model is developed exclusively by full time professional staff employed by AIR.

D. Describe the modeler’s services.

AIR provides catastrophe risk assessment and management products and services to primary insurance companies, reinsurers, intermediaries, involuntary markets, state funds, and other insurance industry organizations such as ISO and the NCCI. We also provide services to investment banks, investors in catastrophe bonds, corporate and government risk managers.

CATRADER enables a reinsurer to estimate potential catastrophe losses from reinsurance treaties covering insurance companies worldwide. Both systems cover the perils of

Standard G-2: Qualifications of Modeler Personnel and Independent Experts hurricanes, tornadoes, hailstorms, straight-line windstorms, extratropical cyclones, earthquakes and fire-following earthquakes in over 40 countries throughout North and South America, the Caribbean, Europe and the Asia-Pacific region. This system is used in pricing treaties, for risk selection and for overall portfolio management. CATRADER provides detailed output that allows analysis of risk transfer options, other than reinsurance, thus making it an appropriate tool for primary insurance companies as well as reinsurers. Our reinsurer clients collectively represent over 85 percent of the global capacity for U.S. catastrophe reinsurance and over 60 percent of the European catastrophe reinsurance capacity.

AIR’s CLASTM service and CLASIC/2TM software products are designed to help clients estimate their potential losses from hurricanes, extratropical cyclones, tornadoes, hailstorms, straight-line windstorms, earthquakes, fire-following earthquakes, terrorism, flood, and wildfire. Designed to analyze risks at the location (i.e., geocode) level, CLASIC/2 takes full advantage of risk-specific structural details, occupancy, age and height, locational characteristics, such as site-specific geographical and geological information, and insurance policy and reinsurance treaty terms. Our clients use these loss estimates in the areas of ratemaking, underwriting, risk mitigation strategies, risk-transfer decision-making, and overall portfolio management. Our primary company clients account for over 50 percent of the total homeowners market in the U.S.

AIR provides hurricane loss estimation services for involuntary markets, including the Florida Hurricane Catastrophe Fund (FHCF). AIR also provides hurricane loss estimation services to the investment community in conjunction with various catastrophe bond offerings that have been issued. Finally, AIR is servicing an increasing number of corporate and government clients in order to help risk managers understand their exposure to extreme event risk from both natural and man-made catastrophes.

E. Indicate how long the model has been used for analyzing insurance company exposures or other such uses. Describe these uses.

Primary insurance company clients have been utilizing our hurricane loss analysis services since 1993. The North Carolina Rate Bureau has been a client of AIR since 1992. Our clients use these loss estimates in the areas of ratemaking, underwriting, risk mitigation strategies, risk-transfer decision-making, and overall portfolio management.

F. Indicate if the modeling organizationhas ever been involved in litigation or challenged by a statutory authority where the credibility of one of its U. S. hurricane model versions was disputed. Describe the nature of the case and the conclusion.

AIR has never been involved in litigation nor has been challenged by a statutory authority with respect to the credibility of its U.S. hurricane model.

G-2: Qualifications of Modeler Personnel and Independent Experts

2. Professional Credentials

A. Provide in a chart format (a) the highest degree obtained (discipline and University), (b) employment or consultant status and tenure in years, and (c) relevant experience and responsibilities of individuals involved in the primary development of or revisions to the following aspects of the model:

1. Meteorology 2. Vulnerability 3. Actuarial Science 4. Statistics 5. Computer Science

In total, the AIR technical staff consists of more than 200 professionals, more than 25 of whom hold Ph.D. credentials in their fields of expertise. While many of these professionals have been directly or indirectly involved with the hurricane model, we list below only those individuals who have contributed significantly to model development, enhancement, testing and/or validation.

Credentials are summarized below. Note that all of the individuals named are full time employees of AIR.

Standard G-2: Qualifications of Modeler Personnel and Independent Experts

1. Meteorology

(a) (b) (c) Name, Education Years at AIR Relevant Experience Dailey, Peter 6 Dr. Dailey is Director of Atmospheric Science at AIR and works on various aspects of the model, including physical properties (local Ph.D., Meteorology, effects due to varying land use and frictional effects on the wind UCLA field). Prior to joining AIR in 2001, Dr. Dailey was Principal Meteorologist at Litton-TASC where he served as technical lead on numerous projects, both governmental and commercial, involving the development of numerical weather prediction technologies. He was Principal Investigator on projects aimed at developing operational weather modeling technology for the media, including a modeling system currently used at The Weather Channel, and led an effort to develop Synthetic Natural Environments for military training simulations. Dr. Dailey received his Ph.D. and M.S. degrees in Meteorology from the University of California, Los Angeles. His background includes work modeling systems as diverse as the U. S. economy and traffic flow. Dr. Dailey is the author of numerous papers related to the application of Numerical Weather Prediction models for various military and civilian technologies with a focus on modeling at the regional scale.

Cox, Justin 1 Dr. Cox is a Research Scientist at AIR responsible for meteorological aspects of various models at AIR including the US Hurricane Model. Ph.D., Meteorology, He graduated from Cornell University in 1999 with a B.S. in University of Utah Meteorology, and received M.S. (2002) and Ph.D. (2006) degrees in Meteorology from the University of Utah. In 1999, he received an American Meteorological Society Graduate Fellowship sponsored by ITT. His M.S. thesis consisted of a detailed examination of the structure of a Wasatch mountain winter storm. He used data from NOAA research aircraft and ground-based Doppler radar measurements to reconstruct the system’s wind field at high resolution. For his doctoral dissertation, Dr. Cox worked with an interdisciplinary research team to investigate the sensitivity of local meteorology to changes in land cover. He employed a NWP (Numerical Weather Prediction) model to simulate the impact of urbanization and irrigation on local winds.

Dima, Ioana M. <1 Dr. Ioana Dima is a Research Scientist at AIR, working on various meteorological aspects of the hurricane model. She joined the Ph.D., Atmospheric company in November 2006. Her undergraduate studies were Science, University completed at University of Bucharest, Romania, where she earned a of Washington Bachelor of Science in Physics. She continued on to graduate school through a research assistantship at University of Washington (Seattle, WA). There she received her Master of Science in 2002 and her PhD in 2005, both in Atmospheric Science. Her doctoral thesis was entitled “An Observational Study of the Tropical Tropospheric Circulation” and offered a diagnosis of various elements of the low latitude general circulation. She is presently interested in investigating the interaction

G-2: Qualifications of Modeler Personnel and Independent Experts

(a) (b) (c) Name, Education Years at AIR Relevant Experience between smaller scale (tropical cyclones) and larger scale (ENSO, MJO, PDO etc) patterns of variability in the tropics.

Deng, Yi <1 Dr. Deng is a Research Scientist at AIR and works on the evaluation of the model’s hazard component. Dr. Deng received his Ph.D. in Ph.D., Atmospheric Atmospheric Sciences and M.S. degree in Statistics from the Sciences, University University of Illinois at Urbana-Champaign. From the perspective of of Illinois at Urbana- dry dynamics, his thesis research provides an explanation for the Champaign differences between the intraseasonal variability of the Pacific and Atlantic storm tracks. Prior to joining AIR in 2006, Dr. Deng was a Postdoctoral Fellow at the Institute for Geophysics, University of Texas at Austin, where he did a comprehensive study comparing precipitation extremes simulated in climate models to those observed by satellites. He was also a key member in a National Science Foundation project that aims to quantify parametric uncertainties in climate models using the method of Bayesian stochastic inversion. Dr. Deng is the author of six peer-reviewed papers related to tropical air- sea coupling, monsoon and dynamics of midlatitude storm tracks.

Butke, Jason <1 Mr. Butke is a research scientist at AIR responsible for the hazards component. He received a Bachelor of Arts degree in Geography with M.S., Geography, a minor in quantitative analysis from Bowling Green State University University of in 2003. Mr. Butke finished his Master of Science degree in Delaware Geography at the University of Delaware in 2006. The title of his master’s thesis was “A comparison between a point snow model and a mesoscale model for regional climate simulations.” While at the University of Delaware, Mr. Butke worked as a research assistant for the Delaware Environmental Observing System aiding in weather station installation and real-time web-based weather data dissemination. He has several publications in press in the field of mesoscale meteorology and spatial resolution error of gridded precipitation fields.

Standard G-2: Qualifications of Modeler Personnel and Independent Experts

2. Vulnerability

While many outside experts have been consulted on various aspects of the vulnerability functions, the primary full-time employees currently involved in model development are:

(a) (b) (c) Name, Education Years at AIR Relevant Experience Guin, Jayanta 10 Dr. Guin is Vice President responsible for operational and strategic management of the AIR Research and Modeling team. Under his Ph.D., Civil leadership the team has expanded the global coverage of natural Engineering, catastrophe models and continues to enhance existing models. He has State University over ten years of experience in probabilistic risk analysis for natural of New York, catastrophes worldwide. Dr. Guin has led the research effort on a Buffalo number of capital markets transactions that involved transfer of risk due to earthquakes, cyclones and windstorms. Prior to his graduate studies, he worked as a structural engineer with a leading design consultant in India. Dr. Guin received his B.S. in Civil Engineering from Jadavpur University in India. He earned his M.S. and Ph.D. in Civil Engineering from the State University of New York, Buffalo, with a specialization in dynamic soil-structure interaction and computational mechanics. As a member of the Computational Mechanics Laboratory at SUNY, Buffalo, Dr. Guin gained extensive experience in finite-element and boundary-element analyses for a wide range of engineering problems.

He, Hua 1 Dr. He is a Research Engineer specializing in the vulnerability functions of the AIR wind damage projection model. Dr. He has Ph. D., Civil conducted damage reconnaissance surveys in the aftermath of Engineering, Texas Hurricanes Rita (2005), Katrina (2005), Ivan (2004) and Charley Tech University (2004). He compiled paper and PDA survey forms for the Wind Science and Engineering Center (WISE) at TTU. Dr. He has conducted detailed research work on ASCE 7. His dissertation topic is data-based probabilistic wind damage estimation of building components. Dr. He earned his Ph.D. in Civil Engineering at Texas Tech University.

Jain, Vineet Kumar 3 Dr. Jain is a Research Engineer for AIR’s Research and Modeling group and has been with AIR for 2 years and is responsible for the Ph.D., Civil and development of models for the assessment of the vulnerability of Environmental structures to winds. Dr. Jain previously worked as a Senior GIS Engineering, Cornell Engineer, performing spatial analysis and modeling for natural hazard University modeling projects. He has a Ph.D. in Civil and Environmental Engineering from Cornell University.

Rowe, John 10 Mr. Rowe is Manager of AIR’s Exposures Group. He has twelve years of modeling experience, seven of which were gained at AIR. Before M.S., Structural being promoted to manager, Mr. Rowe worked on the catalogs and Engineering, City wind damage function development for numerous countries. He also University London worked on the development of the AIR storm surge flood model for and Technical the U.K. Mr. Rowe holds his M.S. in Structural Engineering from City University of

G-2: Qualifications of Modeler Personnel and Independent Experts

(a) (b) (c) Name, Education Years at AIR Relevant Experience Denmark, University London and the Technical University of Denmark, Copenhagen Copenhagen. He specialized in numerically modeling the behavior of structures.

Standard G-2: Qualifications of Modeler Personnel and Independent Experts

3. Actuarial

(a) (b) (c) Name, Education Years at AIR Relevant Experience Lalonde, David 11 Mr. Lalonde is Senior Vice President responsible for the Consulting Services group, providing Catastrophe Loss Analysis Services BMath., Actuarial (CLAS™) and Risk Transfer Services (RTS™). He is extensively Science/Statistics, involved in helping clients utilize, interpret and integrate catastrophe University of modeling technologies into effective risk management programs. Mr. Waterloo, FCAS, Lalonde has a wealth of experience in the securitization of insurance FCIA, MAAA risk and the use of dynamic financial analysis models for pricing, reserving, and corporate planning. He is responsible for ensuring AIR models meet all regulatory standards and he regularly assists clients in responding to Department of Insurance requests for information relating to the use of catastrophe models in ratemaking. He has appeared as an expert witness in rate arbitration hearings and has provided catastrophe modeling expertise on due diligence teams. Prior to joining AIR, he was a Director of Coopers & Lybrand where he directed a team of actuaries who provided a wide variety of consulting services to insurance companies and self-insured organizations to help them assess and manage their risk. Prior to that, Mr. Lalonde was Chief Actuary at the Insurance Corporation of British Columbia, where he reported to the company’s Board of Directors on a wide range of strategic and operational issues, including recommending appropriate levels of surplus based on stochastic planning models. Mr. Lalonde is a Fellow of the Casualty Actuarial Society, a Fellow of the Canadian Institute of Actuaries, and a Member of the American Academy of Actuaries. Mr. Lalonde earned his B.Math. (Honors) in Actuarial Science with Statistics from the University of Waterloo. He is a past member and Chair of the CAS Advisory Committee on Securitization/Risk Financing and a member of the AAA Extreme Event Risk Committee.

G-2: Qualifications of Modeler Personnel and Independent Experts

4. Statistics

(a) (b) (c) Name, Education Years at AIR Relevant Experience Lalonde, David 11 See Disclosure 2.A.3 above. BMath., Actuarial Science/Statistics, University of Waterloo, FCAS, FCIA, MAAA

Ljung, Greta 8 Dr. Ljung is a Senior Research Statistician at AIR. Her responsibilities include statistical modeling of historical data Ph.D., Statistics, including parameter estimation and goodness-of-fit testing, University of Wisconsin- modeling of hurricane tracks, and stochastic event generation. Madison Prior to joining AIR, Dr. Ljung was Vice President for Statistical Services at the Institute for Business & Home Safety (IBHS) where she was responsible for the development of a national database of insured losses from natural disasters. In addition, she conducted research and taught for ten years at Massachusetts Institute of Technology (MIT). Dr. Ljung is a member of the American Statistical Association and the Institute for Mathematical Statistics. Her work has been published in leading academic journals such as, the Journal of the Royal Statistical Society, Journal of Time Series Analysis, and Biometrika. She has also served on the editorial boards of Technometrics, Journal of Forecasting, and the International Journal of Forecasting, and she is a current member of the Technometrics Management Committee. Dr. Ljung holds a Ph.D. in Statistics from the University of Wisconsin-Madison.

Healy, Mary <1 Dr. Healy is a Research Statistician at AIR. Her responsibilities Ph. D., Statistics, Boston are primarily statistical support for the stochastic event University generation. Prior to joining AIR, Dr. Healy was a postdoctoral fellow at the Centers for Disease Control and Prevention / National Center for Health Statistics (CDC/NCHS). Her primary focus was spatial analysis and the effects of aggregation, especially in scale, as it pertained to mapping of disease and small area estimation. Dr. Healy is a member of the American Statistical Association (ASA), Institute for Mathematical statistics (IMS), and International Society for Bayesian Analysis (ISBA). Her work has been published in academic journals such as Statistics in Medicine and Statistics and Probability Letters. Dr. Healy holds a Ph.D. in Statistics from Boston University.

Standard G-2: Qualifications of Modeler Personnel and Independent Experts

5. Computer Science

(a) (b) (c) Name Years at AIR Relevant Experience Zvezdov, Ivelin 1 Ivelin Zvezdov is the Manager of Statistical Analysis/Numerical Models in our Software Services group. Ivelin previously M.A., Economics, St. worked as Manager of Risk Controls Group for Calpine Andrews University; Corporation. He holds a M.A. in Economics from St. Andrews Masters, European University, UK; a Masters in European Integration from Integration, University of University of Oxford, UK; and is working towards a M.S. in Oxford Applied Math. Mr. Daraskevich is a member of AIR's Research and Modeling Daraskevich, Glen 6 group where his primary focus is operational management. B.S., Civil Engineering, Previously, he was a member of AIR's Consulting group where University of he acted as a lead analyst in several securitization transactions. Connecticut Prior to joining AIR, Mr. Daraskevich acted as a consultant for M.S., Environmental the Massachusetts Department of Environmental Protection Engineering, where he contributed recommendations for re-engineering the University of New Haven agency's information systems. Before that, he worked as an M.S., Information Environmental Engineer at the Connecticut Department of Systems, Environmental Protection, where he performed environmental Boston University modeling and analysis of atmospheric pollutants. Mr. Daraskevich holds a B.S. in Civil Engineering from the University of Connecticut, a M.S. in Environmental Engineering from the University of New Haven, as well as a M.S. in Information Systems from Boston University. Davidson, Boris 12 Mr. Davidson joined AIR in 1994 and currently holds the position of Technical Manager and Chief Software Architect. B.S., Computer Mr. Davidson’s responsibilities include software architecture Engineering, design and product evolution, as well as day-to-day technical Northeastern University assistance with numerous on-going projects. He played an integral role in the development of AIR’s CATMAP/2, CATRADER and CLASIC/2 software systems and, more recently, has developed flexible application programming interfaces that allow the integration of AIR software into any Windows-based underwriting system. Prior to joining AIR, Mr. Davidson worked as software engineer for SYTRON (archival systems) and Dragon Systems (speech recognition). He earned a B.S. in Computer Engineering from Northeastern University.

Potharaju, Sudhir 6 Sudhir Kumar Potharaju is a Senior Software Engineer for our Kumar Software R & D Group. Previously, Sudhir spent two and a half years as a Software Engineer at AIT in Hyderabad. Before that, M.S., Electronics and he worked at the Electronics Corporation of India Ltd. (ECIL), Communications, involved in design, coding, and testing of various projects. In Osmania University 1998, he completed his engineering degree in Electronics and Telecommunications at IETE, Osmania University Campus.

Grenier, Roger 7 Dr. Grenier is Assistant Vice President of AIR’s Software

G-2: Qualifications of Modeler Personnel and Independent Experts

(a) (b) (c) Name Years at AIR Relevant Experience M.S., Water Resources Services group, responsible for service and support for Engineering, Stanford CLASIC/2 clients, quality assurance testing, and software University release, installation, and training. Previously, Dr. Grenier was manager of AIR’s Consulting Services group where he managed Ph.D., Civil Engineering, the scope, timelines, and production of AIR’s consulting North Carolina State projects. Before that, Dr. Grenier was a member of AIR’s University Research and Modeling team and was responsible for the development and implementation of AIR’s U.K. flood model. Dr. Grenier has extensive experience in numerical modeling. He has published numerous papers on hydrodynamic modeling and coastal hazard assessment. Dr. Grenier earned an M.S. in Water Resources Engineering from Stanford University, and a Ph.D. in Civil Engineering from North Carolina State University.

Hevron, Glenn 2 Glenn Hevron is Vice President of the Software Services group, directing client-supporting activities for CLASIC/2. Previously, B.S., Physics, Glenn worked as a Principal at Vertex Inc. in Sarasota, Florida. Computer/ Information Mr. Hevron received his B.S. in Physics and Services, Texas A&M; Computer/Information Services from Texas A&M and his MBA MBA, Southern from Southern Methodist University. Methodist University Lewis, Peter 7 Mr. Lewis is a Microsoft® Certified Professional specializing in systems support and database administration. Mr. Lewis B.S., Business provides client support for AIR’s CLASIC/2 system. Prior to Administration, joining AIR, Mr. Lewis worked as a Senior Claims Adjuster for Northeastern University Travelers Property Casualty Insurance as well as a Systems Engineer for Allmerica Property Casualty Company and Wausau Insurance Company. He has a B.S. in Business Administration from Northeastern University with a concentration in Finance.

Richards, David 5 Mr. Richards is Vice President and Director of AIR’s Software Development Group. Mr. Richards oversees the development of B.S., Electrical AIR’s existing and new products leveraging Internet delivery. Engineering, Lowell Since joining AIR, Mr. Richards has been responsible for Technological Institute significant enhancements to the AIRWeather™, ALERT™ and UNICEDE® web sites. He has also managed the development of AIR’s newest web-based products, including AIRProfiler® and CATStation®. Mr. Richards received his B. S. in Electrical Engineering from Lowell Technological Institute.

Sandri, Praveen 6 Dr. Sandri, who joined AIR in 1997, is Managing Director of India Operations. Prior to joining AIR, Dr. Sandri was a Ph.D., Civil Research Associate at the Wind Engineering Research Center of Engineering, Texas Texas Tech University where he worked in the area of wind Tech University engineering and was the Manager of the Wind Engineering Research Field Laboratory (WERFL). Dr. Sandri applied expert systems and artificial neural networks to wind engineering

Standard G-2: Qualifications of Modeler Personnel and Independent Experts

(a) (b) (c) Name Years at AIR Relevant Experience applications. He was the Project Leader for the development of WIND-RITE™, a knowledge-based expert system for the Insurance Institute for Property Loss Reduction, and he developed the “Expert System as an Alternate Code Compliance Methodology” for the Florida Housing Finance Agency. Dr. Sandri earned his Ph.D. in Civil Engineering at Texas Tech University in the area of application of artificial neural networks to wind induced damage prediction for individual non- engineered buildings. Dr. Sandri has performed post disaster damage surveys for Hurricane Opal, 1995; Hurricane Danny, 1997; Hurricane Bonnie, 1998; tornado and hail damage (TX, 1995); and tornado damage (TX 1995).

Trudeau, Larry 2 Mr. Trudeau joined AIR in 2004 and is AIR’s Software Engineering Manager. Previously, he helped start Axint B.S., Education, Technologies, a company focused on Insurance Automation and Bryant College EDI Software, where he was Vice President of Development. He also helped found ProcessWare, Inc. where he was Vice President of Engineering. Mr. Trudeau’s responsibilities at AIR include management of the desktop applications software group and the database group. Mr. Trudeau earned his B.S. from Bryant College in the field of Education.

Senthil, Tharini 3 Ms. Senthil is a Product Manager in AIR’s Software Services group responsible for managing the CLASIC/2 application B.S., Computer including requirements definition, quality assurance and release Engineering, scheduling. She is also responsible for training and providing University Of Windsor support for CLASIC/2 clients. Previously, Ms. Senthil was a CLASIC/2 Client Services Consultant as well as the Product Manager of AIR's ALERT (AIR Loss Estimates in Real-Time) service. Ms. Senthil earned a B.S. in Computer Engineering and Mass Communication from the University of Windsor.

Holden, Jonathon 1 Jonathan Holden is Sr. Program Manager of AIR’s Software Services group, responsible for service and support for B.A., Geoscience, CLASIC/2 clients, quality assurance testing, and software Franklin & Marshall release, installation, and training. Previously, Mr. Holden was a College Senior Project Manager for EMC Microsoft Practice. He holds M.S., Hydrology, a B.A. in Geoscience from Franklin and Marshall College, and a University of New M.S. in Hydrology from the University of New Hampshire. Hampshire

B. Identify any new employees or consultants (since the previous submission) working on the model.

The following new AIR employees have worked on the model since the previous submission:

G-2: Qualifications of Modeler Personnel and Independent Experts

• Justin Cox, Meteorology. See Disclosure 2.A.1 above. • Ioana M. Dima, Meteorology. See Disclosure 2.A.1 above. • Yi Deng, Meteorology. See Disclosure 2.A.1 above. • Jason Butke, Meteorology. See Disclosure 2.A.1 above. • Mary Healy, Statistics. See Disclosure 2.A.4 above. • Ivelin Zvezdov, Computer Science. See Disclosure 2.A.5 above. • Sudhir Kumar Potharaju, Computer Science. See Disclosure 2.A.5 above. • Tharini Senthil, Computer Science. See Disclosure 2.A.5 above. • Jonathon Holden, Computer Science. See Disclosure 2.A.5 above.

Standard G-2: Qualifications of Modeler Personnel and Independent Experts

C. Provide visual business workflow documentation connecting all personnel related to model design, testing, execution, maintenance, and decision- making.

Figure 4: AIR Hurricane Model Workflow

Research & Modeling Jayanta Guin Hurricane Hazard Team Catalog Generation Wind Field Calculation

Greta Ljung, Peter Dailey

Structural Engineering Team Damageability Relationships

Jayanta Guin, Vineet Kumar Jain

Model Loss Team Loss Estimation

Jayanta Guin, Vineet Kumar Jain, Glen Daraskevich

Technical Production Services Model Validation and Testing

Great Ljung, Peter Dailey

Software Development

David Richards

Software Engineering Computer Standards

Model Porting & Integration Larry Trudeau

Boris Davidson

Software Services Glenn Hevron Core Products Quality Assurance Technical Support Release Engineering Verification & Testing

Tharini Senthil, Jonathan Holden Roger Grenier, Ivelin Zvezdov Peter Lewis

End Users

G-2: Qualifications of Modeler Personnel and Independent Experts

D. Indicate specifically whether individuals listed in A. and B are associated with the insurance industry, consumer advocacy group, or a government entity as well as their involvement with consulting activities.

All individuals listed in A and B are full-time employees of AIR.

3. Independent Peer Review

A. Provide dates of external independent peer reviews that have been performed on the following components as currently functioning in the model:

1. Meteorology

Review by Dr. John Freeman in 1994.

2. Vulnerability

Reviewed by Dr. Joseph Minor, P.E. in 2001, 2002, 2003, 2004, 2005, 2006 and 2007.

3. Actuarial

This component has not been reviewed by independent experts.

4. Statistics

This component has not been reviewed by independent experts.

5. Computer Science

Reviewed by Dr. Mark Wolfskehl in 2002. Reviewed by Dr. John Kam in 2003, 2004 and 2005. Reviewed by Ms. Narges Pourghasemi in 2006 and 2007.

B. Provide documentation of independent peer reviews directly relevantto the modeler’s responses to the current Standards, Disclosures, or Forms. Identify any unresolved or outstanding issues as a result of these reviews.

Dr. Joseph Minor's review of the vulnerability component of the AIR hurricane model and his CV are included as Attachment B. There were no unresolved issues.

Standard G-2: Qualifications of Modeler Personnel and Independent Experts

Ms. Pourghasemi’s review of the model software, along with her CV, are included as Attachment C. There were no unresolved issues.

C. Describe the nature of any on-going or functional relationship the organization has with any of the persons performing the independent peer reviews.

AIR has no on-going functional relationship with any of the persons who performed independent reviews, nor with any of their employees or consultants, nor with any independent organization.

4. Provide a completed Form G-1, General Standards Expert Certification.

A completed Form G-1 is provided on page 44.

5. Provide a completed Form G-2, Meteorological Standards Expert Certification.

A completed Form G-2 is provided on page 45.

6. Provide a completed Form G-3, Vulnerability Standards Expert Certification.

A completed Form G-3 is provided on page 46.

7. Provide a completed Form G-4, Actuarial Standards Expert Certification.

A completed Form G-4 is provided on page 47.

8. Provide a completed Form G-5, Statistical Standards Expert Certification.

A completed Form G-5 is provided on page 48.

9. Provide a completed Form G-6, Computer Standards Expert Certification.

A completed Form G-6 is provided on page 49.

Standard G-3: Risk Location

G-3 Risk Location

A. ZIP Codes used in the model shall be updated at least every 24 months using information originating from the United States Postal Service. The United States Postal Service issue date of the updated information shall be reasonable.

ZIP Codes used in the model are updated annually with information provided by the United States Postal Service (USPS). The USPS issue date of the information currently used in the model is November 2005.

B. ZIP Code centroids, when used in the model, shall be based on population data.

The AIR model uses population-weighted ZIP Code centroids.

C. ZIP Code information purchased by the modeler shall be verified by the modeler for accuracy and appropriateness.

The methodology employed by AIR’s vendor for computing population centroids is identical to the computational methods promulgated by the U.S. Census Bureau.

Additional quality control measures are performed by AIR to verify the positional accuracy of the population centroid in relation to the ZIP Code boundaries and ensure their appropriateness. These measures comprise a set of procedures that overlay the population- weighted centroids with the ZIP Code boundaries.

Disclosures

1. List the current ZIP Code databases used by the model and the components of the model to which they relate. Provide the effective (official United States Postal Service) date corresponding to the ZIP Code databases.

The model uses a valid ZIP Code database and a database of population-weighted centroids. The ZIP Code database is used to locate an exposure for purposes of the hazard and vulnerability components of the model. The ZIP Code database is derived from information issued by the USPS as of November 2005.

2. Describe in detail how invalid ZIP Codes are handled.

A ZIP Code is defined to be “invalid” if it does not match the list of currently valid ZIP Codes.

An invalid ZIP Code is first checked against AIR’s master database of ZIP Codes for all years from 1990 through 2005. This database includes ZIP Codes that were valid in

Standard G-3: Risk Location previous years but later became invalid. Any invalid ZIP Code matched to a ZIP Code in this master database is reassigned to the currently valid ZIP Code.

The model produces a list of ZIP Codes that do not appear in AIR’s master database of valid ZIP Codes. Exposure in any remaining invalid ZIP Code is not modeled. The insurer may choose to allocate or map these exposures back into the exposure data set; this is not part of the model. If AIR makes assumptions regarding allocation or remapping at a client’s request, such assumptions are included in the Project Information Assumption Form (PIAF), which is included in Attachment A of this submission.

G-4: Independence of Model Components

G-4 Independence of Model Components

The meteorological, vulnerability, and actuarial components of the model shall each be theoretically sound without compensation for potential bias from the other two components.

All components of the AIR model are theoretically sound and independently derived, no component compensates for any potential bias in any of the other two components. Furthermore, each component is independently validated.

Form G-1: General Standards Expert Certification

I hereby certify that I have personally reviewed the submission of AIR Atlantic Tropical Cyclone Model V9.0 for compliance with the 2006 Standards adopted by the Florida Commission on Hurricane Loss Projection Methodology and hereby certify:

1) that the model meets the General Standards (G1 – G4), 2) that the Disclosures and Forms related to the General Standards section contain accurate, reliable, unbiased, and complete information, 3) that my review was completed in accordance with the professional standards and code of ethical conduct for my profession, and 4) that in expressing my opinion I have not been influenced by any other party in order to bias or prejudice my opinion.

Jayanta Guin Ph.D., Structural Engineering Name Professional Credentials (Area of Expertise)

February 26, 2007 Signature (original submission) Date

Signature (response to Deficiencies, if any)) Date

April 4, 2007 Signature (final submission) Date

An updated signature is required following modifications to the model and any revisions to the original submission.

NOTE: A facsimile or any properly reproduced signature will be acceptable to meet this requirement.

Form G-2: Meteorological Standards Expert Certification

1) that the model meets the Meteorological Standards (M1 – M6), 2) that the Disclosures and Forms related to the Meteorological Standards section contain accurate, reliable, unbiased, and complete information, 3) that my review was completed in accordance with the professional standards and code of ethical conduct for my profession, and 4) that in expressing my opinion I have not been influenced by any other party in order to bias or prejudice my opinion.

Peter Dailey Ph.D., Senior Research Scientist Name Professional Credentials (Area of Expertise)

February 26, 2007 Signature (original submission) Date

March 15, 2007 Signature (response to Deficiencies, if any) Date

April 4, 2007 Signature (final submission) Date

An updated signature is required following modifications to the model and any revisions to the original submission.

NOTE: A facsimile or any properly reproduced signature will be acceptable to meet this requirement.

Form G-3: Vulnerability Standards Expert Certification

1) that the model meets the Vulnerability Standards (V1 – V2), 2) that the Disclosures and Forms related to the Vulnerability Standards section contain accurate, reliable, unbiased, and complete information, 3) that my review was completed in accordance with the professional standards and code of ethical conduct for my profession, and 4) that in expressing my opinion I have not been influenced by any other party in order to bias or prejudice my opinion.

Joseph E. Minor Civil Engineering/Wind Engineering Name Professional Credentials (Area of Expertise)

February 23, 2007 Signature Date

March 19, 2007 Signature (response to Deficiencies, if any) Date

Signature (final submission) Date

An updated signature is required following modifications to the model and any revisions to the originalsubmission.

NOTE: A facsimile or any properly reproduced signature will be acceptable to meet this requirement.

Form G-4: Actuarial Standards Expert Certification

1) that the model meets the Actuarial Standards (A1 – A9), 2) that the Disclosures and Forms related to the Actuarial Standards section contain accurate, reliable, unbiased, and complete information, 3) that my review was completed in accordance with the professional standards and code of ethical conduct for my profession, and 4) that in expressing my opinion I have not been influenced by any other party in order to bias or prejudice my opinion.

David A. Lalonde FCAS, FCIA, MAAA, AIR Senior VP Name Professional Credentials (Area of Expertise)

February 26, 2007 Signature Date

March 15, 2007 Signature (response to Deficiencies, if any) Date

April 4, 2007 Signature (final submission) Date

An updated signature is required following modifications to the model and any revisions to the original submission.

NOTE: A facsimile or any properly reproduced signature will be acceptable to meet this requirement.

Form G-5: Statistical Standards Expert Certification

1) that the model meets the Statistical Standards (S1 – S6), 2) that the Disclosures and Forms related to the Statistical Standards section contain accurate, reliable, unbiased, and complete information, 3) that my review was completed in accordance with the professional standards and code of ethical conduct for my profession, and 4) that in expressing my opinion I have not been influenced by any other party in order to bias or prejudice my opinion.

Greta M. Ljung Ph.D., Senior Research Statistician Name Professional Credentials (Area of Expertise)

February 26, 2007 Signature Date

March 19, 2007 Signature (response to Deficiencies, if any) Date

April 4, 2007 Signature (final submission) Date

An updated signature is required following modifications to the model and any revisions to the originalsubmission.

NOTE: A facsimile or any properly reproduced signature will be acceptable to meet this requirement.

Form G-6: Computer Standards Expert Certification

1) that the model meets the Computer Standards (C1 – C7), 2) that the Disclosures and Forms related to the Computer Standards section contain accurate, reliable, unbiased, and complete information, 3) that my review was completed in accordance with the professional standards and code of ethical conduct for my profession, and 4) that in expressing my opinion I have not been influenced by any other party in order to bias or prejudice my opinion.

Narges Pourghasemi M.S., Computer Science Name Professional Credentials (Area of Expertise)

February 5, 2007 Signature Date

Signature (response to Deficiencies, if any) Date

Signature (final submission) Date

An updated signature is required following modifications to the model and any revisions to the original submission.

NOTE: A facsimile or any properly reproduced signature will be acceptable to meet this requirement.

Standard M-1: Official Hurricane Set

2006 Meteorological Standards

M-1 Base Hurricane Storm Set

For validation of landfall and by-passing storm frequency in the stochastic storm set, the modeler shall use the latest updated Official Hurricane Set or the National Hurricane Center HURDAT as of June 1, 2006 or later. Complete additional season increments based on updates to HURDAT approved by the Tropical Prediction Center/National Hurricane Center are acceptable modifications to these storm sets. Peer reviewed atmospheric science literature can be used to justify modifications to the Base Hurricane Storm Set.

For landfall frequency analyses, AIR used the latest updated Official Hurricane Set. For by-passing storm frequency, AIR used the latest official hurricane set supplemented with the latest National Hurricane Center (NHC) HURDAT.

Disclosures

1. Identify the Base Hurricane Storm Set, the release date, and the time period included for landfall and by-passing storm frequencies.

The Base Hurricane Storm Set consists of the Official Hurricane Set released November 1, 2006 and spanning the years 1900-2005, supplemented for bypassing storms and the 2006 season with the latest NHC HURDAT data (last released on February 1, 2007) and spanning the years 1900-2006.

2. If the modeler has modified the Base Hurricane Storm Set, provide justification for such modifications.

AIR has not modified its Base Hurricane Storm Set since 2006 except to add the most recent available data from the Official Hurricane Set and NHC HURDAT data.

Standard M-2: Hurricane Characteristics

M-2 Hurricane Characteristics

Methods for depicting all modeled hurricane characteristics, including but not limited to wind speed, radial distributions of wind and pressure, minimum central pressure, radius of maximum winds, strike probabilities, tracks, the spatial and time variant wind fields, and conversion factors, shall be based on information documented by currently accepted scientific literature.

Methods for depicting all modeled hurricane characteristics are based on information documented in the scientific literature.

Disclosures

1. Identify the hurricane characteristics (e.g., central pressure or radius of maximum winds) that are used in the model. Describe the historical data used for each of these characteristics identifying all storms used.

After extensive research and testing, including cross-verification of data, AIR researchers developed probability distributions for each of the meteorological variables that are critical to the characterization of hurricanes. These variables along with their historical data used to develop probability distributions are:

Hurricane Characteristic Historical Data Hurricane Frequency Updated Official Hurricane Set, HURDAT(1900-2006)

Updated Official Hurricane Set, HURDAT(1900-2006), NHC's Tropical Cyclone Reports (Preliminary Reports (1958-2006)) Landfall Location Updated Official Hurricane Set, NHC's Tropical Cyclone Hurricane Intensity Reports (Preliminary Reports(1958-2006)) NWS23 (1900-1975), NWS38 (1900-1984), RPI's Extended Radius of Maximum Winds Best Track Data, EBTRK (1988-1999) Forward Speed HURDAT(1900-2001) Storm Heading at Landfall HURDAT(1900-2001) Storm Track HURDAT(1900-2001)

Along with consideration of storm asymmetry, storm filling over land, surface roughness, radial wind speed profile, wind field of an event is then generated.

2. Describe the dependencies among variables in the wind field component and how they are represented in the model.

With the exception of radius of maximum winds, which is represented using a regression model of the form Rmax = ƒ(cp, latitude) + ε, the individual hurricane parameters are

Standard M-2: Hurricane Characteristics considered independent of each other. The following variables are, however, dependent upon latitude:

• Central pressure • Radius of maximum winds • Forward speed • Storm heading • Filling rate • Air density coefficient • Coriolis parameter

The latitudinal dependence of these parameters is modeled by coastal segment, with three exceptions. The air density coefficient and Coriolis parameter are direct functions of latitude. Rmax is a continuous function of both central pressure and latitude.

3. Describe the process for converting gradient winds to surface winds including the treatment of the inherent uncertainties in the conversion factor with respect to location of the site compared to the radius of maximum winds over time. Justify the variation of the gradient to surface winds conversion factor relative to hurricane intensity.

AIR uses the wind-pressure relationship described in NWS23. The relationship first calculates the maximum gradient wind speed at the radius of maximum winds, and then applies a conversion factor of 0.9 to adjust the gradient wind speed to 10meter, 10-minute over-water wind speed. The conversion factor is called reduction factor in NWS-23. For a complete discussion, see NWS-23, pages 234-235. According to the study by Franklin, Black, and Valde (2003), the value of 0.9 is recommended to convert flight-level winds at 700 mb to 10 meters above at the radius of maximum winds. The constant conversion factor used by AIR is thus justified and spatially consistent with current understanding of the relationship between gradient wind and surface wind.

The work of Franklin, Black, and Valde (2003) and NWS23 shows, however, that there are inherent uncertainties associated with the reduction factor. Additional research is needed before action can be taken to account for the uncertainties. For example, research is needed to investigate how the factor varies with tropical cyclone characteristics temporally and spatially, and to determine an appropriate probability distribution for the conversion factor.

4. Describe how the wind speeds generated in the wind field model were converted from sustained to gust and identify the average time.

Input to the AIR vulnerability module is 1-minute sustained wind, thus there is no need to convert winds to 3-second gusts.

Conversion of ten-minute averaging wind speed to one-minute sustained wind is based on accepted engineering relationships and varies from 1.10 to 1.36, as a function of land

Standard M-2: Hurricane Characteristics use/land cover. For typical suburban terrain, the gust factor is 1.2. Typical references include Simiu and Scanlan “Wind Effects on Structures”(1996), N. Cook “The Designers Guide to Wind Loading of Building Structures, Part 1.”(1985), and ESDU Engineering Sciences Data “Wind Engineering”(1994).

5. Describe how the asymmetric nature of hurricanes is considered in the model.

Asymmetry is handled by adding a wind component to a symmetric wind field on the right of the hurricane track and subtracting the component from winds on the left of the track. The asymmetry factor is calculated as a function of the hurricane forward speed and the angle between the track direction and the surface wind direction:

c 2 ASYM = c1 ( T ) cos (β) where ASYM = the wind component due to asymmetry

c1, c2 = Constants T = the hurricane forward speed β = the angle between the track direction and the surface wind direction

6. Describe the stochastic hurricane tracks and discuss their appropriateness. Describe the historical data used as the basis for the model’s hurricane tracks.

The AIR hurricane model is part of an Atlantic basin-wide model that includes the Caribbean Islands and the U.S. mainland. The methodology used to generate the basin- wide tracks is developed using historical tracks available in the HURDAT database from the period 1900 to 2001. The starting latitude and longitude of each storm is simulated from a bivariate probability distribution derived from historical starting locations using a bivariate Gaussian smoothing technique. The dependence structure in the historical data at successive 6-hourly time intervals is quantified and used to develop time series models that describe the track direction, forward speed, and central pressure as the storm moves across the basin. The analysis of the HURDAT data shows that first-order Markov models are appropriate for track direction and forward speed. Higher order dependence present in the central pressure along the track is represented using a second-order autoregressive time series model. The parameters of these models are estimated using a procedure that captures the spatial variability in the behavior of the storms in different parts of the Atlantic basin.

The HURDAT data is available at 6-hour time intervals and this resolution is sufficient to capture the evolution of each storm across the Atlantic basin. However, at the U.S. coastline, additional information is needed since the track points rarely fall directly on the coastline and the storm characteristics begin to change once the storm moves over land. As discussed elsewhere, AIR uses the more detailed landfall information available in NWS- 23, NWS-38 and Preliminary Reports and Tropical Cyclone Reports from NHC (available at http://www.nhc.noaa.gov/pastall.shtml) to generate storm characteristics at landfall. The

Standard M-2: Hurricane Characteristics landfall information, including locational frequency and storm intensity, is also used to eliminate tropical storms and generate post-landfall hurricane tracks. The Atlantic basin- wide tracks are integrated with the post-landfall hurricane tracks using a spline smoothing technique that ensures consistency in intensity, radius of maximum winds, and storm heading across the tracks. The methodology produces realistic tracks that resemble the storms tracks that have been observed historically across the Atlantic basin and the U.S. mainland.

7. Describe how the coastline is segmented (or partitioned) in determining the parameters for hurricane frequency used in the model. Provide the hurricane frequency distribution by intensity for each segment.

To determine hurricane probabilities along the coast of Florida, the actual number of hurricane occurrences is tabulated for smoothed 50 nautical mile segments. The number of occurrences for each segment is then smoothed by setting it equal to a weighted average of the landfall counts for each segment and the surrounding segments.

The intent of the smoothing procedure is to eliminate or reduce the random variation in the limited historical data while maintaining areas of high and low risk. The smoothing is based on a procedure well-documented in the meteorological literature (see, for example, NOAA Technical Report NWS-38, p. 75). The results are validated by comparing the simulated annual occurrence rate to the actual occurrence rate experienced between 1900 and 2006 for the 100 nautical mile segments along the Florida, Alabama, Georgia and Mississippi as illustrated in Figure 5.

Standard M-2: Hurricane Characteristics

Figure 5: Hurricane Frequency by Coastal Segment, 1900-2006

8. For hurricane characteristics modeled as random variables, describe the probability distributions.

Hurricane Frequency. The AIR hurricane model uses a Negative Binomial distribution to generate the number of hurricane landfalls per year. The graph below shows the actual historical data from 1900-2006 along with the modeled distribution for the entire US coastline. The modeled distribution fits the historical data very well. The average number of landfalls per year is 1.785, which includes multiple landfalls by a single hurricane. The average number of hurricanes per year making landfall in the U.S. is 1.60. We make no other assumptions as to future hurricane activity.

Figure 6: Number of Landfalls Per Year, Modeled vs Actual, 1900-2006

0.35 0.30 Historical 0.25 Modeled 0.20 0.15

Frequency 0.10 0.05 0.00 012345678+ Number of Landfalls

Standard M-2: Hurricane Characteristics

Landfall Location. In the AIR hurricane model there are 3,100 possible landfall points at each one nautical mile of smoothed coastline from Texas to Maine. In Florida there are over 900 possible landfall locations. Historical hurricane occurrences since 1900 are used to estimate a smoothed locational frequency distribution. The smoothing technique employed produces a landfall distribution that maintains areas of high versus low frequency while smoothing out variations due to limitations on completeness in the historical record.

Hurricane Intensity. The AIR hurricane model utilizes central pressure as the primary hurricane intensity variable. The historical data are modeled using Weibull distributions where the parameters are estimated for each 100 nautical-mile coastal segment as well as for larger regional segments, with the final distribution being a weighted combination of the two. The Weibull form was selected based on “goodness-of-fit” tests with actual historical data. The use of the Weibull distribution for storm central pressure is documented in the scientific literature. When the Weibull generates a storm that is inconsistent with the historical record, the central pressure is reset to be no lower than 884 mb (close to the lowest central pressure found in the Atlantic basin).

Radius of Maximum Winds. The probability distribution for radius of maximum winds (Rmax) is modeled using a regression model of the form:

Rmax = ƒ(cp, latitude) + ε , where ƒ(cp, latitude) represents the mean of Rmax for given values of central pressure and latitude. The error term, ε, is assumed to be Normally distributed. The parameters in this regression model are estimated using data available in NWS 38, supplemented with subsequent storms. The final distribution is truncated using limits that depend on central pressure and are consistent with the range of historically observed values.

Forward Speed. Forward speed is modeled using a Lognormal distribution with parameters estimated for each of the 31 one hundred nautical mile segments of coastline. Separate distributions are estimated for each of these segments to capture the dependence of this variable upon geographical location, particularly latitude.

Storm Heading at Landfall. Separate distributions for storm heading at landfall are estimated for variable length segments of coastline. The length of each segment is governed by the general orientation of that segment. Standard 100 nautical mile segments cannot be used because the orientation of the coastline might change dramatically within these segments. The corresponding probability distributions are combined normal distributions with bounds based on the historical record and meteorological expertise.

Storm Tracks. The starting latitude and longitude of each storm is simulated from a bivariate probability distribution derived from historical starting locations using a bivariate Gaussian smoothing technique. The dependence structure in the historical data at successive 6-hourly time intervals is quantified and used to develop time series models that describe the track direction, forward speed, and central pressure as the storm moves across the basin. The analysis of the HURDAT data shows that first-order Markov models are

Standard M-2: Hurricane Characteristics appropriate for track direction and forward speed. Higher order dependence present in the central pressure along the track is represented using a second-order autoregressive time series model. A more complete discussion of the methodology used to generate storm tracks is provided in the answer to Disclosure 6 above.

Wind Field Generation Component. For each simulated event, the AIR hurricane model simulates the storm’s movement along its track. A complete time profile of wind speeds is developed for each location affected by the storm, thus capturing the effect of duration of wind on structures as well as peak wind speed. Calculations of local intensity take into account the effects of the asymmetric nature of the hurricane wind field, storm filling over land, surface roughness, and relative wind speeds as the distance from the radius of maximum winds increases.

9. Identify any changes in the functional representation of hurricane characteristics during an individual storm event life cycle .

AIR wind field does capture the wind speed discontinuity at the coastal boundary. Overland wind speeds take into account the impact of local land use/land cover. The overland storm intensities of simulated tropical cyclones change according to AIR’s overland filling rates.

10. Describe how your model’s wind field is consistent with the inherent differences in wind fields for such diverse storms as Hurricane Charley, Hurricane Katrina, and Hurricane Wilma, for example.

Historical data shows that tropical cyclones affecting Florida can be quite diverse. Sources of diversity in the wind fields of Florida hurricanes include intensity at landfall, storm decay after landfall, forward speed, and radius of maximum winds. The AIR wind field model accounts for the response of a hurricane winds to each of these parameters.

The landfalling hurricanes of 2004 and 2005 provide a good sample of diversity in these storm parameters. In these seasons, hurricane intensity spanned the spectrum from very intense (e.g., Charley 2004) to weak (e.g., Katrina 2005 Florida landfall). The rate of intensity decay after landfall (typically referred to as filling) in some storms was relatively slow (e.g., Wilma 2005), while in others it was rapid (e.g., Charley 2004). Forward speed varied from slow (e.g., Frances 2004) to fast (e.g., Wilma 2005). Finally, the radius of maximum winds varied from small (e.g., Dennis 2004) to large (e.g., Wilma 2005). Despite the diversity in these basic storm parameters, all of which relate directly to the wind field, AIR modeled wind speeds are realistic and unbiased.

Standard M-3: Landfall Intensity

M-3 Landfall Intensity

Models shall use maximum one-minute sustained 10-meter wind speed when defining hurricane landfall intensity. This applies both to the Base Hurricane Storm Set used to develop landfall strike probabilities as a function of coastal location and to the modeled winds in each hurricane which causes damage. The associated maximum one-minute sustained 10-meter wind speed shall be within the range of wind speeds (in statute miles per hour) categorized by the Saffir-Simpson scale.

The AIR hurricane model uses maximum one-minute average sustained wind speeds, which are valid for 10-meter elevation, to define intensity.

Disclosures

1. Define an “event” in the model. Discuss how storms that intensify or decay at or below the Category 1 level are accounted for in the model.

An event, in the context of the AIR Atlantic Tropical Cyclone Model, is a simulated hurricane. For Category 1 hurricanes that intensify or decay at or below the Category 1 level, losses are calculated until the wind speed is less than 40 mph.

2. Describe how the model handles events with multiple landfalls and by-passing storms. Be specific with respect to how by-passing storms are handled in the model when the wind speeds are less than hurricane force winds.

An event, or simulated hurricane in the context of the AIR Atlantic Tropical Cyclone Model, may comprise several landfalls or bypasses, where a bypass is defined as a hurricane that does not make landfall but causes damaging winds over land. Since the AIR model follows each simulated hurricane from the time of its inception until it dissipates, multiple landfalls and bypassing hurricanes are part of the simulation. For by-passing storms, the wind speed is monitored hourly to ensure that it meets the definition of a by- passer. The simulated frequency of multiple landfalls and bypassing storms is consistent with the historical frequency of these storms during the period 1900-2006.

3. Provide all model derived characteristics of the Florida hurricane in the stochastic storm set with the greatest over water intensity at the time of landfall.

• Landfall lon/lot: -80.5530, 25.2997 • CP (mb) = 884.97 • Rmax = 8.65 • FS = 23.73 • Landfall angle = -60.10 • 1-min 10-m Maximum wind (mph) = 176.51

Standard M-4: Hurricane Probabilities

M-4 Hurricane Probabilities

A. Modeled probability distributions for hurricane intensity, forward speed, radii for maximum winds, and storm heading shall be consistent with historical hurricanes in the Atlantic basin.

AIR modeled probability distributions for hurricane strength, forward speed, radii for maximum winds and storm heading at landfall are consistent with observed historical hurricanes in the Atlantic basin and are bounded by observed global extremes as documented in accepted scientific literature available.

Hurricane strength is modeled as a combination of Weibull distributions. Weibull distributions are fitted to the historical data for each 100 nautical-mile coastal segment. Separate Weibulls are then estimated for various regions along the coast. The intensity distribution used for each segment is a combination, with appropriate weights applied, of the regional Weibulls and the segment Weibull. The Weibull form was selected based on “goodness-of-fit” tests with actual historical data. The Weibull parameters, α and β, are estimated using the maximum likelihood estimation method.

While this particular procedure is unique to the AIR hurricane model, the use of a Weibull distribution for hurricane strength is well documented in the literature. The advantage of the AIR technique is that it provides the appropriate level of smoothing of the historical data and results in the appropriate change in risk, when moving along the Florida coastline. Also, it leads to more stable results than empirical tail-fitting procedures.

AIR uses theoretical distributions rather than empirical tail-fitting techniques because of the advantages that result from use of theoretical distributions. Theoretical distributions lead to more reliable and stable results, they can be tested for goodness-of-fit and they are not overly sensitive to new or corrected data. Use of theoretical distributions also produces results that are more indicative of future events because theoretical distributions, unlike empirical procedures, are not overly reliant on actual historical events, which for some segments, could be few or even none at all. Extreme values in the tail are reset so as not to be inconsistent with the historical record.

The radius of maximum winds is simulated using a regression model in which the mean is a function of central pressure and latitude. The model incorporates the fact that stronger storms tend to have a smaller radius than less intense storms. Also, because of the dependence on latitude, the average radius increases as one moves from south to north. The average simulated radius for Florida landfalling hurricanes is 25 miles. The average radius calculated from actual storms occurring between 1900 and 2006 is also 25 miles.

Forward speed is modeled using a Lognormal distribution. Use of this distribution is documented in the literature. The average simulated forward speed for the state of Florida is 14.4 mph and the average forward speed calculated from hurricanes occurring between 1900 and 2006 is 14.5 mph. The slowest speed simulated in Florida is 4.61 mph. The maximum forward speed is bounded by coastal segment according to NWS-23.

Standard M-4: Hurricane Probabilities

The AIR modeled maximum one-minute sustained 10-meter wind speed does not exceed nor is the modeled minimum central pressure less than the global extreme for a tropical cyclone, as documented in accepted scientific literature. The AIR modeled maximum one- minute sustained 10-meter wind speed is 177 mph. The AIR modeled minimum central pressure is 884 mb (close to the lowest central pressure found in the Atlantic basin).

B. Modeled hurricane probabilities shall reflect the Base Hurricane Storm Set used for category 1 to 5 hurricanes and shall be consistent with those observed for each coastal segment of Florida and neighboring states (Alabama, Georgia, and Mississippi).

AIR modeled hurricane probabilities for category 1 - 5 hurricanes reasonably match the historical record through 2006 and are consistent with those observed for each geographical area of Florida, Alabama, Georgia and Mississippi.

Disclosures

1. List assumptions used in creating the hurricane characteristic databases.

The AIR hurricane simulation model is based on meteorological data for hurricanes occurring between 1900 and 2006. The meteorological data utilized to develop the model can be found in various source documents listed previously in the response to Standard G1 Disclosure 4, most notably:

Storm track, wind speed, and central pressure data for all tropical cyclones affecting the U.S. since 1900 were obtained from the National Hurricane Center and the National Climatic Data Center. In addition to the aforementioned publications and sources, verification of storm parameters and wind speeds were obtained from the National Hurricane Center’s Preliminary Reports for Specific Hurricanes for individual storms.

The primary databases used by the AIR model, as well as the aspect of the model to which they relate, are provided in the table below.

Standard M-4: Hurricane Probabilities

Table 2: Primary Databases

Database Model Component Public/Proprietary

Historical hurricane activity Event Generation Public Smoothed coastline file with 3,100 Event Generation Proprietary points at each one nautical mile Wind Speed Surface terrain characteristics Public Generation Wind Speed ZIP Code centroid database Public Generation High resolution coastline file with Wind Speed Proprietary (created over 70,000 points Generation by vendor for AIR) “Effective” coastline file for Wind Speed Proprietary calculating surface roughness Generation

2. If the model incorporates short term and long term variations in annual storm frequencies, describe how this is incorporated.

The model does not incorporate either short term or long term variation in expected annual storm frequency.

3. Provide a completed Form M-1, Annual Occurrence Rates.

A completed Form M-1 is provided on page 68.

Standard M-5: Land Friction and Weakening

M-5 Land Friction and Weakening

A. The magnitude of land friction coefficients shall be consistent with currently accepted scientific literature relevant to current geographic surface roughness distributions and shall be implemented with appropriate geographic information system data.

The method used by AIR to model the effects of land friction on wind speeds is based on scientific methods and provides realistic wind speed transitions between adjacent ZIP Codes, counties, and territories. The magnitude of the friction coefficients is consistent with the accepted scientific literature. Typical references include Simiu and Scanlan “Wind Effects on Structures”, N. Cook “The Designers Guide to Wind Loading of Building Structures. Part 1.” and ESDU Engineering Sciences Data “Wind Engineering.”

In the AIR model, the initial step in calculating the friction coefficient for each point of interest is to obtain an estimate of the surface roughness at the site. At ground level, horizontal drag forces induced by the surface roughness are exerted on the wind flow, causing retardation of the wind near the ground. The surface roughness is estimated based on digital USGS land use/land cover data. The land use/land cover categories vary from urban or built-up land, to agricultural land, to forest land or wetlands, to water. Each terrain type has a different “roughness value” that will lead to different frictional effects on wind speeds. In general, the more rough the terrain the larger the frictional effect on wind speeds. The magnitudes of the friction coefficients in the AIR method are consistent with accepted scientific literature and empirical hurricane wind speed data.

It is important to note that as a hurricane travels inland and encounters different types of terrain, the hurricane wind speeds do not adjust immediately upon experiencing a new terrain type. In practice, the wind has to blow over a certain length, or fetch, before the boundary layer is in equilibrium with the underlying surface, i.e., the variation of wind speed with height does not change as the fetch of the upwind uniform terrain increases. Downwind of a change in terrain roughness, such as the edge of an urban area, a new internal boundary layer begins to grow. Within this new layer the flow is not in equilibrium and the wind profile changes as distance increases. The wind profile change is modeled by using a function that varies the upwind friction coefficient to the equilibrium or fully developed friction factor over the appropriate fetch. The transition is consistent with the accepted scientific literature and the actual available hurricane wind speed data.

Figure 7 below shows wind speeds and friction factors, respectively, for a simulated storm making landfall on the panhandle of Florida. Figure 8 shows wind speeds and friction factors, respectively, for a simulated storm making landfall on the southeastern part of Florida. It can be seen that there are realistic wind speed transitions between adjacent ZIP Codes.

Standard M-5: Land Friction and Weakening

Figure 7: Florida Panhandle Storm: Wind speeds and Friction Factors for Simulated Storm

Figure 8: South Florida Storm: Wind Speeds and Friction Factors for Simulated Storm

B. The hurricane overland weakening rate methodology used by the model shall be consistent with historical records.

The model’s overland weakening rates, or filling rates, compare favorably with the historical records for storms of all intensities.

Disclosures

1. Describe and justify the functional form of hurricane decay rates used by the model.

Standard M-5: Land Friction and Weakening

Once over land, the hurricane moves away from its source of energy, i.e., warm ocean water. As a result, the eye “fills” and winds degrade as the time since landfall increases. AIR’s filling functions give the reduction in over water wind speed as a function of time since landfall, rather than distance. A faster moving storm will cause hurricane force winds further inland than a slow moving storm with the same initial intensity (wind speed). Therefore the distance of hurricane force winds inland as a function of Saffir-Simpson category is not an output of the AIR model.

The wind speed at the radius of maximum winds is calculated each hour after landfall for each simulated storm. The wind speed, taking into account an adjustment factor for the filling effect, is calculated as follows:

Vh = V (Adjfach ) where Vh = the wind speed at the eye at hour h

V = the wind speed

b Adjfach = exp (-a*t ) where a and b are dependent on the coastal segment. t is the time since landfall.

Note that the function varies by coastal region and smoothing algorithms are applied such that there is no sudden jump between regions.

The AIR hurricane filling functions provide reliable weakening rates for the state of Florida and fall within plus or minus 20% of the Kaplan and DeMaria values.

2. Describe the relevance of the gust factor used in the model.

The AIR model doesn’t use any gust factors that are typically referred to the conversion factor between hourly mean wind and 3-sec peak gust.

The AIR model uses a factor to convert 10-minute wind to 1-minute wind. The conversion of ten-minute averaging wind speed to one-minute sustained wind is based on accepted engineering relationships and varies from 1.10 to 1.36, as a function of land use/land cover. For typical suburban terrain, the gust factor is 1.2. Typical references include Simiu and Scanlan “Wind Effects on Structures” (1996), N. Cook “The Designers Guide to Wind Loading of Building Structures, Part 1.” (1985), and ESDU Engineering Sciences Data “Wind Engineering” (1994).

3. Identify all non-meteorological variables that affect wind speed estimation (e.g., surface roughness, topography, etc.).

Standard M-5: Land Friction and Weakening

Surface roughness and the distance from the coast are other variables that affect wind speed estimation.

4. Provide the collection and publication dates of the land use and land cover data used in the model and justify their timeliness for Florida.

The model uses digital, satellite-derived land use/land cover data dating from the early to mid-90s and published by the USGS (http://landcover.usgs.gov/natllandcover.asp) in 2000. The AIR hurricane model covers 27 states, including Florida, and therefore requires a nationwide land use/land cover (LULC) database. At the time of the model update, the LULC database used was the most recent national land cover database available. The National Land Cover Database 2001 (NLCD 2001), based on satellite data from 2001, was initially released in January 2007, and will be implemented in AIR’s hurricane model as soon as is practicable.

5. Provide a graphical representation of the modeled degradation rates for Florida hurricanes over time compared to wind observations. Reference to the Kaplan-DeMaria decay rates alone is not acceptable.

AIR’s overland weakening rates, or filling rates, are a function of time, as faster moving storms will cause more damage inland than slower moving storms with the same initial intensity. The filling characteristics of North American storms vary with geographical location. The AIR model is based on a reanalysis of the data in NWS38 and combines fifty nautical mile segments along the Gulf and East Coast of the U.S. into regions. The model provides for smooth transitions between adjacent regions.

Figure 9 shows actual and modeled filling rates for 2004 Hurricanes. The degradation of model-generated winds is consistent with that of observed winds.

Figure 9: Actual and modeled filling rates for 2004 Hurricanes

Actual Model Filling Rate

024681012 Hours After Landfall

Standard M-5: Land Friction and Weakening

6. The spatial distribution of model- generated winds should be demonstrated to be consistent with the observed winds.

Spatial variability of winds across the footprint of a hurricane is a function of a variety of factors. These factors include (a) the radial profile of winds from the eye to storm’s periphery, (b) the radius of maximum winds, (c) the storm’s forward speed, (d) the storm’s latitude, and (e) local surface roughness characteristics. Each of these factors is accounted for in the AIR wind field formula, and is therefore reflected in model-generated wind speeds. Figure 9a demonstrates that the distribution of model-generated winds for two storms (Hurricanes Dennis and Charley) is consistent with the distribution of observed winds (overlaid).

Figure 9a. Observed and modeled wind speeds for Hurricanes Dennis and Charley

7. Document any differences between the treatment in the model of decay rates for stochastic hurricanes compared to historical hurricanes affecting Florida.

Unless specified, before 1999 historical hurricanes affecting Florida use the same NWS-38 based decay rates as for stochastic hurricanes. The exceptions, all of which are bypassing storms, are No Name 18 (1933), No Name 4 (1935), Able (1951), Fox (1952), Ginny (1953), Abby (1968), Alison (1995), and Danny (1997). After 2000 historical hurricanes use the actual observed central pressure over their life time.

8. Provide a completed Form M-2, Maps of Maximum Winds.

A completed Form M-2 can be found on page 70.

Standard M-6: Logical Relationships of Hurricane Characteristics

M-6 Logical Relationships of Hurricane Characteristics

A. The magnitude of asymmetry shall increase as the translation speed increases, all other factors held constant.

The magnitude of asymmetry increases as the translation speed increases, all other factors held constant.

The time variant wind field, including the radial distribution of wind speeds, is consistent with accepted scientific principles and with historical hurricane characteristics.

B. The mean wind speed shall decrease with increasing surface roughness (friction), all other factors held constant.

The mean wind speed decreases with increasing surface roughness (friction), all other factors held constant.

Disclosure

1. Provide a completed Form M-3, Radius of Maximum Winds.

A completed Form M-3 is provided on page 72.

Form M-1: Annual Occurrence Rates

A. A. Provide annual occurrence rates for landfall from the data set defined by marine exposure that the model generates by hurricane category (defined by wind speed in the Saffir-Simpson scale) for the entire state of Florida and selected regions as defined in Figure 3. List the annual occurrence rate (probability of an event in a given year) per hurricane category. Annual occurrence rates should be rounded to two decimal places.

Table 3: Modeled Annual Occurrence Rates Entire State Region A – NW Florida Region B – SW Florida Category Historical Modeled Historical Modeled Historical Modeled 1 0.25 0.21 0.10 0.09 0.08 0.05 2 0.12 0.16 0.04 0.06 0.02 0.05 3 0.16 0.16 0.03 0.04 0.05 0.05 4 0.04 0.07 0.00 0.02 0.02 0.03 5 0.02 0.01 0.00 0.00 0.01 0.00

Florida By-Passing Region C – SE Florida Region D – NE Florida Hurricanes Category Historical Modeled Historical Modeled Historical Modeled 1 0.07 0.06 0.00 0.01 0.06 0.06 2 0.05 0.05 0.02 0.01 0.03 0.03 3 0.08 0.05 0.00 0.00 0.02 0.03 4 0.02 0.03 0.00 0.00 0.00 0.01 5 0.01 0.00 0.00 0.00 0.00 0.00

Region E – Georgia Region F – Alabama/Mississippi Category Historical Modeled Historical Modeled 1 0.00 0.01 0.07 0.05 2 0.04 0.01 0.03 0.04 3 0.00 0.00 0.06 0.04 4 0.00 0.00 0.00 0.02 5 0.00 0.00 0.01 0.00

Note: Except where specified, Number of Hurricanes does not include By-Passing Storms

B. The historical frequencies below have been derived from the Commission’s Official Hurricane Set. If the National Hurricane Center’s HURDAT or other hurricanes in addition to the Official Hurricane Set as specified in Standard M- 1 are used, then the historical frequencies should be modified accordingly.

AIR includes the 2006 hurricane season in modeling landfall frequency. Historical frequencies as specified in Standard M-1 have been modified accordingly.

Form M-1: Annual Occurrence Rates

C. Describe model variations from the historical frequencies.

The modeled frequencies are based on historical frequencies for the period 1900-2006. There are no variations from these frequencies.

D. Provide vertical bar graphs depicting distributions of hurricane frequencies by category by region of Florida (Figure 3) and for the neighboring states of Alabama/Mississippi and Georgia. For the neighboring states, statistics based on the closest milepost to the state boundaries used in the model are adequate.

Note: Except where specified, Number of Hurricanes does not include By-passing Storms

Figure 10: State of Florida and Neighboring States By Region

Form M-2: Maps of Maximum Winds

A. Provide a color contour map of the maximum winds for the modeled version of the Base Hurricane Storm Set.

Figure 11: ZIP Code Level Maximum Wind Speed  Historical 107-year Period

Form M-2: Maps of Maximum Winds

B. Provide a color contour map of the maximum winds for a 100-year return period from the stochastic storm set.

Figure 12: ZIP Code Level Maximum Wind Speed  Simulated 100-year Return Wind Speeds

C. Provide the maximum winds plotted on each contour map

The maximum 1-min 10-m sustained wind from Figure 11 is 152.9 mph.

The maximum 1-min 10-m sustained wind from Figure 12 is 123.1 mph.

Form M-3: Radius of Maximum Winds

A. For the central pressures in the table below, provide ranges for radius of maximum winds used by the model to create the stochastic storm set.

Table 4: Range of Rmax by Central Pressure Central Pressure (mb) Range of Rmax (mi) 900 5 - 28 910 5 - 34 920 5 – 40 930 6 – 45 940 6 – 50 950 6 – 55 955 6 – 60 960 6 – 62 965 6 – 63 970 7 – 65 975 7 – 65 980 7 – 65 985 7 – 65 990 7 – 65

B. Identify the other variables that influence Rmax.

The probability distribution for Rmax is a function of central pressure and latitude.

Form M-3: Radius of Maximum Winds

C. Provide a representative scatter plot of Central Pressure (x-axis) versus Rmax (y-axis) to demonstrate relative populations and continuity of sampled hurricanes in the stochastic storm set. “Representative” means that the relative distribution of hurricane frequencies across both Central Pressure and Rmax ranges should be evident.

Below is a representative scatter plot of central pressure versus Rmax.

Figure 13: State of Florida and Neighboring States

Official Hurricane Set

Landfall Code Landfall Code A B 11/1/2006 Standards Enter/ Central Wind Enter/ Central Wind Name YYYYMMDD Landfall Code Exit Pressure Speed Category Exit Pressure Speed Category NONAME 4 19010802 HRLA1MS1AL1 NONAME 3 19030909 HRCFL2AFL1 Enter 980 75 1 NONAME 2 19060614 HRCFL1 NONAME 6 19060919 HRMS2AL2 NONAME 8 19061008 HRCFL2 Enter 967 125 3 NONAME 8 19091006 By-Passing HRLA3 MS2 NONAME 5 19101009 HRBFL3 Enter 941 121 3 NONAME 1 19110808 HRAFL1 AL1 Enter 990 81 1 NONAME 2 19110823 HRGA2SC2 NONAME 3 19120910 HRAL1 NONAME 4 19150831 HRAFL1 Enter 982 92 1 NONAME 1 19160629 HRMS3AL3 NONAME 13 19161012 HRAL2AFL2 Enter 974 115 2 NONAME 14 19161111 HRBFL1 Enter 990 81 1 NONAME 3 19170921 HRAFL3 Enter 964 104 2 NONAME 2 19190902 By-Passing NONAME 6 19211020 HRBFL3DFL2 Enter 952 104 2 NONAME 4 19240913 HRAFL1 Enter 994 75 1 NONAME 7 19241014 HRBFL1 Enter 972 93 1 NONAME 2 19251129 HRBFL1 Enter 994 75 1 NONAME 1 19260722 HRDFL2 NONAME 6 19260911 HRCFL4BFL3AFL3 AL3 Enter 950 121 3 Exit 950 121 3 NONAME 10 19261014 By-Passing NONAME 1 19280803 HRCFL2 NONAME 4 19280906 HRCFL4DFL2 GA1 SC1 NONAME 2 19290922 HRCFL3AFL2 Enter 980 75 1 NONAME 3 19320826 HRAL1 NONAME 5 19330725 HRATX2CFL1 NONAME 12 19330831 HRCFL3 NONAME 2 19350829 HRBFL5AFL2 Enter 985 86 1 Enter 892 173 5 NONAME 6 19351030 HRCFL2 Exit 973 75 1 NONAME 5 19360727 HRAFL3 Enter 973 90 1 NONAME 2 19390807 HRCFL1AFL1 Exit 990 80 1 NONAME 3 19400805 HRGA2SC2 NONAME 5 19411003 HRCFL2BFL2AFL2 Enter 990 75 1 Exit 960 109 2 NONAME 11 19441012 HRBFL3DFL2 Enter 949 117 3 NONAME 1 19450620 HRAFL1 Enter 982 92 1 NONAME 9 19450912 HRCFL3 NONAME 5 19461005 HRBFL1 Enter 993 75 1 NONAME 4 19470904 HRCFL4 LA3 MS3BFL2 Exit 978 97 2 NONAME 8 19471009 HR GA2 SC2CFL1 Enter 975 80 1 NONAME 7 19480918 HRBFL3CFL2 Enter 963 115 3 NONAME 8 19481003 HRCFL2 NONAME 2 19490823 HRCFL3 BAKER 19500820 HRAL1 EASY 19500901 HRAFL3 Enter 958 102 2 K NG 19501013 HRCFL3 FLORENCE 19530923 HRAFL1 Enter 982 92 1 FLOSSY 19560921 HR LA2AFL1 Enter 974 92 1 DONNA 19600829 HRBFL4 NC3 NY3DFL2 CT2 RI2 MA1 NH1 ME1 Enter 930 132 4 ETHEL 19600914 HRMS1 CLEO 19640820 HRCFL2 DORA 19640828 HRDFL2 ISBELL 19641008 HRBFL2CFL2 Enter 964 107 2 BETSY 19650827 HRCFL3 LA3 ALMA 19660604 HRAFL2 Enter 970 98 2 INEZ 19660921 HRBFL1 Enter 977 76 1 GLADYS 19681013 HRAFL2DFL1 Enter 977 86 1 CAMILLE 19690814 HRLA5MS5 AGNES 19720614 HRAFL1 NY1 CT1 Enter 978 85 1 ELOISE 19750913 HRAFL3 Enter 955 119 3 DAVID 19790825 HRCFL2DFL2 GA2 SC2 FREDERIC 19790829 HRAL3MS3 ELENA 19850828 By-Passing AFL3HRAL3MS3 KATE 19851115 HRAFL2 Enter 967 92 1 FLOYD 19871009 HRBFL1 Enter 993 75 1 ANDREW 19920816 HRCFL5BFL3 LA3 Exit 950 126 3 ER N 19950731 HRCFL1AFL2 Enter 974 98 2 OPAL 19950927 HRAFL3 Enter 942 113 3 DANNY 19970716 HRLA1AL1 EARL 19980831 HRAFL1 Enter 987 81 1 GEORGES 19980915 By-Passing BFL2HRMS2 IRENE 19991012 HRBFL1 Enter 987 80 1 CHARLEY 20040809 HRBFL4CFL1DFL1SC1NC1 Enter 941 150 4 FRANCES 20040825 HRCFL2BFL1 Exit 969 92 1 IVAN 20040902 By-Passing AFL3HRAL3 JEANNE 20040913 HRCFL3BFL1AFL1 Exit 965 86 1 Exit 965 86 1 DENNIS 20050704 HRAFL3 Enter 946 121 3 KATR NA 20050823 HRCFL1LA3MS3AL1 RITA 20050918 By-Passing LA3CTX2 WILMA 20051015 HRBFL3CFL2 Enter 950 121 3 Total by Landfall Code 26 25

The Codes: AFL = Northwest Florida; BFL = Southwest Florida; CFL = Southeast Florida; DFL = Northeast Florida

NOTE Category defined by wind speed; HURDAT Landfall Code defined by central pressure

Official Hurricane Set

Landfall Code Landfall Code Landfall Code Adjacent States C D By-Pass Enter/ Central Wind Enter/ Central Wind Region Central Wind Name YYYYMMDD Exit Pressure Speed Category Exit Pressure Speed Category Affected Pressure Speed Category State Category State Category NONAME 4 19010802 MS 1 AL 1 NONAME 3 19030909 Enter 977 98 2 NONAME 2 19060614 Enter 979 86 1 NONAME 6 19060919 MS 2 AL 2 NONAME 8 19061008 Exit 967 81 1 NONAME 8 19091006 C 978 98 2 MS 2 NONAME 5 19101009 NONAME 1 19110808 NONAME 2 19110823 GA 2 NONAME 3 19120910 AL 1 NONAME 4 19150831 NONAME 1 19160629 MS 3 AL 3 NONAME 13 19161012 NONAME 14 19161111 NONAME 3 19170921 NONAME 2 19190902 B 929 132 4 NONAME 6 19211020 Exit 980 92 1 NONAME 4 19240913 NONAME 7 19241014 NONAME 2 19251129 NONAME 1 19260722 Enter 960 109 2 NONAME 6 19260911 Enter 931 134 4 NONAME 10 19261014 C 968 110 2 NONAME 1 19280803 Enter 977 98 2 NONAME 4 19280906 Enter 935 128 3 NONAME 2 19290922 Enter 948 114 3 NONAME 3 19320826 AL 1 NONAME 5 19330725 Enter 990 81 1 NONAME 12 19330831 Enter 948 132 4 NONAME 2 19350829 NONAME 6 19351030 Enter 973 75 1 NONAME 5 19360727 NONAME 2 19390807 Enter 990 81 1 NONAME 3 19400805 GA 2 NONAME 5 19411003 Enter 954 121 3 NONAME 11 19441012 NONAME 1 19450620 NONAME 9 19450912 Enter 951 116 3 NONAME 5 19461005 NONAME 4 19470904 Enter 947 125 3 NONAME 8 19471009 Exit 993 85 1 GA 2 NONAME 7 19480918 Exit 964 92 1 NONAME 8 19481003 Enter 963 86 1 NONAME 2 19490823 Enter 954 116 3 BAKER 19500820 AL 1 EASY 19500901 K NG 19501013 Enter 955 112 3 FLORENCE 19530923 FLOSSY 19560921 DONNA 19600829 Exit 969 110 2 ETHEL 19600914 MS 1 CLEO 19640820 Enter 967 99 2 DORA 19640828 Enter 961 99 2 ISBELL 19641008 Exit 968 105 2 BETSY 19650827 Enter 952 115 3 ALMA 19660604 INEZ 19660921 GLADYS 19681013 Exit 966 86 1 CAMILLE 19690814 MS 5 AGNES 19720614 ELOISE 19750913 DAVID 19790825 Enter 968 98 2 Exit 971 98 2 GA 2 FREDERIC 19790829 AL 3 MS 3 ELENA 19850828 A 953 127 3 AL 3 MS 3 KATE 19851115 FLOYD 19871009 ANDREW 19920816 Enter 922 165 5 ER N 19950731 Enter 984 86 1 OPAL 19950927 DANNY 19970716 AL 1 EARL 19980831 GEORGES 19980915 B 975 104 2 MS 2 IRENE 19991012 Exit 984 75 1 CHARLEY 20040809 Exit 970 86 1 Exit 970 86 1 FRANCES 20040825 Enter 960 104 2 IVAN 20040902 A 946 121 3 AL 3 JEANNE 20040913 Enter 950 121 3 DENNIS 20050704 KATR NA 20050823 Enter 984 81 1 MS 3 RITA 20050918 B 974 62 2 WILMA 20051015 Exit 955 110 2 31 7

Standard V-1: Derivation of Vulnerability Functions

2006 Vulnerability Standards

V-1 Derivation of Vulnerability Functions

A. Development of the vulnerability functions is to be based on a combination of the following: (1) historical data, (2) tests, (3) structural calculations, (4) expert opinion, or (5) site inspections. Any development of the vulnerability functions based on structural calculations or expert opinion shall be supported by tests, site inspections, or historical data.

The original AIR hurricane model vulnerability functions, developed in 1985, were primarily based on structural engineering research publications, damage surveys conducted by wind engineering experts, and analyses of available loss data. The original damage functions were fine-tuned over time based on the results of post-disaster field surveys and on analyses of additional detailed loss data from clients.

B. The method of derivation of the vulnerability functions shall be theoretically sound.

The AIR vulnerability functions have been developed by experts in wind and structural engineering and based on published engineering research. The functions have been validated based on results of damage surveys and on actual claims data provided by client companies. The method has been peer reviewed internally and by external experts and is theoretically sound.

C. Any modification factors/functions to the vulnerability functions or structural characteristics and their corresponding effects shall be clearly defined and be theoretically sound.

There are two types of modifications to vulnerability functions: the AIR wind model uses a modification factor to vulnerability functions to capture the effectiveness of building code, and modifications functions are used to account for mitigation measures. All modifications for building code effectiveness for existing construction fall within the range increment of plus or minus 25% and are theoretically sound. A discussion of modeled individual risk characteristics and their mitigating effects can be found in the response to Standard V-2 below.

D. Construction type and construction characteristics shall be used in the derivation and application of vulnerability functions.

The AIR hurricane model uses vulnerability functions for 20 different residential construction types, which are enumerated in the disclosures below. The model also

Standard V-1: Derivation of Vulnerability Functions includes an individual risk analysis module that accounts for a wide range of construction characteristics.

E. In the derivation and application of vulnerability functions, assumptions concerning building code revisions and building code enforcement shall be justified.

The model differentiates between four different building code regions in Florida and the model’s vulnerability functions are modified accordingly. It is assumed, for example, that buildings in the southern part of the state are characterized by higher degree of wind resistivity. The model also differentiates buildings built in different time periods in Florida. Engineering judgment went into the actual selection of the modification factors. These modification factors reflect the evolution of building codes and higher level of engineering attention in newer construction relative to older construction. They have been validated by comparing actual losses with simulated losses for different areas and time periods in Florida and have been found to be reasonable and theoretically sound. The model does not explicitly consider building code enforcement.

For buildings built to the minimum requirements of Florida Building Code (FBC 2001), separate vulnerability functions have been developed. Six unique building categories were identified in the entire state of Florida by taking into consideration the design wind speed, the terrain exposure category, and the requirements of Wind-borne Debris Region and High-Velocity Wind Zone, as specified in Florida Building Code 2001. The vulnerability functions for these six unique building categories were derived using the building features and mitigation measures from the AIR individual risk module that meet the minimum requirements of the Florida Building Code 2001.

F. Vulnerability functions shall be separately derived for building structures, mobile homes, appurtenant structures, contents, and additional living expense.

AIR engineers have developed separate vulnerability functions for the primary structure, the appurtenant structures, contents and additional living expenses. The AIR model output provides losses for each coverage separately, and combined, as well as by construction type. Output is provided gross and net of deductibles.

G. The minimum wind speed that generates damage shall be reasonable.

The model starts calculating losses at 40 mph one-minute average wind speed, which is reasonable.

Standard V-1: Derivation of Vulnerability Functions

Disclosures

1. Provide a flow chart documenting the process by which the vulnerability functions are derived and implemented.

Figure 14: Derivation and Implementation of AIR Vulnerability Functions

2. Describe the nature and extent of actual insurance claims data used to develop the model’s vulnerability functions. Describe in detail what is included, such as, number of policies, number of insurers, date of loss, and number of units of dollar exposure, separated into personal lines, commercial, and mobile home.

Validation data comes from multiple sources and client companies. Not all companies or sources provide the same level of detail. Loss data for actual events normally consists of claim counts, paid losses, by ZIP Code, and by line of business. Loss data are also frequently provided by policy form, construction type, coverage, and date of loss. AIR has been provided detailed loss data by 6 client companies covering about $600 billion of personal and about $6 billion of mobile homes exposure.

Standard V-1: Derivation of Vulnerability Functions

AIR was first provided actual claims data in 1986 when developing the model for the E.W. Blanch Company. Three primary companies submitted detailed loss data for hurricanes Alicia (1983), Diana (1984), Bob (1985), Danny (1985), Elena (1985), Gloria (1985), Juan (1985), and Kate (1985). The loss data included, by county or state, by line of business, claim counts, paid and incurred losses. Detailed exposure data was also available for each company, which included number of risks, amount of insurance and deductibles by five digit ZIP Code.

In 1989, at the request of a client, AIR performed a “blind” validation test for hurricanes Frederic (1979), Allen (1980), Alicia (1983), Diana (1984), Danny (1985), Elena (1985), Gloria (1985), Juan (1985), and Kate (1985) based on detailed exposure data (also number of risks, amount of insurance and deductibles by five digit ZIP Code). Aggregate losses for each hurricane were provided to AIR after the test. The AIR test results were judged by the client to be “quite good”.

Since 1992, many of our primary company clients, from whom we have detailed exposure data, have also provided for validation purposes detailed loss data for hurricanes Andrew, Opal, Erin, Bertha and Fran. Data sets from 14 such permutations of client company/storm losses have been analyzed. Loss data received from these companies is at the ZIP Code level and often by construction class and coverage.

Additional data were provided by our clients for 1998 Hurricanes Bonnie, Earl, Georges and Tropical Storm Frances. Data sets from several permutations of client company/storm losses have been analyzed. Loss data received from these companies is at the ZIP Code or policy level.

More recently AIR has begun receiving claims data from several client companies for 2004 hurricanes Charley, Frances, Jeanne and Ivan.

In summary, the AIR simulation model vulnerability functions are based on loss data spanning many companies and several hurricanes affecting different geographical areas, not just Florida. New data is analyzed as it becomes available. Examples of the comparison between actual loss data and the AIR vulnerability functions are shown in the graphs below.

Standard V-1: Derivation of Vulnerability Functions

Figure 15: Actual and Simulated Damage Ratios vs Wind Speed: Coverage A - Single Company, Single Storm Damage Ratio Damage Actual Simulated

Wind Speed (mph)

Figure 16: Actual and Simulated Damage Ratios vs Wind Speed: Coverage C - Single Company, Single Storm

Actual Damage Ratio Damage Simulated

Wind Speed (mph)

Standard V-1: Derivation of Vulnerability Functions

Figure 17: Actual and Simulated Damage Ratios vs Wind Speed: Coverage D - Single Company, Single Storm

Actual

Damage Ratio Damage Simulated

Wind Speed (mph)

Figure 18: Actual and Simulated Damage Ratios vs Wind Speed: Mobile Homes - Single Company, Single Storm

Actual

Damage Ratio Damage Simulated

Wind Speed (mph)

Standard V-1: Derivation of Vulnerability Functions

Figure 19: Actual and Simulated Damage Ratios vs Wind Speed: Frame - Single Company, Single Storm

Actual

Damage Ratio Damage Simulated

Wind Speed (mph)

Figure 20: Actual and Simulated Damage Ratios vs Wind Speed: Masonry - Single Company, Single Storm

Actual

Damage Ratio Damage Simulated

Wind Speed (mph)

Standard V-1: Derivation of Vulnerability Functions

3. Summarize site inspections, including the source, and a brief description of the resulting use of these data in development, validation, or verification of vulnerability functions.

AIR engineers have performed post-disaster field surveys for every U.S. landfalling hurricane since Hugo in 1989, including the notable ones like Andrew, Fran, Georges, Floyd, Charley, Frances, Ivan, Jeanne, Dennis, Katrina, Rita, and Wilma. Additionally, AIR engineers have also surveyed damage in the aftermath of major storms outside of the mainland U.S., including Hurricane Floyd in the Bahamas and Hurricane Fabian in Bermuda. These surveys improve our understanding of the response of structures to winds, including damage mechanisms. Such damage investigations provide, for example, information on the relative vulnerability of various construction types and building components used in the development and validation of the vulnerability relationships.

Based on damage investigations, buildings can be classified as “engineered” or “non- engineered” structures. Most residential dwellings are generally classified as non- engineered. A typical example is a wood-frame single-family dwelling, a construction that may not have received much attention from a structural engineer. Most commercial structures, built in accordance with building codes and under the supervision of a structural engineer, are classified as engineered structures. A typical example of an engineered structure is a high-rise reinforced concrete building. In general, an engineered building is more wind resistant than a non-engineered building. 2004 and 2005 post-disaster surveys, conducted by AIR engineers indicate that low-rise commercial and residential-commercial wood frame and masonry buildings--that do not get as much engineering attention as the high-rise buildings--have similar vulnerability as their residential counterparts. Recent damage surveys also indicated that year-built of a residential dwelling provides important information in determining its vulnerability. Newly built buildings were observed to perform better than the old buildings.

Wind damage primarily affects non-structural elements, such as different components of the building envelope. Field surveys indicate that the failure of houses with wood-framed roofs often occurs first at the roof, often at improper fastening between the roof sheathing and building frame. Damage investigations have identified certain building connections that have failed in windstorms, for example, a common failure initiation point on roof systems occurs where the roof membranes are attached to edges and corners. Several roof system failures could be attributed to the lifting and peeling of metal edge flashings. Uplift of the roof edges allows the wind to penetrate underneath the roof membrane, resulting in pressure rise beneath the membrane and removal of the roof covering. At high wind speeds, the integrity of the entire structure can be compromised, particularly in cases where the roof provides the lateral stability by supporting the tops of the building’s walls. Thus, three damage regimes can be identified for residential buildings: a) the low damage regime corresponding to wind speeds of less than about 90 mph, where damage is limited to roof covering and cladding, b) the medium damage regime where damage propagates to roof sheathing, connections and openings and c) the catastrophic damage regime corresponding to wind speeds in excess of 130 mph, where the roof framing is severely damaged, resulting in lateral instability of walls causing their collapse and complete destruction of

Standard V-1: Derivation of Vulnerability Functions the building. In case of engineered buildings, damage typically occurs to non-structural components like mechanical equipment, roofing, cladding and windows; complete structural collapse is extremely rare.

In certain parts of the United States, masonry building systems are the prevalent construction method. When masonry is used as the exterior wall material, the walls are normally constructed to full height and then wood floors and the roof are framed into the masonry. Damage investigations have confirmed that such construction results in continuous exterior walls and thus a stronger structural frame, resulting in exterior walls that are more resistant to winds and windborne debris impacts as compared to wood frame buildings.

While exploring the damage caused by Hurricane Georges and Tropical Storm Frances, AIR engineers validated the effects of wind duration on damage estimation, long a component of the AIR model. Damage resulting from a slower moving (longer duration) storm will be higher because of the cumulative effects of wind.

Information such as detailed above, obtained from post-disaster damage investigations, has been incorporated in the development and validation of the vulnerability model.

4. Describe the research used in the development of the model’s vulnerability functions.

The AIR hurricane model uses engineering research publications, damage surveys conducted by wind engineering experts, and analysis of loss data to develop vulnerability functions (Figure 14). AIR engineers have been doing post-disaster field surveys for each hurricane since 1989. AIR also receives detailed claims and loss data from client companies and ISO. Damage survey findings and analysis of loss data have been used to fine-tune vulnerability functions.

5. Describe the number of categories of the different vulnerability functions. Specifically, include descriptions of the structure types, lines of business, and coverages in which a unique vulnerability function is used.

The AIR hurricane model uses 8 basic construction types for residential construction, 8 for apartments and condominiums, and 4 for mobile homes. Thus 20 unique vulnerability functions have been developed separately for each of four coverages:

A – Building B – Appurtenant Structures C – Contents D – Time Element

Standard V-1: Derivation of Vulnerability Functions

Table 5: Residential Construction Types in the AIR Model

Construction Type Description

Residential and Apartment or Condominium Buildings Wood Frame Wood frame structures tend to be mostly low rise (one to three stories, occasionally four stories). Stud walls are typically constructed of 2 inch by 4 or 6 inch wood members vertically set 16 or 24 inches apart. These walls are braced by plywood or by diagonals made of wood or steel. Many detached single and low- rise multiple family residences in the United States are of stud wall wood frame construction. Masonry veneer Wood frame structures with one width of non-load bearing concrete, stone or clay brick attached to the stud wall. Unreinforced Masonry Unreinforced masonry buildings consist of structures in which there is no steel reinforcing within a load bearing masonry wall, floors, roofs, and internal partitions in these bearing wall buildings are usually of wood. Reinforced Masonry Reinforced masonry construction consists of load bearing walls of reinforced brick or concrete-block masonry. Floor and roof joists constructed with wood framing are common. Reinforced Concrete Reinforced concrete buildings consist of reinforced concrete columns and beams. Steel Steel frame buildings consist of steel columns and beams. Light Metal Light Metal buildings are made of light gauge steel frame and are usually clad with lightweight metal or asbestos siding and roof, often corrugated. They typically are lowrise structures. Unknown Represents a weighted average of all of the above construction types except the mobile homes

Construction Type Description

Mobile Homes Mobile home with no tie-downs This code would be used for mobile homes (manufactured homes) with no anchoring systems present. Mobile home with partial tie- This code would be used for mobile homes (manufactured homes) downs when the tie downs are either over-the-top ties, or frame ties but not both or with fewer ties than recommended by the manufacturer. Mobile home with full tie- This code would be used for mobile homes (manufactured homes) downs when the anchoring systems are both over-the-top ties, and frame ties. Typically 10 frame ties and 7 over the top ties are required for full tie down in singlewide mobile homes. Mobile Homes Represents a weighted average of tie-down types, including no tie- downs. This code would be used for mobile homes (manufactured homes) when the tie down information is unknown.

Standard V-1: Derivation of Vulnerability Functions

These construction types are typically provided by primary insurer client companies.

Apartments and condominiums usually receive a similar degree of engineering attention as commercial construction. From a structural viewpoint, therefore, commercial construction and apartments/condominiums are quite similar. Nevertheless, apartments and condominiums have some building components that make them more susceptible to windstorms than commercial construction. The more vulnerable components found in apartments and condominiums include balconies, awnings, and double sliding glass doors etc. These components are less engineered at the design and construction stages and hence lead to apartments/condominiums being more vulnerable than commercial construction. Therefore, AIR engineers have developed separate damage functions for several height ranges (1, 2-3, 4-7 and > 7 stories) for apartment/condominium structures.

Claims data from recent hurricanes in Florida indicate that buildings built prior to 1995 are significantly more vulnerable than post-1994 buildings. The AIR model includes year-built categories: pre-1995, 1995-2001, and post-2001.

6. Identify the one-minute average sustained wind speed at which the model begins to estimate damage.

The model starts calculating losses at 40 mph one-minute average wind speed. Damages continue to be calculated until wind speeds fall below 40 mph.

7. Describe how the duration of wind speeds at a particular location over the life of a hurricane is considered.

The AIR hurricane model considers the duration of wind speed at each location over the life of the storm while wind speeds are above 40 mph. The model estimates damage using sustained wind speeds. Duration is an important consideration in damage estimation. For example, consider two storms of equal intensity but different duration. Damages for the longer duration storm will be higher because of the cumulative effects of wind. Recent examples of this include Cyclone Tracy, Hurricanes Luis and Georges.

When estimating damage to a property at any point in time, the model takes into account the extent of the damage that has occurred in the preceding time period, thereby calculating the cumulative effects of winds. That is, each successive damage ratio is applied, in succession, to the remaining undamaged portion of the exposure from the preceding period.

8. Provide a completed Form V-1, One Hypothetical Event.

A completed Form V-1 is provided on page 91.

Standard V-2: Mitigation Measures

V-2 Mitigation Measures (*Significant Revision due to Form V-2)

A. Modeling of mitigation measures to improve a building’s wind resistance and the corresponding effects on vulnerability shall be theoretically sound. These measures shall include fixtures or construction techniques that enhance:

• Roof strength • Roof covering performance • Roof-to-wall strength • Wall-to-floor-to-foundation strength • Opening protection • Window, door, and skylight strength.

The individual risk analysis module, which accounts for the effects of mitigation measures, was developed using a knowledge-based expert system in a structured approach. Based on structural engineering expertise and building damage observations made in the aftermath of historical hurricanes, 22 building features have been identified as having a significant impact on the building losses. These features include fixtures or construction techniques that enhance roof strength, roof-covering performance, roof-to-wall strength, wall-to-floor- to-foundation strength, opening protection, window, door, and skylight strength. Options for each feature are identified based on construction practice. Algorithms for modifying the vulnerability functions, for both structural and nonstructural damage, are developed based on engineering principles and building performance observations. The module supports any combination of multiple building features on the building damage and produces a modification function to the vulnerability function. The modification function captures the changes to building vulnerability that result when certain building features are present and when information on such building features is known. The modification function varies with the wind intensity to reflect the relative effectiveness of a building feature when subject to different wind speeds.

The building vulnerability component of the AIR model explicitly addresses new construction built in accordance to the new building code, FBC2001. Methods for estimating the effects of mitigation measures, as described in Attachment D: U.S. Hurricane Individual Risk Methodology, are theoretically sound.

B. Application of mitigation measures shall be empirically justified both individually and in combination.

Methods for estimating the effects of mitigation measures, as described in Attachment D, U.S. Hurricane Individual Risk Methodology, are reasonable, both individually and in combination.

Standard V-2: Mitigation Measures

Disclosures

1. Provide a completed Form V-2, Mitigation Measures – Range of Changes in Damage.

A completed Form V-2 is provided on CD-ROM in both Excel and PDF format, and is included below on page 95. Note that AIR’s Form V-2 includes more than the minimum mitigation measures required.

2. Provide a description of the mitigation measures used by the model that are not listed in Form V-2.

Table 6 describes the mitigation measures that are not listed in Form V-2. Please note that Table 7 in V-2 section of the standard provides the modification factor corresponding to these features.

Table 6: Mitigation Measures in the AIR Model

MITIGATION MEASURES DESCRIPTION BUILDING CONDITION A general qualitative description of building condition from visual inspection TREE EXPOSURE Tree Exposure: Descr bes the tree hazard around the building SMALL DEBRIS SOURCE Descr bes the potential of small debris in a radius of 200 feet LARGE MISSILE SOURCE Descr bes the potential of large missiles in a radius of 100 feet ROOF GEOMETRY Descr bes the shape of the roof ROOF PITCH Addresses the roof slope ROOF COVERING The nature of material used to cover the roof ROOF DECK Material and construction type of the roof deck of building ROOF COVER Nature of the connections used to secure the roof covering to the roof ATTACHMENT deck ROOF DECK Nature of the connections used to secure the roof deck to the underlying ATTACHMENT roof support system ROOF ANCHORAGE Nature of connections used to secure the roof support systems to the walls WALL TYPE Materials used for external walls of the building WALL SIDING Materials used for weathering protection of walls GLASS TYPE Type of glass used in building GLASS PERCENT The percent area of the walls covered by glass WINDOW PROTECTION Describes the nature of wind protection systems used EXTERIOR DOORS Descr bes the nature of the exterior doors in the building BUILDING FOUNDATION Connection type between the structure and foundation CONNECTION ROOF ATTACHED Description of the mechanical and other equipment on top of roofs STRUCTURES WALL ATTACHED Components of a property that are not an integral part of the main STRUCTURES building but are physically attached to it APPERTENANT Components of a property that are not an integral part of the main STRUCTURES building and are not connected to it ROOF BUILT The year the roof was put in place

Form V-1: One Hypothetical Event

A. Wind speeds for 336 ZIP Codes are provided in the file named “FormV1Input06.xls.” The wind speeds and ZIP Codes represent a hypothetical hurricane track. The modeler is instructed to model the sample exposure data provided in the file named “FormA2Input06.xls” against these wind speeds at the specified ZIP Codes and provide the damage ratios summarized by wind speed (mph) and construction type.

The wind speeds provided are one-minute sustained 10-meter wind speeds. The sample exposure data provided consists of three structures (one of each construction type – wood frame, masonry, and mobile home) individually placed at the population centroid of each of the ZIP Codes provided. Each ZIP Code is subjected to a specific wind speed. For completing Part A, Estimated Damage for each individual wind speed range is the sum of loss to all buildings in the ZIP Codes subjected to that individual wind speed range. Subject Exposure is all exposures in the ZIP Codes subjected to that individual wind speed range. For completing Part B, Estimated Damage is the sum of the loss to all buildings of a specific type (wood frame, masonry, or mobile home) in all of the wind speed ranges. Subject Exposure is all exposures of that specific type in all of the ZIP Codes.

One base structure for each of the construction types should be placed at the population center of the ZIP Codes.

Reference Frame Structure: Reference Masonry Structure: One story One story Unbraced gable end roof Unbraced gable end roof Normal shingles (55mph) Normal shingles (55mph) ½” plywood deck ½” plywood deck 6d nails, deck to roof members 6d nails, deck to roof members Toe nail truss to wall anchor Toe nail truss to wall anchor Wood framed exterior walls Masonry exterior walls 5/8” diameter anchors at 48” centers No vertical wall reinforcing for wall/floor/foundation connections No shutters No shutters Standard glass windows Standard glass windows No door covers No door covers No skylight covers No skylight covers Constructed in 1980 Constructed in 1980 Reference Mobile Home Structure: Tie downs Single unit Manufactured in 1980

Form V-1: One Hypothetical Event

Part A

Wind Speed* (mph) Estimated Damage**/ Subject Exposure 41 – 50 0.009%-0.029% 51 – 60 0.072%-0.149% 61 – 70 0.183%-0.387% 71 – 80 0.477%-0.922% 81 – 90 1.028%-2.023% 91 – 100 2.009%-4.565% 101 – 110 4.965%-9.246% 111 – 120 9.876%-23.870% 121 – 130 23.228%-38.989% 131 – 140 42.227%-67.812% 141 – 150 66.686%-78.455% 151 – 160 75.450%-83.048% 161 – 170 77.270%-84.254%

Part B

Estimated Damage/ Construction Type Subject Exposure Wood Frame 4.465%-5.125% Masonry 3.813%-4.682% Mobile Home 5.581%-6.083%

*Wind speeds are one-minute sustained, ten-meter wind speeds. **Total estimated damage is the sum of estimated damage to all buildings in that category. For example, the total estimated damage to all buildings affected by 58 mph or greater winds or the total estimated damage to all buildings with wood frame construction.

B. Confirm that the structures used in completing the Form are identical to those in the above table. If additional non-structural assumptions are necessary to complete this Form (for example, regarding duration or surface roughness), the modeler should provide the reasons why the assumptions were necessary as well as a detailed description of how they were included.

The structures used in completing the Form are identical to those in the above table. The AIR vulnerability model requires a complete time profile of one minute sustained wind speeds in order to calculate damage for a particular risk. Therefore using just the peak wind speeds is not sufficient to calculate damage ratios. In order to provide information for Form V-1, we chose a historical storm that produces all the range of wind speeds given in Form V-1. Two scenarios were created by doubling (Case 1) and halving (Case 2) the

Form V-1: One Hypothetical Event actual forward speed of the storm while holding all other parameters constant. The range of damage ratios provided in Form V-1 are arrived at by simulating these two scenarios; that is, the lower and upper bounds correspond to Case 1 and Case 2, respectively. The results clearly show the impact of storm duration in the AIR model. The AIR model considers actual terrain surface roughness to calculate wind speeds that is used in Form V-1.

C. Provide a plot of the Form V-1 Part A data.

Figure 21: Total Loss Percentages by Wind Speed

100%

90% fast moving storm 80% slow moving storm 70% 60% 50% 40% 30% 20%

Total Loss / Subject Exposure / Subject Loss Total 10% 0% Wind Speed (mph)

Form V-2: Mitigation Measures – Range of Changes in Damage

A. Provide the change in the zero deductible personal residential reference structure damage rate (not loss cost) for each individual mitigation measure listed in Form V-2 as well as for the combination of the four mitigation measures provided for the Mitigated Frame Structure and the Mitigated Masonry Structure below.

A completed Form V-2 is provided on page 95 and on CD-ROM in both Excel and PDF formats.

B. If additional assumptions are necessary to complete this Form (for example, regarding duration or surface roughness), the modeler should provide the rationale why for the assumptions as well as a detailed description of how they are included.

No additional assumptions are necessary to complete Form V-2.

C. Provide this Form on CD-ROM in both Excel and PDF format. The file name should include the abbreviated name of the modeler, the Standards year, and the Form name. A hard copy of Form V-2 should be included in the submission.

This form is additionally provided on the accompanying CD-ROM in both Excel and PDF format.

Reference Frame Structure: Reference Masonry Structure: One story One story Unbraced gable end roof Unbraced gable end roof Normal shingles (55 mph) Normal shingles (55 mph) ½” plywood deck ½” plywood deck 6d nails, deck to roof members 6d nails, deck to roof members Toe nail truss to wall anchor Toe nail truss to wall anchor Wood framed exterior walls Masonry exterior walls 5/8” diameter anchors at 48” centers for No vertical wall reinforcing wall/floor/foundation connections No shutters No shutters Standard glass windows Standard glass windows No door covers No door covers No skylight covers No skylight covers Constructed in 1980 Constructed in 1980 Mitigated Frame Structure: Mitigated Masonry Structure: Rated shingles (110mph) Rated shingles (110mph) 8d nails, deck to roof members 8d nails, deck to roof members Truss straps at roof Truss straps at roof Plywood Shutters Plywood Shutters

Form V-2: Mitigation Measures – Range of Changes in Damage

Reference and mitigated structures are $100,000 fully insured structures with a zero deductible policy as indicated under “Owners” Policy Type for Form A-6.

Place the reference structure at the population centroid for ZIP Code 33921 located in Lee County.

Form V-2: Mitigation Measures – Range of Changes in Damage Form V-2: Mitigation Measures – Range of Changes in Damage PERCENTAGE CHANGES IN DAMAGE ((REFERENCE DAMAGE RATE - MITIGATED DAMAGE RATE) / REFERENCE DAMAGE RATE) * 100 INDIVIDUAL MITIGATION MEASURES FRAME STRUCTURE MASONRY STRUCTURE WIND SPEED (MPH) WIND SPEED (MPH) 60 85 110 135 160 60 85 110 135 160 REFERENCE STRUCTURE ⎯ ⎯ ⎯ ⎯ ⎯ ⎯ ⎯ ⎯ ⎯ ⎯

BRACED GABLE ENDS 9.3 11.9 11.0 7.5 4.2 9.2 11.8 10.9 7.5 4.2 ROOF

STRENGTH HIP ROOF 13.7 17.2 15.7 10.8 6.2 13.5 17.0 15.6 10.8 6.2

RATED SHINGLES (110 MPH) 6.2 4.7 2.8 1.5 0.7 6.2 4.7 2.8 1.5 0.7 MEMBRANE 0.0 5.6 5.7 0.0 0.0 0.0 5.8 5.9 0.0 0.0 NAILING OF DECK (8d) 2.5 5.4 6.7 5.7 3.3 2.5 5.4 6.6 5.6 3.3 ROOF COVER NG

CLIPS 1.0 4.9 10.4 14.6 14.2 1.0 4.8 10.3 14.5 14.2 STRAPS 1.0 5.9 13.0 18.3 18.2 1.0 5.9 12.8 18.2 18.1 STRENGTH

ROOF-WALL ROOF-WALL TIES OR CLIPS 0.0 0.1 0.4 1.7 4.3 0.0 0.1 0.4 1.6 4.3 STRAPS 0.0 0.4 1.3 3.9 7.8 0.0 0.4 1.3 3.8 7.8

STRENGTH WALL-FLOOR WALL-FLOOR

LARGER ANCHORS OR CLOSER 0.0 0.1 0.4 1.7 4.3 ⎯ ⎯ ⎯ ⎯ ⎯ SPACING STRAPS 0.0 0.4 1.3 3.9 7.8 ⎯ ⎯ ⎯ ⎯ ⎯ STRENGTH WALL-FOUNDATION VERTICAL REINFORCING ⎯ ⎯ ⎯ ⎯ ⎯ 0.0 0.5 1.4 4.3 9.5

PLYWOOD 6.6 9.5 9.2 6.7 4.2 6.4 9.2 8.9 6.5 4.1 WINDOW SHUTTER STEEL 6.6 9.5 9.2 6.7 4.2 6.4 9.2 8.9 6.5 4.1 ENGINEERED 9.5 14.4 14.8 11.5 7.4 9.2 13.8 14.2 11.2 7.2

OPENING PROTECTION OPENING PROTECTION DOOR AND SKYLIGHT COVERS 2.4 4.5 5.1 4.2 2.8 2.3 4.3 4.9 4.1 2.8

WINDOWS LAMINATED 4.0 5.5 5.1 3.4 2.0 3.9 5.3 4.9 3.4 2.0 IMPACT GLASS 9.0 13.9 14.4 11.2 7.2 8.7 13.3 13.8 10.9 7.0 SKYLIGHT SKYLIGHT STRENGTH

WINDOW, DOOR, WINDOW, PERCENTAGE CHANGES IN DAMAGE (REFERENCE DAMAGE RATE – MITIGATED DAMAGE RATE) /

MITIGATION MEASURES IN REFERENCE DAMAGE RATE) * 100 COMBINATION FRAME STRUCTURE MASONRY STRUCTURE WIND SPEED (MPH) WIND SPEED (MPH)

60 85 110 135 160 60 85 110 135 160

MITIGATED STRUCTURE 17.4 27.8 34.3 34.3 27.4 17.1 27.2 33.7 33.8 27.2

STRUCTURE

Form V-2: Mitigation Measures – Range of Changes in Damage

The following modification factors to the vulnerability functions are based on structural characteristics. A positive value implies a mitigation credit, while a negative value implies a debit or increase in damage.

Table 7: Modification Factors to Vulnerability Functions

OWNERS FRAME

MODIFICATION FACTORS 60 mph 85 mph 110 mph 135 mph 160 mph AVERAGE 0.4 0.5 0.5 0.4 0.4 GOOD 0.9 1.1 1.1 0.9 0.7

BU LDING POOR CONDITION -2.6 -3.1 -3.1 -2.7 -2.3 NO 0.1 0.1 0.2 0.3 0.3 RE

TREE TREE YES

EXPOSU -0.4 -1.2 -1.7 -2.0 -2.2

NO 2.3 4.1 4.9 4.6 4.1 YES SMALL SMALL DEBRIS DEBRIS

SOURCE -0.8 -0.9 -0.8 -0.6 -0.5

NO 1.5 4.1 6.2 6.6 6.0 LARGE LARGE MISSILE MISSILE SOURCE YES -0.6 -0.7 -0.7 -0.5 -0.4

FLAT 0.1 0.1 0.1 0.1 0.1

GABLE W/O BRACING 0.0 0.0 0.0 0.0 0.0

HIP 13.7 17.2 15.7 10.8 7.8

COMPLEX 2.2 2.9 2.7 2.0 1.6

STEPPED 5.5 7.2 6.7 4.8 3.6

SHED -1.0 -1.4 -1.3 -0.9 -0.7

ROOF GEOMETRY ROOF GEOMETRY MANSARD 12.3 15.7 14.7 10.5 7.7

GABLE W/BRACING 9.3 11.9 11.0 7.5 5.4

PYRAMID 11.9 15.0 14.0 10.0 7.4

GAMBREL 5.1 6.7 6.3 4.6 3.5

LOW -2.4 -3.2 -3.0 -2.2 -1.6

MEDIUM -0.3 -0.5 -0.5 -0.3 -0.2

ROOF PITCH HIGH 2.3 3.2 3.0 1.9 1.3

ASPHALT SHINGLES 0.0 0.0 0.0 0.0 0.0

WOODEN SHINGLES -0.6 -0.5 -0.3 -0.3 -0.3

CLAY/CONCRETE TILES 3.1 2.6 1.8 1.0 0.7

LIGHT METAL PANELS -6.1 -5.3 -3.5 -2.0 -1.3

SLATE 5.5 4.4 2.7 1.4 0.9

BUILT-UP ROOF WITH GRAVEL 6.5 5.1 3.1 1.6 1.1

SINGLE PLY MEMBRANE 5.8 4.6 2.9 1.5 1.0 ROOF COVERING STANDING SEAM METAL ROOFS 6.1 4.8 3.0 1.5 1.0

BUILT-UP ROOF WITHOUT GRAVEL 6.2 5.0 3.2 1.7 1.1

SINGLE PLY MEMBRANE BALLASTED 6.2 4.7 2.8 1.5 1.0

FBC EQUIVALENT 6.2 4.7 2.8 1.5 1.0

PLYWOOD 0.0 0.0 0.0 0.0 0.0

WOOD PLANKS 2.5 3.7 3.6 2.5 1.8

PARTICLE BOARD/OSB 1.2 1.9 1.8 1.2 0.9 ROOF DECK METAL DECK W/ INSULATION BOARD 2.3 3.7 3.7 2.6 1.8 96 Information Submitted in Compliance with the 2006 Standards of the Florida Commission on Hurricane Loss Projection Methodology © 2007 AIR Worldwide Corporation

Form V-2: Mitigation Measures – Range of Changes in Damage METAL DECK W/ CONCRETE 2.5 3.9 4.4 3.9 3.3

PRE-CAST CONCRETE SLABS 2.6 4.1 4.5 3.8 3.3

REINFORCED CONCRETE SLABS 2.7 4.1 4.4 3.8 3.3

LIGHT METAL -2.9 -3.7 -3.1 -1.8 -1.3

SCREWS 2.3 2.0 1.3 0.7 0.5

NAILS/STAPLES -0.6 -0.5 -0.3 -0.2 -0.1

ADHESIVE/EPOXY -3.1 -2.4 -1.5 -0.8 -0.6 ROOF COVER ATTACHMENT MORTAR -0.6 -0.5 -0.3 -0.2 -0.1

SCREWS/BOLTS 2.5 5.4 6.7 5.7 4.3

NAILS/STAPLES 0.0 0.0 0.0 0.0 0.0

ADHESIVE/EPOXY -1.8 -3.6 -4.4 -3.9 -2.9

STRUCTURALLY CONNECTED 2.9 5.8 7.0 5.9 4.5

6D NAILS @ 6" SPACING, 12" OC 0.0 0.0 0.0 0.0 0.0

8D NAILS @ 6" SPACING, 12" OC 2.5 5.4 6.7 5.7 4.3 ROOF DECK ATTACHMENT 8D NAILS @ 6" SPACING, 6" OC 2.5 5.4 6.7 5.7 4.3

HURRICANE TIES 1.0 5.9 13.0 18.3 19.1

NAILS/SCREWS 0.0 0.0 0.0 0.0 0.0

ANCHOR BOLTS 1.0 4.9 10.4 14.6 15.0

GRAVITY/FRICTION -0.4 -2.7 -6.1 -8.7 -8.9

ADHESIVE/EPOXY 0.3 -0.6 -2.3 -3.5 -3.6

ROOF ANCHORAGE STRUCTURALLY CONNECTED 1.0 5.9 13.0 18.5 19.5

CLIPS 1.0 4.9 10.4 14.6 15.0

BRICK/UNREINFORCED MASONRY -2.4 -2.7 -2.4 -1.8 -1.4

REINFORCED MASONRY 2.5 3.0 2.9 2.3 1.8

PLYWOOD 0.0 0.0 0.0 0.0 0.0

WOOD PLANKS 2.5 2.9 2.7 2.0 1.5

PARTICLE BOARD/OSB 0.7 0.9 0.9 0.6 0.5

WALL TYPE METAL PANELS -3.6 -4.2 -3.9 -2.9 -2.2

PRE-CAST CONCRETE ELEMENTS 2.5 3.0 2.9 2.3 1.8

CAST-IN-PLACE CONCRETE 2.5 3.0 2.9 2.3 1.8

GYPSUM BOARD -3.2 -3.5 -2.9 -2.0 -1.5

VENEER BRICK/MASONRY 2.7 2.5 1.7 1.0 0.7

WOOD SHINGLES 0.5 0.4 0.2 0.1 0.1

CLAPBOARDS 1.6 1.6 1.2 0.6 0.4

ALUMINUM/VINYL SIDING -5.0 -4.8 -3.5 -1.9 -1.2

WALL S D NG STONE PANELS 3.3 2.8 1.9 1.2 0.8

EIFS -12.0 -10.6 -7.3 -4.4 -3.1

STUCCO -3.6 -3.4 -2.3 -1.3 -0.8

ANNEALED 0.0 0.0 0.0 0.0 0.0

TEMPERED 3.2 4.0 3.5 2.2 1.6

HEAT STRENGTHENED 2.6 3.1 2.6 1.7 1.3

GLASS TYPE TYPE GLASS LAMINATED 4.0 5.5 5.1 3.4 2.5

INSULATING 2.1 2.5 2.2 1.5 1.1

LESS THAN 5 0.7 1.0 1.0 0.8 0.6

BETWEEN 5 AND 20 0.3 0.3 0.3 0.2 0.1

BETWEEN 20 AND 60 -2.4 -3.6 -3.5 -2.4 -1.7

GLASS PERCENT PERCENT GLASS GREATER THAN 60 -3.4 -5.0 -4.8 -3.2 -2.2 Information Submitted in Compliance with the 2006 Standards of the 97 Florida Commission on Hurricane Loss Projection Methodology © 2007 AIR Worldwide Corporation

Form V-2: Mitigation Measures – Range of Changes in Damage NONE 0.0 0.0 0.0 0.0 0.0

NON-ENGINEERED 6.6 9.5 9.2 6.7 5.0 WINDOW

PROTECTION ENGINEERED 9.5 14.4 14.8 11.5 8.9

SINGLE WIDTH 0.2 0.4 0.6 0.5 0.4

DOUBLE WIDTH -1.3 -2.7 -3.3 -2.8 -2.3

REINFORCED SINGLE WIDTH 1.2 2.7 3.5 3.3 2.8

REINFORCED DOUBLE WIDTH 0.4 0.9 1.2 1.1 1.0

EXTERIOR DOORS SLIDING -1.6 -3.5 -4.3 -3.9 -3.3

REINFORCED SLIDING -0.2 -0.4 -0.5 -0.5 -0.4

HURRICANE TIES 0.0 0.4 1.3 3.9 6.2

NAILS/SCREWS 0.0 0.0 0.0 0.0 0.0

ANCHOR BOLTS 0.0 0.1 0.4 1.7 3.1

GRAVITY/FRICTION 0.0 -0.3 -1.0 -3.0 -4.8 CONNECTION ADHESIVE/EPOXY 0.0 -0.1 -0.4 -1.5 -2.7 BU LDING FOUNDATION STRUCTURALLY CONNECTED 0.0 0.5 1.4 4.3 7.3

CHIMNEY 0.9 1.5 1.3 0.7 0.4

A/C UNIT 0.3 0.4 0.3 0.2 0.1

SKYLIGHTS 0.0 0.0 0.0 0.0 0.0

PARAPET WALLS 2.1 2.3 1.5 0.5 0.1

OVERHANG/RAKE (8 TO 36 INCHES) 0.9 1.5 1.3 0.7 0.4

DORMERS 0.7 1.3 1.3 0.8 0.4

OTHER 1.0 1.5 1.3 0.7 0.4

NO ATTACHED STRUCTURES 2.0 2.9 2.6 1.6 1.1

OVERHANG/RAKE (LESS THAN 8 INCHES) 1.9 2.7 2.4 1.5 1.0 ROOF ATTACHED STRUCTURES OVERHANG/RAKE (GREATER THAN 36 INCHES) 0.8 1.3 1.2 0.8 0.5

WATERPROOF MEMBRANE/FABRIC 2.0 2.9 2.7 1.7 1.2

SECONDARY WATER RESISTANCE 0.0 5.6 5.7 0.0 0.0

CARPORTS/CANOPIES/PORCHES -1.8 -2.5 -2.4 -1.8 -1.3

SINGLE DOOR GARAGE 0.4 0.6 0.5 0.3 0.2

DOUBLE DOOR GARAGE -2.9 -3.9 -3.7 -2.9 -2.3

REINFORCED SINGLE DOOR GARAGE 1.8 2.8 2.5 1.5 0.9

REINFORCED DOUBLE DOOR GARAGE 0.4 0.6 0.5 0.3 0.2

SCREENED PORCHES/GLASS PATIO DOORS -2.2 -2.9 -2.7 -2.1 -1.6

BALCONY -1.7 -2.4 -2.2 -1.6 -1.2 WALL ATTACHEDSTRUCTURES NO ATTACHED WALL STRUCTURES -0.1 -0.1 -0.1 0.0 0.0

DETACHED GARAGE -0.3 -0.6 -0.8 -0.7 -0.6

POOL ENCLOSURES -13.7 -20.9 -16.6 -6.0 -1.3

NO POOL ENCLOSURES 1.5 1.9 1.8 1.3 1.0

SHED -0.3 -0.6 -0.8 -0.7 -0.6

MASONRY BOUNDARY WALL -0.3 -0.6 -0.8 -0.7 -0.6

OTHER FENCE -0.3 -0.6 -0.8 -0.7 -0.6

APPERTENANT STRUCTURES NO APPURTENANT STRUCTURES 1.9 2.5 2.3 1.6 1.3

0-10 YEARS OLD 2.0 3.0 2.8 1.6 0.9

11-20 YEARS OLD 0.4 0.5 0.5 0.3 0.2

ROOF BU LT 21+ YEARS OLD -3.1 -3.8 -3.6 -2.7 -2.1

Form V-3: Mitigation Measures – Mean Damage Ratio Trade Secret List Item

A. Provide the mean damage ratio to the reference structure for each individual mitigation measure listed in Form V-3 as well as the percent damage for the combination of the four mitigation measures provided for the Mitigated Frame Structure and the Mitigated Masonry Structure below.

B. If additional assumptions are necessary to complete this Form (for example, regarding duration or surface roughness), the modeler should provide the rationale for the assumptions as well as a detailed description of how they are included.

C. Provide a graphical representation of the vulnerability curves for the reference structure and the fully mitigated structure.

Reference Frame Structure: Reference Masonry Structure: One story One story Unbraced gable end roof Unbraced gable end roof Normal shingles (55mph) Normal shingles (55mph) ½” plywood deck ½” plywood deck 6d nails, deck to roof members 6d nails, deck to roof members Toe nail truss to wall anchor Toe nail truss to wall anchor Wood framed exterior walls Masonry exterior walls 5/8” diameter anchors at 48” centers for No vertical wall reinforcing wall/floor/foundation connections No shutters No shutters Standard glass windows Standard glass windows No door covers No door covers No skylight covers No skylight covers Constructed in 1980 Constructed in 1980

Mitigated Frame Structure: Mitigated Masonry Structure: Rated shingles (110mph) Rated shingles (110mph) 8d nails, deck to roof members 8d nails, deck to roof members Truss straps at roof Truss straps at roof Plywood Shutters Plywood Shutters

Reference and mitigated structures are $100,000 fully insured structures with a zero deductible policy as indicated under “Owners” Policy Type for Form A-6.

Place the reference structure at the population centroid for ZIP Code 33921 located in Lee County.

Form V-3 will be presented at the closed meeting.

Standard A-1: Modeled Loss Costs

2006 Actuarial Standards

A-1 Modeled Loss Costs

Modeled loss costs shall reflect all damages from storms that reach hurricane strength and produce minimum damaging wind speeds or greater on land in Florida.

Modeled loss costs reflect all damages from storms that reach hurricane strength and produce minimum damaging wind speeds or greater on land in Florida.

Disclosure

1. Describe how damage from model generated storms (landfalling and bypassing) is excluded or included in the calculation of loss costs for the state of Florida.

The calculation of loss costs includes the losses from all hurricanes that make landfall in Florida or are Florida by-passers. Damage is included in the calculation of loss costs from the time the hurricane first causes damaging wind speeds on land in Florida.

Standard A-2: Underwriting Assumptions

A-2 Underwriting Assumptions

A. When used in the modeling process or for verification purposes, adjustments, edits, inclusions, or deletions to insurance company input data used by the modeler shall be based upon accepted actuarial, underwriting, and statistical procedures.

Any adustments, edits, inclusions or deletions made to client company input data are based upon accepted actuarial underwriting and statistical procedures.

B. For loss cost estimates derived from or validated with historical insured hurricane losses, the assumptions in the derivations concerning (1) construction characteristics, (2) policy provisions, (3) claim payment practices, and (4) relevant underwriting practices underlying those losses, as well as any actuarial modifications, shall beappropriate.

The typical construction types provided in the loss data include Frame, Masonry Veneer, Masonry, and Mobile Home. Coverage losses, if included, are typically by Coverage A (building), Coverage C (contents) and Coverage D (additional living expenses). AIR uses this detailed loss data to validate the vulnerability functions used in the model. All applicable policy provisions including limits, deductibles, coinsurance, etc. are taken into consideration and modeled appropriately. AIR discusses and documents all assumptions related to validation data with our clients. If data is excluded or adjusted, this is documented in the PIAF, which is included as Attachment A of this submission, and in the project file.

Disclosures

1. Identify the assumptions used to develop loss costs for unknown residential construction types.

The vulnerability curve used for unknown residential construction types is obtained from a weighted averaging of the vulnerability curves of the individual residential construction classes. For example, for residential single-family dwellings we utilize eight construction classes. These are standard construction types provided by our client companies. The composite vulnerability curve, used for unknown, is based on a weighted average of each of the different construction classes. Note that our unknown residential function does not include mobile homes. Our clients are almost always able to supply mobile home exposures separately. Since the vulnerability of mobile homes is much greater than other construction types, it would not be appropriate to use this construction class in the composite function for residential “unknown”. The same methodology is used for the apartment and condominium composite vulnerability functions.

2. Describe how the modeled loss costs take into consideration storm surge and flood damage to the infrastructure.

Standard A-2: Underwriting Assumptions

The storm surge component of the AIR hurricane model generally follows that of FEMA’s “Coastal Flooding: Hurricane Storm Surge Model,” 1985. It is a fully probabilistic simulation methodology, using a set of values stochastically drawn from the probability distributions governing key meteorological, geographical, and oceanographic parameters. These include central barometric pressure, radius of maximum winds, forward speed, storm heading at landfall, coastline orientation, coastal elevation, tide height, and bathymetry, or water depth.

The storm surge component of the model can be in either “on” or “off” position. Storm surge loss was assumed to be 0% (i.e., in the “off” position) for purposes of this submission.

AIR does not explicitly model flood damage to infrastructure.

3. Describe the assumptions included in model development and validation concerning insurance company claim payment practices.

Clients are asked to separately disclose any non-wind hurricane damage and these losses are excluded.

4. Identify depreciation assumptions and describe the methods and assumptions used to reduce insured losses on account of depreciation. Provide a sample calculation for determining the amount of depreciation and the actual cash value (ACV) losses.

The model requires replacement value and insured limit by coverage. The model makes no depreciation assumptions. Insurers may make specific assumptions for any depreciation adjustments that reduce replacement value to actual cash value. In such cases, the actual cash value is entered for the replacement value and thus the damage ratio is applied to the actual cash value, or depreciated replacement value. Loss amounts are then capped at insured limit by coverage. If AIR makes assumptions regarding depreciation and ACV at a client’s request, such assumptions are included in the PIAF.

Table 8: Sample Calculation for Determining Depreciation and ACV Losses

Actual Input Coverage A limit $100,000 $100,000 Coverage C limit $50,000 $50,000 ACV (Coverage C) $25,000 Replacement Value C $100,000 $25,000 Percent of Contents Damaged 1% Contents Losses $250 5. Identify property value assumptions and describe the methods and assumptions used to determine the true property value and associated losses. Provide a sample calculation for determining the property value and guaranteed replacement cost losses.

Standard A-2: Underwriting Assumptions

The model requires replacement value and insured limit by coverage. Damages are calculated based on replacement value and are capped at the policy limit. Insurers may determine replacement value from insured limit and insurance-to-value assumptions. If AIR makes assumptions regarding property value at a client’s request, such assumptions are included in the PIAF.

If the replacement value is greater than the policy limit and a guaranteed replacement cost factor (GRC) applies, the model can calculate guaranteed replacement cost losses by inputting Policy Limit × GRC for Policy Limit. Note that modeled losses cannot be greater than the specified replacement value.

Table 9: Sample Calculation for Determining Property Value and Guaranteed Replacement Cost Losses

Actual Input Replacement Value $120,000 $120,000 Policy Limit $100,000 $120,000 GRC Factor 1.2

Scenario A* Scenario B* Damage Ratio .75 .9 Modeled Loss $90,000 $108,000 Insured Loss (with GRC) $90,000 $108,000 * ignoring application of deductible

6. Describe how loss adjustment expenses are considered within the loss cost estimates.

The modeled losses include losses to the primary structure, the appurtenant structures, contents and additional living expenses. Actual loss data from client companies used for validation purposes does not include LAE. If provided, it is identified separately and not used in the validation.

Standard A-3: Loss Cost Projections

A-3 Loss Cost Projections

A. Loss cost projections produced by hurricane loss projection models shall not include expenses, risk load, investment income, premium reserves, taxes, assessments, or profit margin.

AIR’s hurricane model produces pure loss estimates. Model loss costs do not include risk load, investment income, premium reserves, taxes, assessment or profit margin.

B. Loss cost projections shall not make a prospective provision for economic inflation.

The model does not make a prospective provision for economic inflation. Insurer inforce exposures or projected exposures are input to the model.

Disclosures

1. Describe the method or methods used to estimate annual loss costs needed for ratemaking. Identify any source documents used and research performed.

For a given set of exposures by coverage, ZIP Code or other geographical grouping, and by construction type, losses are estimated and aggregated for thousands of simulated storms. Average annual losses by location, by construction type are calculated by dividing the total losses net of policy conditions for all simulated storms by the number of iterations. Loss costs for a given territory or county are calculated by dividing the average annual losses for all locations within a territory or county by the corresponding exposures.

2. Identify the highest level of resolution for which loss costs can be provided. Identify the resolution used for the reported output ranges.

Loss costs can be provided at any geographic level desired: state, county, ZIP Code, rating territory, latitude, longitude square or specific latitude, longitude point. The reported output ranges use 5-digit ZIP Code resolution.

Standard A-4: Demand Surge

A-4 Demand Surge (*New Standard)

A. Demand surge shall be included in the model’s calculation of loss costs

AIR's modeling of loss costs contained herein reflects use of a function to account for the effects of temporary inflation that can result from increased demand for materials and services to repair and rebuild damaged property after a major catastrophe event ("demand surge").

B. The methods, data, and assumptions used in the estimation of demand surge shall be actuarially sound.

The methods, data, and assumptions used in the estimation of demand surge are actuarially sound. The demand surge function is derived by analyzing component-level building damage and economic data reflecting component repair costs for historical events. However, the demand surge function is estimated based on very few historical data points and therefore subject to significant uncertainty.

Disclosures

1. Describe how the model incorporates demand surge in the calculation of loss costs.

Historical evidence from major catastrophic events in the past 15 years suggests that after a major event, increased demand for materials and services to repair and rebuild damaged property can put pressure on prices, resulting in temporary inflation. This phenomenon is often referred to as demand surge and it results in increased losses to insurers.

Key Factors Leading to Demand Surge: Sudden Increase in Demand A catastrophic event causes widespread damage to property, which leads to a sharp increase in the need for building material and services following the event. Demand also increases sharply for resources such as labor, transportation, equipment and storage in the affected area. Resource availability in any regional economy is typically sufficient to accommodate normal demand, even taking into account some buffer. However, an unexpected increase in demand can lead to shortages and price increases.

Time Element Losses When there is widespread damage to property, low regional capacity to meet the increased demand can result in longer than normal repair times. This, in turn, results in greater business interruption losses and additional living expenses. Infrastructure damage, delayed building permit processes and a shortage of available building inspectors are also factors in increasing time element loss.

How the Model Incorporates Demand Surge:

Standard A-4: Demand Surge

AIR has related the amount of demand surge in a particular event to the amount of total industry insurable losses from the event. The factor is dependent on coverage. A table incorporated into the software contains the corresponding demand surge factors, by coverage, for industry losses. For a given event, the demand surge factors by coverage are applied to the corresponding ground-up losses, based on the industry loss for that event. Policy conditions are then applied probabilistically. The sum of these losses by coverage yields the total event loss with demand surge included. Very few data points exist to create and validate a demand surge curve resulting in significant uncertainty related to the level of demand surge in an event. This uncertainty is illustrated by examining the sensitivity of the demand surge curve. For example, by increasing the demand surge curve by 10%, the resulting average annual losses increase by 6.5%.

2. Provide citations to published papers, if any, that were used to develop how the model estimates demand surge.

No published papers on demand surge were used to develop how the model estimates demand surge.

Standard A-5: User Inputs

A-5 User Inputs

All modifications, adjustments, assumptions, and defaults, and treatments of missing values necessary to use the inputs used in the model shall be actuarially sound and included with the model output. Treatment of missing values for user inputs required to run the model shall be actuarially sound and described with the model output.

Assumptions that relate to an insurer’s input data are identified on the analysis option form, an output report of CLASIC/2™. When AIR performs analyses for client, the company also provides the (PIAF) as included in Attachment A. Assumptions related to any missing values are detailed in the PIAF. The PIAF is given to clients and requires approval prior to the hurricane analysis. It is also included with the final report.

Insurance to Value The model requires replacement value and insured limit by coverage. Damages are calculated based on replacement value and are capped at the policy limit. Insurers may determine replacement value from insured limit and insurance-to-value assumptions. This calculated value could then be input as replacement value. If AIR makes assumptions regarding property value at a client’s request, such assumptions are included in the PIAF.

Demographic Assumptions AIR defaults assume average structural characteristics, maintenance, state of repair, etc.

Appurtenant Structures The model requires replacement value and insured limit by coverage. Insurers provide actual replacement value and insured limit for appurtenant structures. If AIR makes assumptions regarding appurtenant structures at a client’s request, such assumptions are included in the PIAF.

Contents The model requires replacement values and insured limits by coverage. Damages are calculated based on the replacement value for the contents (which may be higher or lower than the limit) and are capped at the contents policy limit. The model uses insurer provided replacement values. Insurers may determine replacement value from insured limit and insurance-to-value assumptions. This calculated value could then be input as replacement value. If AIR makes assumptions regarding property value at a client’s request, such assumptions are included in the PIAF.

Additional Living Expenses The AIR model does not use the additional living expense limits to calculate damages. Losses due to Additional Living Expense (Time Element) coverage are estimated using the mean building damage, the time estimated to make repairs or to reconstruct the damaged building and the estimated cost of ALE per time period. ALE losses are then capped using the limits. Therefore, changing the ALE limits will not affect the loss calculation unless the ALE damage is already above the limit.

Standard A-5: User Inputs

Insurer Exposures By ZIP Code Insurer exposure data is checked for valid ZIP Codes before running the hurricane model. ZIP Codes are checked for validity by comparing insurer supplied ZIP Codes with AIR’s most recent ZIP Code database. If an insurer ZIP Code cannot be matched to the current ZIP Code database, AIR’s database of historical ZIP Codes will be searched. This database includes ZIP Codes that were valid in previous years but later changed. ZIP Codes which are matched to a valid ZIP Code in a prior year are reassigned to the appropriate currently valid ZIP Code.

The model produces a list of ZIP Codes that do not appear in AIR’s master database of valid ZIP Codes. Exposure in any invalid ZIP Code is not modeled. The insurer may choose to allocate or map these exposures back into the exposure data set; this is not part of the model. If AIR makes assumptions regarding allocation or remapping at a client’s request, such assumptions are included in the PIAF.

Disclosures

1. Describe the methods used to distinguish among policy form types (e.g., homeowners, dwelling property, mobile home, tenants, condo unit owners).

The AIR hurricane model can distinguish among any policy form type. Exposures are distinguished based on vulnerability characteristics, such as construction type and occupancy. Policy form is carried as a reporting field and not explicitly modeled. As discussed in section I.A.7, for all policy forms, losses are estimated separately by coverage for coverage A, B, C and D, or any combination. The model can produce loss costs for defined groups of policies if vulnerability characteristics are known. Such characteristics are described in Standard V-1.5.

2. Disclose, in a model output report, the specific type of input that is required to use the model or model output in a personal residential property insurance rate filing. Such input includes, but is not limited to, optional features of the model, type of data to be supplied by the model user and needed to derive loss projections from the model, and any variables that a model user is authorized to set in implementing the model. Include the model name and version number on the model output report. All items included in the output form submitted to the Commission should be clearly labeled and defined.

The UNICEDE®/px Preparer’s Guide describes the data format used to transfer detailed exposure information to the CLASIC/2™ software, and the data elements necessary for deriving loss estimates from the model. However, clients are not required to submit their data to AIR in this format and, as such, there is no input form for exposure data. AIR will remap client data to the UNICEDE/px format as necessary. A copy of the Preparer’s Guide will be available to the Professional Team during its on-site visit.

Standard A-5: User Inputs

The CLASIC/2™ User’s Guide describes the analysis options that may be selected for generating loss results, including variables that a user may set in implementing the model. The analysis options appear as header information in all loss exports provided by the software. They may also be exported as a text file. A sample text file is included in the PIAF (Attachment A). A copy of the User’s Guide will be available to the Professional Team during its on-site visit.

3. Provide a copy of the input form used by a model user to provide input criteria to be used in the model. The modeler should demonstrate that the input form relates directly to the model output. Include the model name and version number on the input form. All items included in the input form submitted to the Commission should be clearly labeled and defined.

If the client has contracted with AIR on a service basis rather than licensing the software, all modifications, adjustments, or assumptions are noted on the PIAF. This form is presented to the client for approval before modeling begins and a copy is also included with the Final report. The PIAF (see Attachment A), relates directly to model input requirements and output to be produced. It includes the model name and version number.

4. Describe actions performed to ensure the validity of insurer data used for model inputs or validation/verification.

Insurer data, whether used as the exposure input to a loss analysis or for model validation, undergoes a set of structured processing procedures. These include checks to determine the quality and completeness of the data, its reasonability, and the existence of any unique conditions or special reporting features. If data is excluded or adjusted, this is noted in the Project Information and Assumptions Form (see Attachment A). Insurer approval of the form is required prior to analysis.

Standard A-6: Logical Relationships to Risk

A-6 Logical Relationship to Risk

A. Loss costs shall not exhibit an illogical relation to risk, nor shall loss costs exhibit a significant change when the underlying risk does not change significantly.

The AIR model produces loss costs that are logical in relation to risk.

B. Loss costs produced by the model shall be positive and non-zero for all valid Florida ZIP Codes.

The loss costs are positive and non-zero for all ZIP Codes. Loss cost maps by ZIP Code are provided in Form A-2.

C. Loss costs cannot increase as the quality of construction type, materials and workmanship increases, all other factors held constant.

Loss costs do not increase as the quality of construction, material or workmanship increases, all else being equal. Loss cost maps for wood, masonry, and mobile home construction types are provided in Form A-2.

D. Loss costs cannot increase as the presence of fixtures or construction techniques designed for hazard mitigation increases, all other factors held constant.

Loss costs do not increase as the presence of mitigation devices or techniques increases, all else being equal.

E. Loss costs cannot increase as the quality of building codes and enforcement increases, all other factors held constant.

Loss costs do not increase as the quality of building codes and enforcement increases, all else being equal.

F. Loss costs shall decrease as deductibles increase, all other factors held constant.

Loss costs do decrease as deductibles increase, all else being equal.

Standard A-6: Logical Relationships to Risk

G. The relationship of loss costs for individual coverages (e.g., structures and appurtenant structures, contents, and loss of use/additional living expense) shall be consistent with the coverages provided.

The relationship of losses for building, appurtenant structures, contents, and additional living expense to the total loss as produced by the model is reasonable, as demonstrated by comparing the relationships between actual and modeled losses for historical events.

Table 10: Comparison between Actual and Modeled Losses for Different Coverages

Cov A Cov B Cov C Cov D Total Actual 0.744 0.074 0.143 0.038 1.000 Modeled 0.732 0.073 0.133 0.062 1.000

Standard A-6: Logical Relationships to Risk

Disclosures

1. Demonstrate that loss cost relationships by type of coverage (structures, appurtenant structures, contents, additional living expenses) are consistent with actual insurance data.

The graphs below show actual versus simulated damage ratios for coverage A for available company-storm combinations. The actual ratios reflect actual loss data in multiple ZIP Codes for multiple storms. The variability in the modeled mean damage ratio reflects the impact of duration of damaging wind speeds at the corresponding ZIP Codes. The lower of the pairs of graphs in this and the following examples simply reverses the order of the series for purposes of revealing the data beneath.

Figure 22: Actual and Modeled Damage Ratios vs Wind Speed, Coverage A

Actual and Modeled Damage Ratios vs Wind Speed, CovA Damage Ratio

Actual Simulated

Wind Speed (mph)

Actual and Modeled Damage Ratios vs Wind Speed, CovA Damage Ratio

Simulated Actual

Wind Speed (mph)

Standard A-6: Logical Relationships to Risk

The graphs below show actual versus simulated damage ratios for coverage C for available company storm combinations.

Figure 23: Actual and Modeled Damage Ratios vs Wind Speed, Coverage C

Actual and Modeled Damage Ratios vs Wind Speed, CovC Damage Ratio

Actual Simulated

Wind Speed (mph)

Actual and Modeled Damage Ratios vs Wind Speed, CovC Damage Ratio

Simulated Actual

Wind Speed (mph)

Standard A-6: Logical Relationships to Risk

The time element losses are a function of the time it takes to repair/reconstruct a damaged building. The following graphs display actual and simulated damage ratios for Coverage D for available company storm combinations.

Figure 24: Actual and Modeled Damage Ratios vs Wind Speed, Coverage D

Actual and Modeled Damage Ratios vs Wind Speed, CovD

Damage Ratio Actual Simulated

Wind Speed (mph)

Actual and Modeled Damage Ratios vs Wind Speed, CovD Damage Ratio Damage Simulated Actual

Wind Speed (mph)

Standard A-6: Logical Relationships to Risk

2. Demonstrate that loss cost relationships by construction type or vulnerability function (frame, masonry, and mobile home) are consistent with actual insurance data.

The graphs below show actual versus simulated damage ratios for mobile homes for available company storm combinations.

Figure 25: Actual and Modeled Damage Ratios vs Wind Speed, Mobile Home

Actual and Modeled Damage Ratios vs Wind Speed, Mobile Home

Actual Damage Ratio Simulated

Wind Speed (mph)

Actual and Modeled Damage Ratios vs Wind Speed, Mobile Home

Damage Ratio Simulated Actual

Wind Speed (mph)

Standard A-6: Logical Relationships to Risk

The graphs below show actual versus simulated damage ratios for frame construction for available company storm combinations.

Figure 26: Actual and Modeled Damage Ratios vs Wind Speed, Frame

Actual and Modeled Damage Ratios vs Wind Speed, Frame Damage Ratio

Actual Simulated

Wind Speed (mph)

Actual and Modeled Damage Ratios vs Wind Speed, Frame Damage Ratio

Simulated Actual

Wind Speed (mph)

Standard A-6: Logical Relationships to Risk

The graphs below show actual versus simulated damage ratios for masonry construction for available company storm combinations.

Figure 27: Actual and Modeled Damage Ratios vs Wind Speed, Masonry

Actual and Modeled Damage Ratios vs Wind Speed, Masonry Damage Ratio

Actual Simulated

Wind Speed (mph)

Actual and Modeled Damage Ratios vs Wind Speed, Masonry Damage Ratio

Simulated Actual

Wind Speed (mph)

3. Demonstrate that loss cost relationships among coverages, territories, and regions are consistent and reasonable.

The relationship between loss costs of different regions or territories is most sensitive to the frequency of hurricanes striking that region or territory and the distance from the coastline of the particular region or territory. ZIP Codes in geographic areas with historically high frequencies of hurricanes will have higher loss costs than ZIP Codes in areas of lower frequencies. Similarly, within areas of high or low frequency, ZIP Codes nearer the coastline will have higher loss costs than ZIP Codes away from the coastline.

Standard A-6: Logical Relationships to Risk

The maps in Form A-2 show the simulated average annual loss costs by ZIP Code. The exposure assumptions are consistent with the Output Range specifications, and loss costs are related to the Coverage “A” amount. ZIP Codes along the coastline have higher loss costs than do inland ZIP Codes. Additionally, adjacent ZIP Codes are generally within one range of average loss costs. That is, there are smooth transitions between ZIP Codes.

4. Explain any anomalies or special circumstances that might preclude any of the above conditions from occurring.

At the level of precision modeled, there are no anomalies or special circumstances that preclude loss costs from exhibiting logical relationship to risk. Site-specific conditions can vary and may not be completely known. The model captures uncertainty in local intensity and damageability by providing a mean damage ratio and a corresponding probability distribution.

5. Provide a completed Form A-1, Loss Costs.

Form A-1 is provided on CD-ROM in both Excel and PDF format. Losses are provided by coverage net of deductible. This year the policy deductible is applied in proportion to ground-up loss by coverage.

6. Provide a completed Form A-2, Zero Deductible Loss Costs by ZIP Code.

Form A-2 is provided on CD-ROM in PDF format, and is also included on page 135.

7. Provide a completed Form A-3, Base Storm Hurricane Set Average Annual Zero Deductible Statewide Loss Costs.

Form A-3 is provided on CD-ROM in both Excel and PDF format, and is also included on page 137.

8. Provide a completed Form A-4, Hurricane Andrew Percent of Losses.

Form A-4 is provided on CD-ROM in both Excel and PDF format, and is also included on page 139.

9. Provide a completed Form A-5, Distribution of Hurricanes by Size of Loss.

Form A-5 is provided on CD-ROM in both Excel and PDF format, and is also included on page 143.

Standard A-7: Deductibles and Policy Limits

A-7 Deductibles and Policy Limits

A. The methods used in the development of mathematical distributions to reflect the effects of deductibles and policy limits shall be actuarially sound.

The AIR damageability functions generate a mean damage ratio for a given wind speed.

For any estimated mean damage ratio, there is a mixed probability distribution, f D , that includes finite probabilities of damage at zero and 100 percent. This representation of the damage ratio fits well with observed data. A sample distribution is shown below:

Figure 28: Probability Distribution around the Mean Damage Ratio

f D (x)

Thus, the effects of deductibles, coinsurance and other policy conditions can be properly calculated as illustrated in the sample insured loss calculation shown below: 1 Expected Insured Loss = f (x) max 0, Coins% ∗ min x ∗ R , P - DED dx ∫ D {}[]()V L x=0 where Coins% = Coinsurance Percentage RV = Replacement Value PL = Policy Limit DED = Deductible x = Random variable for damage ratio

In application, f D (x) is discretized and numerical integration is used to estimate the expected insured loss.

B. The relationship among the modeled deductible loss costs shall be reasonable.

The relationship among the modeled deductible loss costs is reasonable. Loss costs do decrease as deductibles increase.

Standard A-7: Deductibles and Policy Limits

C. Deductible loss costs shall be calculated in accordance with s. 627.701(5)(a), F.S.

The AIR model explicitly enables the application of seasonal deductibles in accordance with s.627.701(5)(a),F.S. The statute requires the application of the hurricane deductible to the first event, and the greater of the remaining deductible and the all other perils deductible to losses from subsequent events in the same calendar year.

Standard A-7: Deductibles and Policy Limits

Disclosures

1. Describe the methods used in the model to treat deductibles (both flat and percentage), policy limits, replacement costs, and insurance-to-value when projecting loss costs.

For any estimated mean damage, there is a probability distribution around that mean. Thus, expected damages above different deductible levels can be readily calculated. Flat dollar deductibles are applied directly while deductibles that are a percentage of coverage amount are converted to the flat dollar equivalent and applied. The model calculates losses to replacement costs and caps losses at policy limits. Insurance-to-value is addressed by the user and identified in the PIAF (see Attachment A).

2. Provide an example of how insurer loss (loss net of deductibles) is calculated. Discuss data or documentation used to confirm or validate the method used by the model.

The following are examples of how insurer losses (net of deductibles) are calculated.

Table 11: Calculating Losses Net of Deductibles (D)=(A)*(C) (A) (C) Zero (E) Structure Policy (B) Damage Deductible Loss Net of Value Limit Deductible Ratio Loss Deductible* 100,000 90,000 500 2% 2,000 1,688 100,000 90,000 500 5% 5,000 4,573

*The calculation of the Loss Net of Deductible is based on actuarial theory of deductibles and limits as illustrated in the Expected Insured Loss equation on page 119. The literature source is:

Hogg, R.V. and Klugman, S. A., Loss Distributions, John Wiley and Sons, 1984.

The distributions are validated using engineering studies and actual loss data.

3. Describe how the model calculates annual deductibles.

In the AIR model, annual frequency of occurrence has always been a key model parameter. Each simulated year may have zero, one, or multiple events, just as might be observed in an actual year. This approach for generating the event catalog makes it straightforward and to determine the probability of multiple-event seasons. The functionality to calculate losses net of an annual deductible was enabled during 2005 in CLASIC/2. Prior to this, the user would have performed the calculations outside the software. The deductible is applied in CLASIC/2 as follows: for the first hurricane event in a year, apply the hurricane

Standard A-7: Deductibles and Policy Limits deductible to the loss distribution. Calculate the “applicable deductible” for this event (see definition below), and then calculate the “remaining deductible” as (hurricane deductible - applicable deductible). The deductible that will apply to the next event is the MAXIMUM of (Remaining deductible, all other perils deductible). The remaining deductible is recalculated and stored after each event per location.

Applicable Deductible = Σ min(loss, deductible)*probabilityj,

where the summation covers from j = 1 to the number of points in the damage ratio distribution and probabilityj is the probability of loss at the jth point in the distribution.

Example: User enters $10,000 as the hurricane deductible (DED1) and $500 as the all other perils deductible (DED2). A first event occurs, and the applicable deductible is calculated as $7000, with a remaining deductible of $3000. For the next event in the same year, the deductible will be max ($3000, $500), or $3000. A second event occurs, and the applicable deductible is calculated as $2700. From this point on, the $500 all other perils deductible would apply to each subsequent event in the year (note that the remaining deductible concept only applies to multiple events within a year – the full hurricane deductible ($10,000) will apply if there is only a single event in a given year).

Standard A-8: Contents

A-8 Contents

A. The methods used in the development of contents loss costs shall be actuarially sound.

The AIR hurricane model represents damages to contents separately from buildings and appurtenant structures since some policies cover contents only and some provide no contents coverage. Contents vulnerability is a function of both building damage and wind speed and the graph below represents the relationship of mean building damage to mean contents damage used in the AIR models. At low wind speeds, the contents damage function is below the building damage function. This relationship reverses at high wind speeds.

Figure 29: Contents Vulnerability Relationship

B. The relationship between the modeled structure and contents loss costs shall be reasonable, based on the relationship between historical structure and contents losses.

The relationship among the modeled structure and contents loss costs is reasonable based on comparisons to client data, as shown in the following chart.

Standard A-8: Contents

Figure 30: Relationship of Cov C to Cov A Loss Cost

Relationship of Cov C to Cov A Loss Cost

Actual Simulated Coverage C Damage Ratio

Coverage A Damage Ratio

Disclosures

1. Describe the methods used in the model to calculate loss costs for contents coverage associated with personal residential structures (including mobile homes), tenants, and condo unit owners.

The contents vulnerability function is a function of both building damage and wind speed. At lower wind speeds, the damage ratio for contents is typically less than the corresponding building damage ratios because a certain level of building damage is usually required before significant contents damages occur. At higher wind speeds this relationship reverses and contents damage ratios eventually exceed those of the building. Validation data comparing actual and simulated damages for Coverage C are illustrated for several storm and company combinations on the following chart.

Standard A-8: Contents

Figure 31: Actual and Modeled Losses, Coverage C Actual Losses

Modeled Losses

Standard A-9: Additional Living Expense (ALE)

A-9 Additional Living Expense (ALE)

A. The methods used in the development of Additional Living Expense (ALE) loss costs shall be actuarially sound.

The AIR hurricane model represents losses to Additional Living Expense (ALE) separately from building, contents and appurtenant structures. The methods used in the development of ALE are actuarially sound.

B. ALE loss cost derivations shall consider the estimated time required to repair or replace the property.

Loss due to Additional Living Expense (Time Element) coverage is based on the mean building damage, the time estimated to make repair or to reconstruct the damaged building and the estimated cost of ALE per time period. Implicit in the time to make repairs is damage to the infrastructure.

C. The relationship between the modeled structure and ALE loss costs shall be reasonable, based on the relationship between historical structure and ALE losses.

The following graph reflects the relationship between repair/reconstruction time and the mean building damage. The ALE loss estimates have been validated using actual client company loss data.

Figure 32: Time Element Vulnerability Relationships

The function corresponds well to actual client loss data, as shown in the following chart.

Standard A-9: Additional Living Expense (ALE)

Figure 33: Relationship of Cov D to Cov A

Relationship of Cov D to Cov A

Actual Simulated CoverageD Damage Ratio

Coverage A Damage Ratio

D. ALE loss costs produced by the model shall appropriately consider ALE claims arising from damage to the infrastructure.

The model considers ALE claims that arise from damage to the infrastructure to the extent that such losses are in the validation data used.

Disclosures

1. Describe the methods used to develop loss cost for additional living expense coverage. State whether the model considers both direct and indirect loss to the structure. For example, direct loss is for expenses paid to house policyholders in an apartment while their home is being repaired. Indirect loss is for expenses incurred for loss of power (e.g., food spoilage).

The time element vulnerability function is based on the mean building damage, the time it takes to repair/reconstruct the damaged building, and the estimated cost for the ALE per time period. At lower wind speeds, the damage to the building is minimal and hence the time to repair is minimal. However, at higher wind speeds, resulting in significant building damage, the time it takes to repair/reconstruct could be several months. This function models losses resulting from expenses incurred while the building is being repaired/reconstructed. The model accounts for direct and indirect losses to the extent that they are in the validation data.

Standard A-9: Additional Living Expense (ALE)

2. State the minimum threshold at which ALE loss is calculated (e.g., loss is estimated for structure damage greater than 20% or only for category 3, 4, 5 events). Provide documentation of validation test results to verify the approach used. ALE loss is a function of mean building damage ratio which starts at a minimum wind speed of 40mph. Validation data of the actual and simulated damages for Coverage D for several company and storm combinations is shown in the following chart.

Figure 34: Actual and Modeled Losses, Coverage D Actual Losses Actual

Modeled Losses

Standard A-10: Output Ranges

A-10 Output Ranges

A. Output ranges shall be logical and any deviations supported.

Output Ranges are logical. There are no deviations other than those inherent to the calculation of county weighted averages. These are explained in Disclosure 1 below.

B. All other factors held constant, output ranges produced by the model shall reflect lower loss costs for:

1. masonry construction versus frame construction,

Output ranges produced by the model reflect lower loss costs for masonry construction versus frame construction.

2. residential risk exposure versus mobile home risk exposure,

Output ranges produced by the model reflect lower loss costs for residential risk exposure versus mobile home risk exposure.

3. in general, inland counties versus coastal counties, and

Output ranges produced by the model reflect lower loss costs, in general, for inland counties versus coastal counties.

4. in general, northern counties versus southern counties.

Output ranges produced by the model reflect lower loss costs, in general, for northern counties versus southern counties.

Disclosures

1. Provide an explanation for all anomalies in the loss costs that are not consistent with the requirements of this Standard.

Loss costs for frame construction are greater than for masonry in every ZIP Code, however the county weighted average loss costs can depart from this when there is more exposure in high loss cost ZIP Codes for masonry construction than for frame. The following example illustrates how the county weighted average loss cost can be greater for masonry than for frame.

Standard A-10: Output Ranges

Table 12: County Weighted Average Loss Costs for Masonry and Frame ZIP Code Construction Exposure Loss Cost per 1,000 A Masonry 10,000 4.0 A Frame 1,000 5.0 B Masonry 1,000 1.0 B Frame 10,000 2.0 County Construction Wtd Avg Masonry 3.73 Frame 2.27

Anomalies of this type have been background shaded in yellow.

2. Provide an explanation of the differences in the output ranges between the prior year and the current year submission.

• The ZIP Code database was updated. • The stochastic catalog was updated. The annual frequency distribution, the probability distribution for landfall location and the probability distribution for hurricane intensity have been updated to reflect the new data. • The relationship of contents damage to building damage for single-family homes and mobile homes has been updated. • Demand Surge was included in the model’s calculation of loss costs in compliance with Standard A-4.

The overall change in the statewide industry residential average loss costs is +10.3%. The isolated impact of each of these changes is shown in the table below.

Table 13: Statewide Impact of Model Updates on Weighted Average Loss Costs ZIP Code Database -0.1% Stochastic Catalog +1.4% Single Family Homes/ Mobile Homes Contents -3.6% Damage Functions Demand Surge +12.6%

3. Provide justification for changes from the prior submission of greater than ten percent in weighted average loss costs for any county, specifically by county.

Standard A-10: Output Ranges

The weighted average loss cost for all counties increase by greater than 10% due to the inclusion of demand surge.

Excluding the impact of demand surge, the contents coverage of all counties for owner frame, masonry and mobile homes decrease by greater than 10% due to the change in the vulnerability function.

The following counties have increase of greater than 10% in weighted average loss cost due to the change in stochastic catalog.

Desoto, Dixie, Franklin, Gadsden, Hamilton, Jefferson, Leon, Liberty, Madison, Union, Wakulla,

The following counties have a decrease of greater than 10% in weighted average loss cost due to the change in stochastic catalog.

Clay, Duval, Flagler, Putnam, Saint Johns and Volusia.

4. Provide justification for changes from the prior submission of ten percent or less in the weighted average loss costs for any county, in the aggregate.

Differences in the output ranges from the prior submission of ten percent or less in the weighted average county loss costs can be attributed to the ZIP Code database update or catalog changes.

5. Provide a completed Form A-6, Output Ranges.

A completed Form A-6 is provided on page 146.

6. Provide a completed Form A-7, Percentage Change in Output Ranges.

A completed Form A-7 is provided on page 187.

7. Provide a completed Form A-8, Percentage Change in Output Ranges by County.

A completed Form A-8 is provided on page 188.

Form A-1: Loss Costs

A. Provide the expected annual loss costs by construction type and coverage for each ZIP Code in the sample data set named “FormA1Input06.xls.” Loss costs should be rounded to six decimal places. There are 1,479 ZIP Codes and three construction types; therefore, the completed file should have 4,437 records in total. The following is a description of the requested file layout. Follow the instructions on Form A-2 below and in the Submission Data description. Note that fields 2-9 are the exposure fields from the sample data set. Fields 10-13 are for the loss costs (net of deductibles).

The expected annual loss costs—rounded to six decimal places—by construction type and coverage for each ZIP Code in the sample data set named FormA1Input06.xls are provided in Form A-1, included on the CD-ROM. The completed file has 4,437 records in total.

B. If there are ZIP Codes in the sample data set that the model does not recognize as “valid,” provide a list in the submission document of such ZIP Codes and provide either a) the new ZIP Code to which the original one was mapped, or b) an indication that the insured values from this ZIP Code were not modeled. Loss cost data should be provided for all ZIP Codes given in the sample data set. That is, if no losses were modeled, the record should still be included in the completed file with loss cost of zero, and, if a ZIP Code was mapped to a new one, the resulting loss costs should be reported with the original ZIP Code.

New ZIP Codes to which the original ZIP Codes were mapped are provided in FormA1Input06.xls, in the worksheet titled Remapped_ZIPs. The ZIP Codes are also included in Attachment E.

Loss cost data is provided for all ZIP Codes given in the sample data set.

Form A-1: Loss Costs

C. Provide the results on CD in both Excel and PDF format using the following file layout. The file name should include the abbreviated name of the modeler, the Standards year, and the Form name.

No. Field Name Description Exposure Fields from Sample Data Set 1 Analysis Date Date of Analysis – YYYY/MM/DD 2 County Code FIPS County Code 3 ZIP Code 5-digit ZIP Code 4 Construction Type Use the following: 1 = Wood Frame, 2 = Masonry, 3 = Mobile Home 5 Annual Deductible 1% (of the Structure Value) policy deductible for each record (i.e., 0.01*$100,000) 6 Structure Value $100,000 for each record 7 Appurtenant Structures Value $10,000 for each record 8 Contents Value $50,000 for each record 9 Additional Living Expense Value $20,000 for each record Loss Costs (net of deductibles) 10 Structure Loss Cost Projected expected annual loss cost for structure divided by the structure value modeled for each record ($100,000) 11 Appurtenant Structures Loss Cost Projected expected annual loss cost for appurtenant structures divided by the appurtenant structures value modeled for each record ($10,000) 12 Contents Loss Cost Projected expected annual loss cost for contents divided by the contents value modeled for each record ($50,000) 13 Additional Living Expense Loss Projected expected annual loss cost for additional living Cost expense divided by the additional living expense value modeled for each record ($20,000)

All deductibles are a percentage of the Structure Value and are policy-level deductibles; however, for reporting purposes, the policy deductible should be pro- rated to the individual coverage losses in proportion to the loss. The default all- other perils deductible is $500.

Example Assume that a model analyzing Wood Frame properties in ZIP Code 33102 (Miami-Dade County) estimated the following:

Field Name Value Analysis Date 1999/11/15 County Code Miami-Dade County = 86 ZIP Code 33102 Construction Type Wood Frame = 1 AnnualDeductible 1% = 0.01*$100,000 = $1,000 Structure Value $100,000 Appurtenant Structures Value $10,000 Contents Value $50,000 Additional Living Expense Value $20,000 Structure Loss Cost* $10,000 Appurtenant Structures Loss Cost* $1,000 Contents Loss Cost* $2,500 Additional Living Expense Loss Cost* $500 *Represents 1st dollar losses (i.e., prior to application of deductibles)

Form A-1: Loss Costs

The $1,000 policy deductible would be applied as follows: Annual Deductible 1% = 0.01*$100,000=$1,000 Structure Loss Cost $10,000-[($10,000÷$14,000)x$1,000]=$9,285.71 Appurtenant Structures Loss Cost $1,000-[($1,000÷$14,000)x$1,000]=$928.57 Contents Loss Cost $2,500-[($2,500÷$14,000)x$1,000]=$2,321.43 Additional Living Expense Loss Cost $500-[($500÷$14,000)x$1,000]=$464.29

The reported Form A-1 data are shown below: Field Name Value Analysis Date 1999/11/15 County Code Miami-Dade County = 86 ZIP Code 33102 Construction Type Wood Frame = 1 Annual Deductible 1% = 0.01 Structure Value $100,000 Appurtenant Structures Value $10,000 Contents Value $50,000 Additional Living Expense Value $20,000 Structure Loss Cost $9,285.71÷$100,000 = 0.092857 Appurtenant Structures Loss Cost $928.57÷$10,000 = 0.092857 Contents Loss Cost $2,321.43÷$50,000 = 0.046429 Additional Living Expense Loss Cost $464.29÷$20,000 = 0.023214

Based on the above information, the data should be reported in the following format: 1999/11/15,86,33102,1,0.01,100000,10000,50000,20000,0.092857,0.092857,0.046429,0.023214

Form A-1 is provided on CD-ROM in both Excel and PDF format

Form A-2: Zero Deductible Loss Costs by ZIP Code

Provide a map color-coded by ZIP Code (with a minimum of 6 value ranges) displaying zero deductible loss costs for frame, masonry, and mobile home.

Figure 35: Loss Costs by ZIP Code for Owners Wood Frame, Zero Deductible

Form A-2: Zero Deductible Loss Costs by ZIP Code

Figure 36: Loss Costs by ZIP Code for Owners Masonry, Zero Deductible

Figure 37: Loss Costs by ZIP Code for Mobile Home, Zero Deductible

Form A-3: Base Hurricane Storm Set Average Annual Zero Deductible Statewide Loss Costs

A. A. Provide the monetary contribution to the average annual personal residential zero deductible statewide loss costs from each specific hurricane in the Base Hurricane Storm Set for the 2002 Florida Hurricane Catastrophe Fund’s aggregate exposure data found in the file named “hlpm2002.exe”. Additional storms that are included in the model’s Base Hurricane Storm Set should be included.

The monetary contribution to the average annual personal residential zero deductible statewide loss costs from each specific hurricane in the Official Storm Set is provided in Form A-4 on the following page.

B. Provide this Form on CD in both an Excel and a PDF format. The file name should include the abbreviated name of the modeler, the Standards year, and the Form name. A hard copy of Form A-4 should be included in the submission.

Form A-3 is provided on CD-ROM in both an Excel and PDF format.

Form A-3: Base Hurricane Storm Set Average Annual Zero Deductible Statewide Loss Costs

Table 14: Form A-3 - Monetary Contribution to the Average Annual Personal Residential Zero Deductible Statewide Loss Costs Date Year Name Contribution Date Year Name Contribution 08/02/1901 1901 NoName 4 117,438 09/18/1948 1948 NoName 7 6,506,118 09/09/1903 1903 NoName 3 3,559,860 10/03/1948 1948 NoName 8 2,815,381 06/14/1906 1906 NoName 2 1,683,188 08/23/1949 1949 NoName 2 35,633,001 09/19/1906 1906 NoName 6 2,954,919 08/20/1950 1950 Baker 462,776 10/08/1906 1906 NoName 8 13,892,411 09/01/1950 1950 Easy 90,663,584 10/06/1909 1909 NoName 8 1 10/13/1950 1950 King 15,841,448 10/09/1910 1910 NoName 5 44,550,393 09/23/1953 1953 Florence 625,377 08/08/1911 1911 NoName 1 677,380 09/21/1956 1956 Flossy 1,805,469 08/23/1911 1911 NoName 2 0 08/29/1960 1960 Donna 93,003,102 09/10/1912 1912 NoName 3 37,189 09/14/1960 1960 Ethel 3,054 08/31/1915 1915 NoName 4 196,053 08/20/1964 1964 Cleo 10,607,542 06/29/1916 1916 NoName 1 3,236,635 08/28/1964 1964 Dora 14,023,106 10/12/1916 1916 NoName 13 1,114,841 10/08/1964 1964 Isbell 5,398,699 11/11/1916 1916 NoName 14 26,626 08/27/1965 1965 Betsy 35,672,670 09/21/1917 1917 NoName 3 3,930,670 06/04/1966 1966 Alma 9,000,747 09/02/1919 1919 NoName 2 26,532,606 09/21/1966 1966 Inez 1,402,070 10/20/1921 1921 NoName 6 41,523,650 10/13/1968 1968 Gladys 8,760,741 09/13/1924 1924 NoName 4 31,910 08/14/1969 1969 Camille 4,208 10/14/1924 1924 NoName 7 1,897,852 06/14/1972 1972 Agnes 350,182 11/29/1925 1925 NoName 2 253,085 09/13/1975 1975 Eloise 3,857,087 07/22/1926 1926 NoName 1 13,471,438 08/25/1979 1979 David 8,542,949 09/11/1926 1926 NoName 6 220,388,706 08/29/1979 1979 Frederic 4,268,471 10/14/1926 1926 NoName 10 4,414,703 08/28/1985 1985 Elena 2,239,470 08/03/1928 1928 NoName 1 6,943,247 11/15/1985 1985 Kate 1,512,152 09/06/1928 1928 NoName 4 144,695,436 10/09/1987 1987 Floyd 23,932 09/22/1929 1929 NoName 2 56,869,229 08/16/1992 1992 Andrew 181,493,411 08/26/1932 1932 NoName 3 284,928 07/31/1995 1995 Erin 2,954,024 07/25/1933 1933 NoName 5 430,671 09/27/1995 1995 Opal 5,973,008 08/31/1933 1933 NoName 12 37,508,150 07/16/1997 1997 Danny 16,488 08/29/1935 1935 NoName 2 18,848,507 08/31/1998 1998 Earl 194,200 10/30/1935 1935 NoName 6 5,593,062 09/15/1998 1998 Georges 372,900 07/27/1936 1936 NoName 5 2,659,602 10/12/1999 1999 Irene 763,233 08/07/1939 1939 NoName 2 394,773 08/09/2004 2004 Charley 53,786,836 08/05/1940 1940 NoName 3 0 08/25/2004 2004 Frances 37,063,138 10/03/1941 1941 NoName 5 35,157,491 09/02/2004 2004 Ivan 13,619,991 10/12/1944 1944 NoName 11 86,562,050 09/13/2004 2004 Jeanne 25,898,038 06/20/1945 1945 NoName 1 2,016,362 07/04/2005 2005 Dennis 4,106,577 09/12/1945 1945 NoName 9 64,261,553 08/23/2005 2005 Katrina 3,835,603 10/05/1946 1946 NoName 5 711,966 09/18/2005 2005 Rita 140,486 09/04/1947 1947 NoName 4 83,433,858 10/15/2005 2005 Wilma 50,837,451 10/09/1947 1947 NoName 8 1,454,992 Additional FL By-Passing Storms in the AIR Base Hurricane Storm Set Date Year Name Contribution Date Year Name Contribution 10/01/1933 1933 NoName18 1,846,947 06/01/1968 1968 Abby 266,228 09/23/1935 1935 NoName4 1,723,,052 06/02/1995 1995 Alison 31,459 05/15/1951 1951 Able 143,530 09/14/2000 2000 Gordon 18,633 10/20/1952 1952 Fox 199 10/29/2001 2001 Michelle 0 10/16/1963 1963 Ginny 921

Form A-4: Hurricane Andrew Percent of Losses

A. Provide the percentage of personal residential zero deductible losses, rounded to four decimal places, from Hurricane Andrew for each affected ZIP Code. Include all ZIP Codes where losses are equal to or greater than $500,000.

The percentage of losses from Hurricane Andrew for each affected ZIP Code are provided in Form A-5 on the following page.

B. Provide a map color-coded by ZIP Code depicting the percentage of total losses from Hurricane Andrewbelow latitude 27ºN using the following interval coding:

Red Over 5% Light Red 2% to 5% Pink 1% to 2% Light Pink 0.5% to 1% Light Blue 0.2% to 0.5% Medium Blue 0.1% to 0.2% Blue Below 0.1%

Figure 38: Percentage of Losses by ZIP Code for Hurricane Andrew

C. Provide this Form on CD in both an Excel and a PDF format. The file name should include the abbreviated name of the modeler, the Standards year, and

Form A-4: Hurricane Andrew Percent of Losses

the Form name. A hard copy of Form A-4 should be included in the submission.

Form A-4 is provided on CD-ROM in both Excel and PDF format.

Rather than using directly a published wind field for Hurricane Andrew, the winds underlying the loss cost calculations must be produced by the model being evaluated and should be the one most emulated by the model. Use the 2002 Florida Hurricane Catastrophe Fund’s aggregate exposure data found in the file named “hlpm2002.exe”.

Form A-4: Hurricane Andrew Percent of Losses

Table 15: Hurricane Andrew Percent of Losses Monetary Monetary Monetary Percent of Percent of Percent of ZIP Contribution ZIP Contribution ZIP Contribution Losses (%) Losses (%) Losses (%) ($) ($) ($) 33004 107,977 0.0595% 33126 861,858 0.4749% 33184 475,337 0.2619% 33009 363,178 0.2001% 33127 746,649 0.4114% 33185 596,666 0.3288% 33010 476,813 0.2627% 33128 123,820 0.0682% 33186 6,053,994 3.3357% 33012 482,434 0.2658% 33129 2,291,786 1.2627% 33187 1,532,725 0.8445% 33013 498,057 0.2744% 33130 498,828 0.2748% 33189 1,041,893 0.5741% 33014 285,601 0.1574% 33131 368,388 0.2030% 33190 173,945 0.0958% 33015 240,886 0.1327% 33132 85,000 0.0468% 33193 1,217,958 0.6711% 33016 229,772 0.1266% 33133 10,560,237 5.8185% 33194 1,565 0.0009% 33018 243,856 0.1344% 33134 6,365,526 3.5073% 33196 2,751,115 1.5158% 33019 389,356 0.2145% 33135 1,992,692 1.0979% 33301 223,270 0.1230% 33020 270,362 0.1490% 33136 201,403 0.1110% 33304 122,322 0.0674% 33021 413,490 0.2278% 33137 947,286 0.5219% 33305 115,782 0.0638% 33023 367,350 0.2024% 33138 1,623,105 0.8943% 33306 42,351 0.0233% 33024 277,654 0.1530% 33139 1,790,318 0.9864% 33308 289,278 0.1594% 33025 178,467 0.0983% 33140 2,021,718 1.1139% 33309 94,574 0.0521% 33026 158,178 0.0872% 33141 781,619 0.4307% 33311 125,946 0.0694% 33027 274,710 0.1514% 33142 1,071,560 0.5904% 33312 286,557 0.1579% 33028 175,962 0.0970% 33143 12,984,623 7.1543% 33313 64,446 0.0355% 33029 341,432 0.1881% 33144 1,250,076 0.6888% 33314 64,726 0.0357% 33030 654,518 0.3606% 33145 4,804,845 2.6474% 33315 90,441 0.0498% 33031 396,646 0.2185% 33146 6,474,150 3.5672% 33316 207,591 0.1144% 33032 804,625 0.4433% 33147 743,810 0.4098% 33317 166,664 0.0918% 33033 1,021,434 0.5628% 33149 3,879,628 2.1376% 33319 77,469 0.0427% 33034 246,505 0.1358% 33150 528,614 0.2913% 33321 57,885 0.0319% 33035 144,438 0.0796% 33154 642,155 0.3538% 33322 92,267 0.0508% 33036 14,503 0.0080% 33155 4,644,823 2.5592% 33323 53,467 0.0295% 33037 334,615 0.1844% 33156 22,301,953 12.2880% 33324 119,686 0.0659% 33039 11,269 0.0062% 33157 18,228,922 10.0438% 33325 105,477 0.0581% 33042 664 0.0004% 33158 4,640,757 2.5570% 33326 118,431 0.0653% 33043 574 0.0003% 33160 596,417 0.3286% 33327 132,870 0.0732% 33050 3,151 0.0017% 33161 733,853 0.4043% 33328 139,290 0.0767% 33054 153,708 0.0847% 33162 464,246 0.2558% 33330 100,985 0.0556% 33055 203,984 0.1124% 33165 2,893,601 1.5943% 33331 157,793 0.0869% 33056 159,896 0.0881% 33166 446,918 0.2462% 33332 64,352 0.0355% 33060 94,024 0.0518% 33167 229,398 0.1264% 33334 124,973 0.0689% 33062 146,784 0.0809% 33168 372,432 0.2052% 33351 44,927 0.0248% 33063 70,506 0.0388% 33169 237,322 0.1308% 33401 4,289 0.0024% 33064 174,798 0.0963% 33170 811,433 0.4471% 33403 1,429 0.0008% 33065 46,568 0.0257% 33172 222,213 0.1224% 33404 4,827 0.0027% 33066 24,278 0.0134% 33173 3,698,058 2.0376% 33405 10,683 0.0059% 33067 55,558 0.0306% 33174 606,372 0.3341% 33406 7,832 0.0043% 33068 59,899 0.0330% 33175 1,767,557 0.9739% 33407 3,683 0.0020% 33069 25,842 0.0142% 33176 12,901,966 7.1088% 33408 7,553 0.0042% 33070 36,056 0.0199% 33177 5,521,268 3.0421% 33409 3,696 0.0020% 33071 72,204 0.0398% 33178 444,746 0.2450% 33410 8,294 0.0046% 33073 41,111 0.0227% 33179 377,043 0.2077% 33411 6,861 0.0038% 33076 45,343 0.0250% 33180 466,843 0.2572% 33412 2,049 0.0011% 33109 190,600 0.1050% 33181 402,877 0.2220% 33413 2,275 0.0013% 33122 12,136 0.0067% 33182 336,564 0.1854% 33414 10,357 0.0057% 33125 2,074,954 1.1433% 33183 1,374,172 0.7571% 33415 5,340 0.0029%

Form A-4: Hurricane Andrew Percent of Losses

Hurricane Andrew Percent of Losses (Continued) Monetary Monetary Monetary Percent of Percent of Percent of ZIP Contribution ZIP Contribution ZIP Contribution Losses (%) Losses (%) Losses (%) ($) ($) ($) 33417 2,752 0.0015% 33916 1,122 0.0006% 34223 2,613 0.0014% 33418 7,792 0.0043% 33917 9,758 0.0054% 34224 2,860 0.0016% 33426 16,591 0.0091% 33919 11,054 0.0061% 34229 536 0.0003% 33428 45,973 0.0253% 33920 1,038 0.0006% 34231 352 0.0002% 33430 1,051 0.0006% 33921 4,937 0.0027% 34238 290 0.0002% 33431 69,396 0.0382% 33922 1,577 0.0009% 34242 507 0.0003% 33432 117,387 0.0647% 33924 6,144 0.0034% 34269 61 0.0000% 33433 93,072 0.0513% 33927 12 0.0000% 34275 1,176 0.0006% 33434 38,306 0.0211% 33928 14,799 0.0082% 34285 541 0.0003% 33435 25,924 0.0143% 33931 15,291 0.0084% 34286 333 0.0002% 33436 37,898 0.0209% 33935 2,034 0.0011% 34287 1,968 0.0011% 33437 36,769 0.0203% 33936 4,049 0.0022% 34288 41 0.0000% 33438 37 0.0000% 33946 900 0.0005% 34289 2 0.0000% 33440 2,057 0.0011% 33947 1,073 0.0006% 34292 1,145 0.0006% 33441 60,205 0.0332% 33948 1,152 0.0006% 34293 2,225 0.0012% 33442 49,412 0.0272% 33950 3,945 0.0022% 34952 470 0.0003% 33444 27,194 0.0150% 33952 1,767 0.0010% 34956 151 0.0001% 33445 45,756 0.0252% 33953 395 0.0002% 34957 789 0.0004% 33446 26,371 0.0145% 33954 344 0.0002% 34990 480 0.0003% 33455 4,717 0.0026% 33955 1,718 0.0009% 34994 270 0.0001% 33458 7,236 0.0040% 33956 5,000 0.0028% 34996 805 0.0004% 33460 12,782 0.0070% 33957 26,644 0.0147% 34997 2,438 0.0013% 33461 8,020 0.0044% 33971 2,049 0.0011% 33462 27,823 0.0153% 33972 2,031 0.0011% 33463 11,923 0.0066% 33980 1,041 0.0006% 33467 15,487 0.0085% 33981 1,086 0.0006% 33469 4,676 0.0026% 33982 1,043 0.0006% 33470 3,252 0.0018% 33983 560 0.0003% 33471 619 0.0003% 339906,837 0.0038% 33476 234 0.0001% 339912,768 0.0015% 33477 5,901 0.0033% 33993 1,169 0.0006% 33478 1,052 0.0006% 34102 162,082 0.0893% 33480 57,817 0.0319% 34103 60,997 0.0336% 33483 48,938 0.0270% 34104 57,461 0.0317% 33484 26,044 0.0143% 34105 51,923 0.0286% 33486 81,825 0.0451% 34108 90,888 0.0501% 33487 77,525 0.0427% 34109 52,485 0.0289% 33493 162 0.0001% 3411043,018 0.0237% 33496 70,238 0.0387% 34112 93,264 0.0514% 33498 23,749 0.0131% 34113 63,866 0.0352% 33901 3,711 0.0020% 3411467,830 0.0374% 33903 11,019 0.0061% 34116 36,395 0.0201% 33904 14,015 0.0077% 34117 21,549 0.0119% 33905 4,396 0.0024% 3411958,504 0.0322% 33907 3,855 0.0021% 3412025,764 0.0142% 33908 25,784 0.0142% 34134 51,614 0.0284% 33909 1,835 0.0010% 3413541,522 0.0229% 33912 24,006 0.0132% 34141 9,684 0.0053% 33913 2,858 0.0016% 34142 1,825 0.0010% 33914 14,494 0.0080% 34145 383,050 0.2111%

Form A-5: Distribution of Hurricanes by Size Form A-5: Distribution of Hurricanes by Size

A. Provide a detailed explanation of how the Expected Annual Hurricane Losses and Return Time are calculated.

The expected annual hurricane loss in each limit range is equal to the total loss within the range divided by the number of years of simulated events, which is 50,000. The return time is equal to the reciprocal of the probability of the loss equaling or exceeding the average loss size in the range. The probability distribution of annual occurrence losses is formulated by selecting the largest event in each simulation year, ranking these events by size, and assigning probability of exceedance according to the percentile of the distribution. For example, the 1,000 year return period loss is likely to be equaled or exceeded 0.1 percent of the time. It represents the 99.0th percentile of the loss distribution. In a 50,000-year simulation, it is the 50th worst simulated event.

B. Complete Form A-5 showing the Distribution of Hurricanes by Size of Loss. For the Expected Annual Hurricane Losses column, provide personal residential, zero deductible statewide loss costs based on the 2003 Florida Hurricane Catastrophe Fund’s aggregate exposure data found in the file named “hlpm2003.exe.”

A completed Form A-5 showing the Distribution of Hurricanes by Size of Loss is provided on page 145. For the Expected Annual Hurricane Losses column, personal residential, zero deductible statewide loss costs based on the 2003 Florida Hurricane Catastrophe Fund’s (FHCF) aggregate exposure data are provided.

C. In the column, Return Time (Years), provide the return time associated with the average loss within the ranges indicated on a cumulative basis.

For example, if the average loss is $4,705 million for the range $4,501 million to $5,000 million, provide the return time associated with a loss that is $4,705 million or greater.

For each range limit in millions ($1,001-$1,500, $1,501-$2,000, $2,001-$2,500) the average loss within that range should be identified and then the return time associated with that loss calculated. The return time is then the reciprocal of the probability of the loss equaling or exceeding this average loss size.

The probability of equaling or exceeding the average of each range should be smaller as the ranges increase (and the average losses within the ranges increase). Therefore, the return time associated with each range and average loss within that range should be larger as the ranges increase. Return times should be based on cumulative probabilities.

A return time for an average loss of $4,705 million within the $4,501-$5,000 million range should be lower than the return time for an average loss of $5,455 million associated with a $5,001- $6,000 million range. The return time associated with the average loss within the ranges indicated is provided in Form A- 5.

Form A-5: Distribution of Hurricanes by Sizes D. Provide a graphical comparison of the current submission Return Times to the prior year’s submission Return Times. Return Time (Years) should be shown on the y-axis on a log 10 scale with Losses in Billions shown on the x-axis. The legend should indicate the corresponding submission with a solid line representing the current year and a dotted line representing the prior year.

Figure 39: Graphical comparison of the current submission Return Times to the prior year’s submission Return Times

5.0 4.5 4.0 Current Year Prior Year 3.5 3.0 2.5 2.0 1.5 1.0

Return Time in Years (Log 10) 0.5 0.0 0 5 - 1.0 1 0 - 1.5 1 5 - 2.0 2 0 - 2.5 2 5 - 3.0 3 0 - 3.5 3 5 - 4.0 4 0 - 4.5 4 5 - 5.0 5 0 - 6.0 6 0 - 7.0 7 0 - 8.0 8 0 - 9.0 9.0 -9.0 1.00 $ - - 0.5 10.0 -10.0 11.0 -11.0 12.0 -12.0 13.0 -13.0 14.0 -14.0 15.0 -15.0 16.0 -16.0 17.0 -17.0 18.0 -18.0 19.0 -19.0 2.00 -20.0 21.0 -21.0 22.0 -22.0 23.0 -23.0 24.0 -24.0 25.0 -25.0 26.0 -26.0 27.0 -27.0 28.0 -28.0 29.0 -29.0 3.00 -30.0 35.0 -35.0 4.00 -40.0 45.0 -45.0 5.00 -50.0 55.0 -55.0 6.00 -60.0 65.0 -65.0 7.00 -70.0 75.0 -75.0 80.0 -80.0 90.0 100 0 - Max 90.0 -90.0 100.0 Loss Range ($ Billions)

E. Provide this Form on CD in both an Excel and a PDF format. The file name should include the abbreviated name of the modeler, the Standards year, and the Form name. A hard copy of Form A-6 should be included in the submission.

Form A-5 is provided on CD-ROM in both Excel and PDF format.

Form A-5: Distribution of Hurricanes by Size Form A-5: Distribution of Hurricanes by Size of Loss

EXPECTED AVERAGE ANNUAL RETURN LOSS RANGE TOTAL LOSS LOSS NUMBER OF HURRICANE TIME (MILLIONS) (MILLIONS) (Millions) HURRICANES LOSSES* (YEARS) $ - To 500 3,245,130 114 28,487 64.9 2.2 501 To 1,000 3,575,240 721 4,961 71.5 3.3 1,001 To 1,500 3,533,290 1,226 2,883 70.7 4.1 1,501 To 2,000 3,331,840 1,733 1,923 66.6 4.8 2,001 To 2,500 2,991,770 2,239 1,336 59.8 5.5 2,501 To 3,000 3,002,220 2,749 1,092 60.0 6.2 3,001 To 3,500 2,954,400 3,239 912 59.1 6.9 3,501 To 4,000 2,630,710 3,737 704 52.6 7.6 4,001 To 4,500 2,367,970 4,229 560 47.4 8.3 4,501 To 5,000 2,400,660 4,735 507 48.0 9.1 5,001 To 6,000 4,647,330 5,467 850 92.9 10.2 6,001 To 7,000 4,201,000 6,473 649 84.0 11.8 7,001 To 8,000 3,827,220 7,475 512 76.5 13.4 8,001 To 9,000 3,797,020 8,475 448 75.9 15.3 9,001 To 10,000 3,230,160 9,500 340 64.6 17.1 10,001 To 11,000 3,045,670 10,502 290 60.9 19.0 11,001 To 12,000 2,886,620 11,455 252 57.7 21.1 12,001 To 13,000 2,638,320 12,504 211 52.8 23.2 13,001 To 14,000 2,144,250 13,486 159 42.9 25.4 14,001 To 15,000 2,453,940 14,520 169 49.1 27.5 15,001 To 16,000 2,188,300 15,520 141 43.8 29.9 16,001 To 17,000 1,948,330 16,511 118 39.0 32.2 17,001 To 18,000 2,137,680 17,522 122 42.8 34.9 18,001 To 19,000 1,909,400 18,538 103 38.2 37.7 19,001 To 20,000 1,852,230 19,497 95 37.0 40.7 20,001 To 21,000 1,950,380 20,530 95 39.0 43.9 21,001 To 22,000 1,526,560 21,501 71 30.5 47.6 22,001 To 23,000 1,734,720 22,529 77 34.7 50.9 23,001 To 24,000 1,339,320 23,497 57 26.8 54.6 24,001 To 25,000 1,200,060 24,491 49 24.0 58.1 25,001 To 26,000 1,629,140 25,455 64 32.6 62.2 26,001 To 27,000 1,059,070 26,477 40 21.2 66.1 27,001 To 28,000 1,100,020 27,500 40 22.0 69.8 28,001 To 29,000 1,424,140 28,483 50 28.5 74.2 29,001 To 30,000 1,414,010 29,458 48 28.3 80.3 30,001 To 35,000 4,753,430 32,336 147 95.1 94.5 35,001 To 40,000 3,886,990 37,375 104 77.7 123.5 40,001 To 45,000 3,683,350 42,337 87 73.7 161.3 45,001 To 50,000 2,760,040 47,587 58 55.2 208.3 50,001 To 55,000 1,462,310 52,225 28 29.2 257.7 55,001 To 60,000 2,066,530 57,404 36 41.3 308.6 60,001 To 65,000 1,692,710 62,693 27 33.9 381.7 65,001 To 70,000 1,211,300 67,294 18 24.2 476.2 70,001 To 75,000 871,576 72,631 12 17.4 543.5 75,001 To 80,000 778,311 77,831 10 15.6 609.8 80,001 To 90,000 2,031,640 84,652 24 40.6 793.7 90,001 To 100,000 1,431,690 95,446 15 28.6 1,136.4 100,001 To $ maximum 4,511,130 118,714 38 90.2 3,125.0 TOTAL 118,459,102 2,417 49,019 2,369.2 5.7

*Personal residential zero deductible statewide loss using 2002 FHCF exposure data – file name: hlpm2002.exe.

Form A-6: Output Ranges

A. Provide output ranges in the format shown in the file named “2006FormA6.xls” by using an automated program or script. A hard copy of the output range spreadsheets should be included in the submission. Provide the output ranges on CD in both Excel and PDF format as specified. The file name should include the abbreviated name of the modeler, the Standards year, and the Form name.

The output ranges are provided in Form A-6 on page 149.

B. Provide loss costs by county. Within each county, loss costs should be shown separately per $1,000 of exposure for personal residential, tenants, condo unit owners, and mobile home; for each major deductible option; and by construction type. For each of these categories using ZIP Code centroids, the output range should show the highest loss cost, the lowest loss cost, and the weighted average loss cost based on the 2002 Florida Hurricane Catastrophe Fund aggregate exposure data provided to each modeler in the file named “hlpm2002.exe”. A file named “02FHCFWts.xls” has also been provided for use in determining the weighted average loss costs. Include the statewide range of loss costs (i.e., low, high, and weighted average). For each of the loss costs provided, identify what that loss cost represents by line of business, deductible option, construction type, and coverages included, i.e., structure, contents, appurtenant structures, or additional living expenses as specified.

All requested loss costs are provided in Form A-6.

C. If a modeler has loss costs for a ZIP Code for which there is no exposure, then the modeler should give the loss costs zero weight (i.e., assume the exposure in that ZIP Code is zero). Provide a list in the submission document of those ZIP Codes where this occurs.

ZIP codes for which there is zero FHCF exposure have been given zero weight. The list of these ZIP Codes is provided below.

Form A-6: Output Ranges

Table 16: List of Zero-weighted ZIP Codes

32008 32221 32445 32766 33514 34480 32008 32222 32448 32767 33523 34484 32009 32226 32449 32784 33525 34498 32013 32227 32455 32798 33527 34602 32038 32234 32460 32814 33534 34604 32040 32254 32462 32815 33538 34609 32044 32258 32464 32820 33576 34610 32046 32266 32465 32827 33584 34614 32052 32305 32466 32829 33585 34705 32053 32306 32502 32830 33597 34731 32054 32307 32531 32831 33598 34739 32058 32310 32533 32832 33610 34753 32059 32317 32534 32833 33619 34756 32060 32320 32535 32909 33620 34762 32061 32321 32536 32948 33621 34771 32062 32324 32542 32949 33811 34773 32063 32327 32544 32950 33834 34786 32064 32331 32564 33010 33838 34947 32066 32332 32565 33013 33839 34956 32071 32333 32567 33018 33841 34972 32083 32334 32568 33030 33843 34981 32087 32336 32570 33031 33849 34984 32091 32340 32571 33039 33850 34987 32094 32343 32577 33042 33857 34988 32096 32344 32580 33054 33865 32097 32347 32611 33055 33868 32130 32348 32617 33056 33873 32134 32350 32618 33109 33875 32139 32351 32619 33122 33890 32140 32352 32621 33128 33896 32145 32355 32622 33130 33897 32148 32356 32626 33147 33927 32159 32358 32628 33150 33956 32162 32403 32631 33158 33960 32179 32420 32640 33167 34117 32180 32421 32643 33170 34141 32181 32423 32648 33184 34211 32187 32425 32668 33187 34219 32189 32426 32680 33194 34251 32190 32427 32693 33199 34286 32195 32428 32694 33317 34288 32202 32430 32696 33327 34289 32208 32431 32709 33332 34433 32209 32437 32732 33430 34434 32212 32438 32735 33438 34449 32214 32440 32736 33471 34461 32215 32442 32744 33493 34473 32219 32443 32754 33498 34475 32220 32444 32759 33513 34479

Form A-6: Output Ranges

D. If the modeler does not have loss costs for a ZIP Code for which there is some exposure, the modeler should not assume such loss costs are zero, but should use only the exposures for which it has loss costs in calculating the weighted average loss costs. Provide a list in the submission document of the ZIP Codes where this occurs.

There are no ZIP Codes in the FHCF data for which AIR does not produce loss costs. FHCF ZIP Codes are remapped to current ZIP Codes by AIR.

E. All anomalies in loss cost that are not consistent with the requirements of Standard A-10 and have been explained in Disclosure A-10.1 should be shaded.

All such anomalies are shaded in yellow.

Output ranges should be computed assuming an average structure.

Output ranges are computed assuming an average structure.

Modelers should indicate if per diem is used in producing loss costs for Coverage D (ALE) in the output ranges. If a per diem rate is used in the submission, a rate of $150.00 per day per policy should be used.

A $150 per diem per policy is used in producing loss costs for Coverage D.

Form A-6: Output Ranges

LOSS COSTS PER $1.000 Personal Residential -- Owners -- FRAME $0 $0 $0 DEDUCTIBLE $0 DEDUCTIBLE $500 $1,000 $2,500 1% 2% 5% DEDUCTIBLE DEDUCTIBLE APPURTENANT ADDITIONAL DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE COUNTY LOSS COSTS STRUCTURE CONTENTS STRUCTURE LIVING EXPENSE TOTAL* TOTAL* TOTAL* TOTAL* TOTAL* TOTAL*

ALACHUA LOW 0.710 0 062 0 071 0 016 0.710 0 656 0 557 0 656 0 583 0.473 HIGH 1 072 0 095 0.107 0 028 1 095 1 013 0 853 1 013 0 897 0.709 WGHTD AVE 0 844 0 073 0 084 0 019 0 847 0.782 0 660 0.782 0 693 0 555

BAKER LOW 0 643 0055 0 064 0 013 0 640 0 591 0503 0 591 0 526 0.426 HIGH 0.718 0 062 0 072 0 016 0.719 0 664 0 564 0 664 0 590 0.477 WGHTD AVE 0.710 0 062 0 071 0 015 0.711 0 656 0 557 0 656 0 584 0.472

BAY LOW 1 253 0.120 0.125 0 035 1 320 1 223 1034 1 223 1 086 0 863 HIGH 5.482 0 920 0 548 0 512 6 836 6 541 5 877 6 541 6 072 5.141 WGHTD AVE 2 880 0 357 0 255 0.167 3 258 3 074 2 688 3 074 2.798 2 296

BRADFORD LOW 0 864 0 075 0 086 0 021 0 872 0 805 0 681 0 805 0.714 0 573 HIGH 0 917 0 080 0 092 0 022 0 929 0 857 0.723 0 857 0.760 0 606 WGHTD AVE 0 873 0 076 0 087 0 021 0 882 0 814 0 688 0 814 0.722 0 578

BREVARD LOW 1.178 0.101 0.118 0 028 1.176 1 086 0 921 1 086 0 965 0.780 HIGH 6 266 0 952 0 627 0 634 7 600 7 285 6 560 7 285 6.775 5.735 WGHTD AVE 2 303 0 238 0 228 0.110 2.483 2 325 2 002 2 325 2 092 1 689

BROWARD LOW 3 295 0 296 0 330 0.125 3.457 3 214 2.714 3 214 2 854 2 239 HIGH 14 039 2 355 1.404 1 872 17 358 16.779 15.405 16.779 15 819 13.782 WGHTD AVE 6.188 0.751 0 607 0 534 7 041 6 692 5 924 6 692 6.147 5.100

CALHOUN LOW 1 065 0 099 0.106 0 025 1.105 1 020 0 858 1 020 0 902 0.715 HIGH 1 594 0.163 0.159 0 057 1.724 1 605 1 364 1 605 1.431 1.134 WGHTD AVE 1.118 0.106 0.112 0 029 1.170 1 082 0 912 1 082 0 959 0.761

CHARLOTTE LOW 2 549 0 240 0 255 0 096 2 676 2.490 2.119 2.490 2 222 1.768 HIGH 5 903 0 812 0 590 0 513 6 936 6 606 5 873 6 606 6 087 5 078 WGHTD AVE 4 087 0.478 0 394 0 274 4 576 4 323 3.786 4 323 3 940 3 231

CITRUS LOW 0 979 0087 0 098 0 026 0 988 0 916 0.784 0 916 0 819 0 667 HIGH 2 034 0 227 0 203 0.116 2 248 2.115 1 844 2.115 1 921 1 574 WGHTD AVE 1.402 0.138 0.139 0 058 1.478 1 380 1.189 1 380 1 242 1 011

CLAY LOW 0 599 0052 0 060 0 012 0 591 0 546 0.469 0 546 0.489 0.405 HIGH 0 887 0 077 0 089 0 022 0 890 0 821 0 693 0 821 0.726 0 584 WGHTD AVE 0.702 0 061 0 070 0 015 0 699 0 646 0 552 0 646 0 577 0.471

COLLIER LOW 2.731 0 239 0 273 0 078 2 813 2 597 2.162 2 597 2 283 1.763 HIGH 9 681 1 547 0 968 1 003 11 896 11.410 10 294 11.410 10 625 9 031 WGHTD AVE 6 588 0 903 0 656 0 571 7.715 7 344 6 522 7 344 6.762 5 639

COLUMBIA LOW 0 599 0 052 0 060 0 012 0 596 0 550 0.468 0 550 0.490 0.400 HIGH 0.765 0 066 0 077 0 017 0.770 0.711 0 599 0.711 0 629 0 502 WGHTD AVE 0 668 0 058 0 067 0 015 0 669 0 618 0 524 0 618 0 549 0.444

DESOTO LOW 1 579 0.134 0.158 0 038 1 567 1.445 1 218 1.445 1 279 1 023 HIGH 1 825 0.160 0.182 0 052 1 848 1.710 1.449 1.710 1 520 1 217 WGHTD AVE 1 590 0.135 0.159 0 038 1 580 1.458 1 229 1.458 1 290 1 032

DIXIE LOW 0 965 0094 0 097 0 032 1 000 0 929 0.795 0 929 0 832 0 676 HIGH 2 062 0 276 0 206 0.139 2 372 2 245 1 979 2 245 2 055 1.709 WGHTD AVE 1 065 0.108 0.105 0 040 1.117 1 041 0 896 1 041 0 936 0.764

Form A-6: Output Ranges

DUVAL LOW 0.495 0042 0 049 0 010 0.485 0.450 0389 0.450 0.405 0 337 HIGH 1 618 0.199 0.162 0.105 1 822 1.723 1 514 1.723 1 574 1 301 WGHTD AVE 0.717 0 068 0 070 0 023 0.733 0 683 0 591 0 683 0 616 0 508

ESCAMBIA LOW 1 279 0.116 0.128 0 031 1 327 1 224 1 025 1 224 1 079 0 850 HIGH 4 879 0.726 0.488 0.411 5 909 5 632 5 015 5 632 5.195 4 339 WGHTD AVE 2 321 0 255 0 230 0.114 2 570 2.407 2 067 2.407 2.163 1.731

FLAGLER LOW 1.149 0.101 0.115 0 029 1.167 1 078 0 909 1 078 0 955 0.759 HIGH 3 269 0.461 0 327 0 280 3 861 3 681 3 283 3 681 3 399 2 846 WGHTD AVE 2 503 0 309 0 239 0.169 2 844 2 694 2 370 2 694 2.464 2 030

FRANKL N LOW 2 597 0 366 0 260 0.176 3 064 2 900 2 554 2 900 2 653 2.198 HIGH 4 693 0 812 0.469 0.433 5 895 5 645 5 086 5 645 5 250 4.470 WGHTD AVE 4 208 0 647 0 375 0 359 5.137 4 907 4 399 4 907 4 547 3 848

GADSEN LOW 0 646 0 059 0 065 0 012 0 655 0 605 0 514 0 605 0 538 0.438 HIGH 0 874 0 080 0 087 0 019 0 898 0 828 0 698 0 828 0.734 0 586 WGHTD AVE 0.727 0 066 0 072 0 015 0.740 0 684 0 580 0 684 0 608 0.492

GILCHRIST LOW 0.787 0 072 0 079 0 022 0 801 0.743 0 635 0.743 0 664 0 540 HIGH 0 846 0 077 0 085 0 023 0 860 0.798 0 681 0.798 0.712 0 578 WGHTD AVE 0 829 0 075 0 083 0 023 0 843 0.781 0 667 0.781 0 698 0 566

GLADES LOW 2 361 0200 0 236 0 061 2 388 2.199 1833 2.199 1 933 1 506 HIGH 2 361 0 200 0 236 0 061 2 388 2.199 1 833 2.199 1 933 1 506 WGHTD AVE 2 361 0 200 0 236 0 061 2 388 2.199 1 833 2.199 1 933 1 506

GULF LOW 1.485 0.152 0.149 0 054 1 599 1.488 1268 1.488 1 329 1 061 HIGH 4.706 0.764 0.471 0.403 5 814 5 551 4 967 5 551 5.137 4 328 WGHTD AVE 4 226 0 636 0 373 0 338 5 094 4 857 4 335 4 857 4.487 3.770

HAMILTON LOW 0.402 0 035 0 040 0 007 0 393 0 363 0 312 0 363 0 325 0 272 HIGH 0 563 0 049 0 056 0 012 0 560 0 517 0.441 0 517 0.461 0 377 WGHTD AVE 0.473 0 041 0 047 0 009 0.467 0.432 0 370 0.432 0 386 0 319

HARDEE LOW 1 268 0.105 0.127 0 026 1 235 1.137 0 962 1.137 1 008 0 818 HIGH 1 379 0.114 0.138 0 031 1 350 1 243 1 048 1 243 1.100 0 888 WGHTD AVE 1 302 0.108 0.130 0 027 1 270 1.169 0 988 1.169 1 036 0 839

HENDRY LOW 2 223 0.189 0 222 0 057 2 245 2 068 1.723 2 068 1 817 1.417 HIGH 2 667 0 227 0 267 0 072 2.717 2 503 2 077 2 503 2.195 1 693 WGHTD AVE 2 366 0 201 0 235 0 061 2 393 2 204 1 834 2 204 1 935 1 504

HERNANDO LOW 1 226 0.109 0.123 0 036 1 249 1.159 0 990 1.159 1 036 0 838 HIGH 2.787 0 356 0 279 0 212 3.188 3 026 2 681 3 026 2.780 2 320 WGHTD AVE 1.752 0.178 0.163 0 083 1 862 1.745 1 512 1.745 1 577 1 288

HIGHLANDS LOW 1 377 0.113 0.138 0 028 1 344 1 236 1 039 1 236 1 091 0 877 HIGH 1 849 0.154 0.185 0 042 1 836 1 688 1.409 1 688 1.484 1.168 WGHTD AVE 1 526 0.126 0.154 0 032 1 500 1 379 1.155 1 379 1 215 0 968

HILLSBOROUGH LOW 1.128 0 094 0.113 0 024 1.103 1 016 0 860 1 016 0 901 0.733 HIGH 3 584 0.450 0 358 0 277 4 098 3 888 3.432 3 888 3 564 2 949 WGHTD AVE 1 509 0.142 0.156 0 058 1 565 1.456 1 248 1.456 1 305 1 058

Form A-6: Output Ranges

HOLMES LOW 1 091 0.101 0.109 0 025 1.126 1 037 0 873 1 037 0 917 0.732 HIGH 1 208 0.112 0.121 0 030 1 256 1.159 0 976 1.159 1 025 0 816 WGHTD AVE 1.114 0.103 0.111 0 026 1.152 1 062 0 893 1 062 0 938 0.749

INDIAN RIVER LOW 1 860 0.162 0.186 0 051 1 894 1.751 1.476 1.751 1 551 1 232 HIGH 5.776 0.755 0 578 0.474 6.743 6.415 5 683 6.415 5 897 4 883 WGHTD AVE 4.160 0.470 0.416 0 271 4 657 4 399 3 843 4 399 4 003 3 262

JACKSON LOW 0 521 0 046 0 052 0 008 0 512 0.472 0.406 0.472 0.423 0 358 HIGH 1.100 0.102 0.110 0 025 1.141 1 053 0 886 1 053 0 931 0.739 WGHTD AVE 0 809 0 073 0 081 0 015 0 821 0.756 0 638 0.756 0 669 0 541

JEFFERSON LOW 0.491 0 044 0 049 0 009 0.489 0.451 0 385 0.451 0.402 0 333 HIGH 0 565 0 051 0 057 0 011 0 567 0 523 0.445 0 523 0.465 0 381 WGHTD AVE 0.498 0 045 0 050 0 009 0.496 0.457 0 390 0.457 0.408 0 338

LAFAYETTE LOW 0 540 0 048 0 054 0 010 0 537 0.496 0.425 0.496 0.443 0 366 HIGH 0 687 0 062 0 069 0 016 0 695 0 642 0 547 0 642 0 572 0.467 WGHTD AVE 0 681 0 061 0 068 0 016 0 688 0 636 0 542 0 636 0 567 0.463

LAKE LOW 0 951 0080 0 095 0 018 0 928 0 856 0.728 0 856 0.761 0 622 HIGH 1.400 0.121 0.140 0 035 1.419 1 308 1 098 1 308 1.155 0 912 WGHTD AVE 1 010 0 085 0.102 0 021 0 994 0 917 0.778 0 917 0 815 0 662

LEE LOW 2.138 0.189 0 214 0 059 2.180 2 015 1695 2 015 1.783 1.408 HIGH 7 990 1 233 0.799 0 801 9 644 9 226 8 282 9 226 8 560 7 239 WGHTD AVE 4 944 0 571 0.427 0 344 5 502 5 207 4 573 5 207 4.755 3 915

LEON LOW 0.559 0051 0 056 0 010 0 563 0 520 0.443 0 520 0.463 0 380 HIGH 0.770 0 072 0 077 0 018 0.795 0.736 0 626 0.736 0 656 0 527 WGHTD AVE 0 605 0 055 0 060 0 012 0 611 0 565 0.481 0 565 0 504 0.411

LEVY LOW 0 883 0076 0 088 0 021 0 885 0 818 0694 0 818 0.727 0 586 HIGH 3 293 0.486 0 329 0 298 3 925 3.748 3 366 3.748 3.477 2 955 WGHTD AVE 1 583 0.164 0.135 0 081 1 680 1 580 1 380 1 580 1.436 1.186

LIBERTY LOW 0 899 0 084 0 090 0 021 0 929 0 859 0.729 0 859 0.764 0 615 HIGH 1 010 0 095 0.101 0 025 1 050 0 970 0 820 0 970 0 861 0 687 WGHTD AVE 0 982 0 092 0 098 0 024 1 020 0 943 0.798 0 943 0 837 0 670

MADISON LOW 0.405 0 035 0 040 0 006 0 395 0 364 0 313 0 364 0 326 0 272 HIGH 0.462 0 040 0 046 0 008 0.457 0.422 0 362 0.422 0 377 0 313 WGHTD AVE 0.419 0 037 0 042 0 007 0.411 0 379 0 326 0 379 0 340 0 283

MANATEE LOW 1 399 0.121 0.140 0 037 1 396 1 290 1 095 1 290 1.147 0 931 HIGH 4 569 0 638 0.457 0.415 5 363 5.115 4 567 5.115 4.728 3 967 WGHTD AVE 2 822 0 313 0 272 0.173 3 097 2 919 2 548 2 919 2 654 2.172

MARION LOW 0.887 0077 0 089 0 020 0 886 0 820 0.700 0 820 0.732 0 596 HIGH 1 293 0.114 0.129 0 035 1 322 1 224 1 035 1 224 1 087 0 864 WGHTD AVE 1 035 0 090 0.104 0 027 1 043 0 965 0 821 0 965 0 860 0 694

MART N LOW 2.411 0206 0 241 0 066 2.458 2 270 1900 2 270 2 001 1 563 HIGH 7 203 0 976 0.720 0 655 8.499 8.110 7 233 8.110 7.491 6 262 WGHTD AVE 5.134 0 627 0 540 0 386 5 889 5 580 4 906 5 580 5.101 4.192

Form A-6: Output Ranges

MIAMI-DADE LOW 3 915 0 365 0 392 0.167 4.194 3 918 3 337 3 918 3 502 2.747 HIGH 17 361 3 219 1.736 2 584 21 900 21 268 19.739 21 268 20 204 17 873 WGHTD AVE 9 257 1 304 0 923 0 998 10 981 10 535 9 511 10 535 9 814 8 351

MONROE LOW 9 259 1.422 0 926 0 907 11 304 10 832 9.747 10 832 10 069 8 508 HIGH 13 838 2 692 1 384 1.789 17 996 17.426 16 058 17.426 16.472 14.412 WGHTD AVE 12 597 2 220 1 201 1 538 16 093 15 550 14 261 15 550 14 649 12.728

NASSAU LOW 0 567 0050 0 057 0 013 0 567 0 525 0.451 0 525 0.470 0 387 HIGH 1 319 0.170 0.132 0 089 1.493 1.415 1 254 1.415 1 300 1 089 WGHTD AVE 1.151 0.142 0.111 0 069 1 279 1 209 1 067 1 209 1.107 0 925

OKALOOSA LOW 1 280 0.119 0.128 0 032 1 332 1 228 1 028 1 228 1 083 0 854 HIGH 5 396 0 876 0 540 0 500 6 681 6 389 5.735 6 389 5 926 5 021 WGHTD AVE 3 372 0.441 0 331 0 226 3 928 3.716 3 262 3.716 3 392 2.791

OKEECHOBEE LOW 2 043 0.172 0 204 0 050 2 053 1 891 1 579 1 891 1 664 1 306 HIGH 2 232 0.189 0 223 0 057 2 255 2 078 1.734 2 078 1 828 1.428 WGHTD AVE 2.159 0.182 0 214 0 054 2.176 2 004 1 673 2 004 1.763 1 380

ORANGE LOW 0 965 0 080 0 097 0 018 0 938 0 864 0.733 0 864 0.768 0 627 HIGH 1.482 0.128 0.148 0 037 1 501 1 386 1.168 1 386 1 227 0 976 WGHTD AVE 1.167 0 098 0.116 0 024 1.149 1 058 0 892 1 058 0 936 0.753

OSCEOLA LOW 1.126 0 093 0.113 0 022 1 097 1 009 0 852 1 009 0 893 0.723 HIGH 1 500 0.126 0.150 0 034 1.491 1 374 1.155 1 374 1 214 0 967 WGHTD AVE 1 223 0.102 0.122 0 025 1 203 1.108 0 933 1.108 0 980 0.787

PALM BEACH LOW 2 649 0 228 0 265 0 078 2.717 2 511 2.104 2 511 2 216 1.732 HIGH 11 559 1.784 1.156 1 350 14 019 13.493 12 263 13.493 12 631 10 825 WGHTD AVE 6.109 0.734 0 608 0 507 6 953 6 605 5 839 6 605 6 062 5 019

PASCO LOW 1.162 0098 0.116 0 026 1.148 1 058 0891 1 058 0 935 0.753 HIGH 3 330 0.431 0 333 0 267 3 853 3 662 3 252 3 662 3 371 2 817 WGHTD AVE 1 959 0 210 0.186 0.105 2.115 1 987 1.731 1 987 1 803 1.482

P NELLAS LOW 1 570 0.146 0.157 0 057 1 638 1 527 1 312 1 527 1 371 1.108 HIGH 5 254 0 815 0 525 0 560 6 337 6 082 5 508 6 082 5 678 4 857 WGHTD AVE 3 397 0.435 0 338 0 274 3 901 3.710 3 298 3.710 3.418 2 859

POLK LOW 1 029 0085 0.103 0 019 0 996 0 917 0.776 0 917 0 813 0 662 HIGH 1 321 0.109 0.132 0 026 1 289 1.186 0 999 1.186 1 049 0 845 WGHTD AVE 1.135 0 094 0.114 0 022 1.102 1 013 0 855 1 013 0 896 0.727

PUTNAM LOW 0.773 0 066 0 077 0 016 0.766 0.706 0 596 0.706 0 626 0 504 HIGH 1 085 0 093 0.108 0 025 1 092 1 008 0 848 1 008 0 891 0.708 WGHTD AVE 0 952 0 082 0 095 0 022 0 956 0 882 0.744 0 882 0.781 0 622

SAINT JOHNS LOW 0 696 0 060 0 070 0 016 0 694 0 642 0 549 0 642 0 574 0.469 HIGH 2 321 0 302 0 232 0.170 2 674 2 535 2 237 2 535 2 323 1 926 WGHTD AVE 1 593 0.189 0.157 0 097 1.771 1 672 1.465 1 672 1 524 1 258

SAINT LUC E LOW 2.140 0.183 0 214 0 058 2.173 2 006 1 684 2 006 1.772 1 396 HIGH 6.451 0 871 0 645 0 569 7 599 7 246 6.451 7 246 6 685 5 572 WGHTD AVE 3.183 0 320 0 318 0.155 3.436 3 216 2.761 3 216 2 889 2 315

Form A-6: Output Ranges

SANTA ROSA LOW 1.451 0.135 0.145 0 041 1 525 1.411 1.187 1.411 1 248 0 984 HIGH 4 935 0.756 0.494 0.427 6 014 5.734 5.114 5.734 5 295 4.440 WGHTD AVE 3.167 0.406 0 305 0 206 3 653 3.452 3 023 3.452 3.146 2 579

SARASOTA LOW 1 531 0.137 0.153 0 047 1 547 1.433 1 219 1.433 1 277 1 034 HIGH 5.425 0.783 0 542 0 522 6.454 6.170 5 535 6.170 5.722 4 830 WGHTD AVE 3.752 0.441 0 339 0 262 4.184 3 963 3.489 3 963 3 625 2 995

SEM NOLE LOW 1 022 0 085 0.102 0 020 0 998 0 919 0.779 0 919 0 816 0 663 HIGH 1 290 0.110 0.129 0 030 1 290 1.190 1 004 1.190 1 054 0 843 WGHTD AVE 1.108 0 093 0.111 0 023 1 092 1 006 0 850 1 006 0 892 0.719

SUMTER LOW 0 972 0 083 0 097 0 022 0 965 0 891 0.758 0 891 0.794 0 645 HIGH 1 357 0.118 0.136 0 037 1 375 1 270 1 071 1 270 1.125 0 897 WGHTD AVE 1 078 0 092 0.108 0 026 1 075 0 994 0 844 0 994 0 884 0.715

SUWANEE LOW 0 519 0 045 0 052 0 010 0 514 0.474 0.405 0.474 0.423 0 349 HIGH 0.734 0 066 0 073 0 018 0.742 0 687 0 585 0 687 0 612 0.497 WGHTD AVE 0 562 0 049 0 056 0 011 0 559 0 516 0.440 0 516 0.460 0 378

TAYLOR LOW 0 635 0058 0 063 0 014 0 641 0 593 0507 0 593 0 530 0.435 HIGH 2 058 0 277 0 206 0.135 2 383 2 256 1 985 2 256 2 063 1.705 WGHTD AVE 0 831 0 084 0 085 0 027 0 869 0 810 0 698 0 810 0.729 0 597

UNION LOW 0 824 0072 0 082 0 020 0 832 0.768 0647 0.768 0 680 0 541 HIGH 0 831 0 073 0 083 0 020 0 840 0.775 0 655 0.775 0 687 0 549 WGHTD AVE 0 825 0 072 0 082 0 020 0 833 0.768 0 647 0.768 0 680 0 542

VOLUSIA LOW 1.116 0 097 0.112 0 026 1.126 1 039 0 876 1 039 0 920 0.731 HIGH 3 245 0.414 0 325 0 238 3.744 3 553 3.136 3 553 3 257 2 689 WGHTD AVE 2.197 0 249 0 212 0.125 2.425 2 284 1 991 2 284 2 074 1 695

WAKULLA LOW 1 021 0.102 0.102 0 031 1 086 1 011 0 866 1 011 0 906 0.730 HIGH 2 306 0 327 0 231 0.155 2.716 2 572 2 271 2 572 2 357 1 961 WGHTD AVE 1 269 0.133 0.114 0 050 1 361 1 275 1.103 1 275 1.151 0 938

WALTON LOW 1.413 0.136 0.141 0 042 1.492 1 381 1.164 1 381 1 224 0 969 HIGH 4 007 0 580 0.401 0 303 4.785 4 544 4 017 4 544 4.169 3.462 WGHTD AVE 3 281 0.422 0 282 0 217 3.775 3 572 3.137 3 572 3 261 2 689

WASH NGTON LOW 1.195 0.112 0.120 0 030 1 246 1.151 0 969 1.151 1 018 0 809 HIGH 1 685 0.172 0.169 0 064 1 813 1 686 1.436 1 686 1 505 1 204 WGHTD AVE 1 257 0.119 0.126 0 033 1 315 1 215 1 024 1 215 1 076 0 855

STATEWIDE LOW 0.402 0035 0 040 0 006 0 393 0 363 0312 0 363 0 325 0 272 HIGH 17 361 3 219 1.736 2 584 21 900 21 268 19.739 21 268 20 204 17 873 WGHTD AVE 2.556 0 272 0 238 0.159 2.767 2 613 2 291 2 613 2 383 1 965

Form A-6: Output Ranges

LOSS COSTS PER $1,000 PERSONAL RESIDENTIAL -Owners -- MASONRY 0% 0% $0 DEDUCTIBLE $0 DEDUCTIBLE $500 $1,000 $2,500 1% 2% 5% DEDUCTIBLE DEDUCTIBLE APPURTENANT ADDITIONAL DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE COUNTY LOSS COSTS STRUCTURE CONTENTS STRUCTURE LIVING EXPENSE TOTAL* TOTAL* TOTAL* TOTAL* TOTAL* TOTAL*

ALACHUA LOW 0 575 0049 0 057 0 011 0 563 0 520 0.445 0 520 0.465 0 384 HIGH 0 868 0 075 0 087 0 020 0 869 0 802 0 678 0 802 0.712 0 571 WGHTD AVE 0 689 0 058 0 069 0 014 0 678 0 625 0 531 0 625 0 556 0.453

BAKER LOW 0 520 0044 0 052 0 009 0 508 0.470 0.402 0.470 0.420 0 347 HIGH 0 581 0 050 0 058 0 011 0 571 0 527 0.451 0 527 0.471 0 388 WGHTD AVE 0 575 0 049 0 058 0 011 0 565 0 522 0.447 0 522 0.466 0 384

BAY LOW 1015 0095 0.102 0 025 1 050 0 970 0 823 0 970 0 863 0 695 HIGH 4.495 0.705 0.450 0 383 5.497 5 240 4 675 5 240 4 840 4 064 WGHTD AVE 2.169 0 247 0 204 0.102 2 387 2 241 1 945 2 241 2 028 1 655

BRADFORD LOW 0 699 0060 0 070 0 015 0 692 0 639 0 543 0 639 0 568 0.464 HIGH 0.743 0 063 0 074 0 016 0.737 0 679 0 577 0 679 0 604 0.490 WGHTD AVE 0.706 0 060 0 071 0 015 0 699 0 645 0 548 0 645 0 574 0.468

BREVARD LOW 0 953 0081 0 095 0 019 0 933 0 862 0.736 0 862 0.769 0 634 HIGH 5.150 0.729 0 515 0.471 6.135 5 858 5 236 5 858 5.419 4 547 WGHTD AVE 2.146 0 222 0 212 0.103 2 310 2.164 1 869 2.164 1 952 1 585

BROWARD LOW 2 672 0231 0 267 0 087 2.741 2 538 2.141 2 538 2 250 1.779 HIGH 11 638 1 831 1.164 1.441 14 203 13 688 12.490 13 688 12 847 11.105 WGHTD AVE 4.733 0.499 0 364 0 288 5 087 4 804 4 202 4 804 4 374 3 585

CALHOUN LOW 0 862 0078 0 086 0 017 0 879 0 810 0 684 0 810 0.717 0 579 HIGH 1 293 0.126 0.129 0 039 1 370 1 271 1 078 1 271 1.131 0 902 WGHTD AVE 0 905 0 083 0 091 0 020 0 929 0 858 0.726 0 858 0.761 0 613

CHARLOTTE LOW 2 067 0.187 0 207 0 067 2.124 1 971 1 677 1 971 1.758 1.410 HIGH 4 844 0 627 0.484 0 381 5 589 5 304 4 688 5 304 4 866 4 036 WGHTD AVE 3 243 0 366 0 324 0.193 3 588 3 376 2 941 3 376 3 064 2 505

CITRUS LOW 0.793 0068 0 079 0 018 0.784 0.727 0 625 0.727 0 652 0 539 HIGH 1 654 0.175 0.165 0 083 1.792 1 682 1.463 1 682 1 524 1 249 WGHTD AVE 1 040 0 095 0.104 0 033 1 061 0 987 0 851 0 987 0 888 0.729

CLAY LOW 0.484 0041 0 048 0 009 0.469 0.434 0 377 0.434 0 392 0 330 HIGH 0.718 0 061 0 072 0 015 0.706 0 651 0 552 0 651 0 577 0.472 WGHTD AVE 0 557 0 047 0 056 0 011 0 544 0 503 0.434 0 503 0.452 0 376

COLLIER LOW 2 212 0.188 0 221 0 055 2 228 2 050 1.708 2 050 1 801 1.410 HIGH 7 960 1.193 0.796 0.753 9 611 9.185 8 229 9.185 8 510 7.173 WGHTD AVE 5 325 0 665 0.498 0.401 6 048 5.732 5 053 5.732 5 248 4 349

COLUMBIA LOW 0.485 0041 0 048 0 009 0.473 0.437 0 375 0.437 0 391 0 326 HIGH 0 619 0 053 0 062 0 012 0 611 0 563 0.478 0 563 0 500 0.407 WGHTD AVE 0 539 0 046 0 054 0 010 0 530 0.489 0.418 0.489 0.437 0 360

DESOTO LOW 1 278 0.106 0.128 0 027 1 242 1.144 0 972 1.144 1 017 0 831 HIGH 1.478 0.126 0.148 0 036 1.465 1 354 1.153 1 354 1 207 0 981 WGHTD AVE 1 309 0.110 0.131 0 028 1 278 1.178 1 001 1.178 1 048 0 855

DIXIE LOW 0.782 0073 0 078 0 023 0.795 0.738 0 634 0.738 0 662 0 545 HIGH 1 685 0 215 0.169 0.104 1 904 1.797 1 580 1.797 1 641 1 364 WGHTD AVE 0 819 0 078 0 081 0 025 0 837 0.778 0 671 0.778 0 699 0 577

Form A-6: Output Ranges

DUVAL LOW 0.400 0034 0 040 0 007 0 385 0 358 0 312 0 358 0 324 0 275 HIGH 1 320 0.154 0.132 0 077 1.458 1 375 1 204 1 375 1 252 1 034 WGHTD AVE 0 524 0 047 0 052 0 013 0 520 0.484 0.421 0.484 0.438 0 367

ESCAMBIA LOW 1 036 0092 0.104 0 022 1 054 0 970 0 816 0 970 0 857 0 686 HIGH 3 993 0 553 0 399 0 300 4.738 4.498 3 976 4.498 4.127 3.418 WGHTD AVE 1 889 0.198 0.189 0 081 2 054 1 917 1 640 1 917 1.717 1 374

FLAGLER LOW 0 930 0080 0 093 0 020 0 925 0 854 0.723 0 854 0.758 0 611 HIGH 2 676 0 352 0 268 0 205 3.102 2 947 2 610 2 947 2.708 2 252 WGHTD AVE 1 986 0 226 0.197 0.114 2 209 2 082 1 817 2 082 1 893 1 549

FRANKL N LOW 2.118 0278 0 212 0.126 2.445 2 306 2 021 2 306 2.101 1.735 HIGH 3 856 0 627 0 386 0 326 4.752 4 535 4 061 4 535 4.199 3 549 WGHTD AVE 3 006 0.420 0 292 0 205 3 541 3 361 2 979 3 361 3 089 2 580

GADSEN LOW 0 523 0047 0 052 0 009 0 521 0.481 0.412 0.481 0.430 0 358 HIGH 0.707 0 064 0 071 0 013 0.714 0 658 0 558 0 658 0 585 0.476 WGHTD AVE 0 589 0 053 0 059 0 010 0 591 0 545 0.466 0 545 0.487 0.401

GILCHRIST LOW 0 638 0057 0 064 0 015 0 636 0 590 0 507 0 590 0 529 0.437 HIGH 0 685 0 060 0 069 0 016 0 683 0 633 0 543 0 633 0 567 0.467 WGHTD AVE 0 672 0 059 0 067 0 016 0 670 0 620 0 533 0 620 0 556 0.458

GLADES LOW 1 911 0.159 0.191 0 043 1 891 1.738 1.454 1.738 1 530 1 214 HIGH 1 911 0.159 0.191 0 043 1 891 1.738 1.454 1.738 1 530 1 214 WGHTD AVE 1 911 0.159 0.191 0 043 1 891 1.738 1.454 1.738 1 530 1 214

GULF LOW 1 205 0.118 0.120 0 037 1 272 1.180 1 005 1.180 1 053 0 847 HIGH 3 858 0 588 0 386 0 300 4 675 4.448 3 954 4.448 4 097 3.426 WGHTD AVE 3.197 0.459 0 310 0 223 3.784 3 592 3.182 3 592 3 300 2.751

HAMILTON LOW 0 325 0028 0 032 0 005 0 312 0 289 0 251 0 289 0 261 0 224 HIGH 0.455 0 039 0 046 0 008 0.445 0.411 0 353 0.411 0 368 0 308 WGHTD AVE 0 392 0 034 0 039 0 006 0 381 0 352 0 305 0 352 0 317 0 268

HARDEE LOW 1 025 0084 0.103 0 018 0 978 0 902 0.771 0 902 0 805 0 668 HIGH 1.116 0 091 0.112 0 022 1 069 0 985 0 839 0 985 0 877 0.723 WGHTD AVE 1 053 0 086 0.105 0 019 1 005 0 926 0.791 0 926 0 826 0 685

HENDRY LOW 1 800 0.150 0.180 0 040 1.778 1 634 1 367 1 634 1.438 1.143 HIGH 2.160 0.180 0 216 0 051 2.151 1 975 1 643 1 975 1.732 1 360 WGHTD AVE 1 985 0.165 0.196 0 045 1 966 1 806 1 506 1 806 1 586 1 252

HERNANDO LOW 0 993 0086 0 099 0 025 0 990 0 918 0.787 0 918 0 822 0 673 HIGH 2 276 0 274 0 228 0.156 2 557 2.419 2.134 2.419 2 216 1 842 WGHTD AVE 1 310 0.127 0.130 0 052 1 367 1 277 1.105 1 277 1.153 0 945

HIGHLANDS LOW 1.114 0091 0.111 0 020 1 064 0 979 0 832 0 979 0 870 0.718 HIGH 1.496 0.123 0.150 0 030 1.454 1 335 1.121 1 335 1.178 0 948 WGHTD AVE 1 243 0.102 0.125 0 023 1.196 1 099 0 928 1 099 0 973 0.794

HILLSBOROUGH LOW 0 912 0 075 0 091 0 017 0 874 0 806 0 690 0 806 0.720 0 600 HIGH 2 932 0 348 0 293 0 203 3 290 3.111 2.732 3.111 2 841 2 340 WGHTD AVE 1.135 0.101 0.114 0 034 1.136 1 054 0 906 1 054 0 946 0.778

Form A-6: Output Ranges

HOLMES LOW 0 883 0080 0 088 0 017 0 895 0 824 0 698 0 824 0.731 0 596 HIGH 0 978 0 089 0 098 0 021 0 999 0 921 0.779 0 921 0 817 0 662 WGHTD AVE 0 900 0 081 0 090 0 018 0 913 0 841 0.712 0 841 0.746 0 607

INDIAN RIVER LOW 1 506 0.128 0.151 0 036 1 502 1 386 1.174 1 386 1 231 0 993 HIGH 4.725 0 580 0.473 0 347 5.413 5.129 4 512 5.129 4 691 3 855 WGHTD AVE 3 662 0 396 0 335 0 214 4 016 3.785 3 297 3.785 3.437 2.797

JACKSON LOW 0.421 0037 0 042 0 006 0.408 0 377 0 330 0 377 0 341 0 297 HIGH 0 891 0 081 0 089 0 018 0 908 0 837 0.707 0 837 0.741 0 598 WGHTD AVE 0 654 0 058 0 066 0 011 0 653 0 601 0 512 0 601 0 535 0.443

JEFFERSON LOW 0 397 0 035 0 040 0 006 0 388 0 359 0 310 0 359 0 322 0 274 HIGH 0.457 0 041 0 046 0 008 0.451 0.416 0 357 0.416 0 372 0 312 WGHTD AVE 0.401 0 035 0 040 0 006 0 392 0 362 0 313 0 362 0 325 0 276

LAFAYETTE LOW 0.437 0038 0 044 0 007 0.426 0 394 0 341 0 394 0 355 0 299 HIGH 0 556 0 049 0 056 0 011 0 551 0 510 0.438 0 510 0.456 0 379 WGHTD AVE 0 554 0 049 0 055 0 011 0 549 0 507 0.436 0 507 0.454 0 378

LAKE LOW 0.769 0064 0 077 0 013 0.737 0 680 0 585 0 680 0 610 0 509 HIGH 1.133 0 096 0.113 0 025 1.125 1 036 0 874 1 036 0 917 0.738 WGHTD AVE 0 832 0 069 0 083 0 015 0 800 0.738 0 632 0.738 0 659 0 547

LEE LOW 1.732 0.149 0.173 0 042 1.728 1 595 1 346 1 595 1.413 1.133 HIGH 6 568 0 954 0 657 0 607 7.798 7.434 6 634 7.434 6 867 5.773 WGHTD AVE 3 263 0 359 0 317 0.181 3 557 3 340 2 898 3 340 3 022 2.466

LEON LOW 0.452 0040 0 045 0 007 0.447 0.413 0 356 0.413 0 370 0 311 HIGH 0 623 0 057 0 062 0 012 0 632 0 584 0.499 0 584 0 522 0.425 WGHTD AVE 0.491 0 044 0 049 0 008 0.488 0.451 0 387 0.451 0.404 0 336

LEVY LOW 0.714 0061 0 071 0 014 0.702 0 648 0 554 0 648 0 579 0.474 HIGH 2.702 0 374 0 270 0 222 3.163 3 012 2 692 3 012 2.784 2 353 WGHTD AVE 0 980 0 094 0 096 0 033 1 007 0 939 0 811 0 939 0 846 0 696

LIBERTY LOW 0.727 0066 0 073 0 015 0.739 0 682 0 582 0 682 0 609 0.497 HIGH 0 818 0 075 0 082 0 017 0 835 0.771 0 654 0.771 0 685 0 556 WGHTD AVE 0.795 0 073 0 079 0 017 0 811 0.749 0 636 0.749 0 666 0 541

MADISON LOW 0 327 0028 0 033 0 005 0 314 0 290 0 253 0 290 0 262 0 225 HIGH 0 374 0 032 0 037 0 006 0 363 0 336 0 291 0 336 0 302 0 256 WGHTD AVE 0 340 0 029 0 034 0 005 0 328 0 303 0 264 0 303 0 274 0 234

MANATEE LOW 1.133 0096 0.113 0 026 1.107 1 023 0 876 1 023 0 914 0.757 HIGH 3.748 0.492 0 375 0 310 4 327 4.113 3 651 4.113 3.786 3.158 WGHTD AVE 2.170 0 226 0 215 0.115 2 330 2.187 1 900 2.187 1 980 1 620

MARION LOW 0.718 0061 0 072 0 014 0.703 0 651 0 560 0 651 0 584 0.483 HIGH 1 047 0 090 0.105 0 025 1 049 0 969 0 823 0 969 0 862 0 695 WGHTD AVE 0 837 0 072 0 084 0 019 0 828 0.766 0 656 0.766 0 685 0 562

MART N LOW 1 952 0.163 0.195 0 047 1 948 1.794 1 505 1.794 1 582 1 255 HIGH 5 903 0.747 0 590 0.481 6 837 6.499 5.755 6.499 5 972 4 951 WGHTD AVE 4 052 0.453 0.417 0 261 4 514 4 257 3.715 4 257 3 870 3.160

Form A-6: Output Ranges

MIAMI-DADE LOW 3.177 0 283 0 318 0.115 3 328 3 096 2 624 3 096 2.756 2.162 HIGH 14.489 2 539 1.449 2 035 18 078 17 510 16.159 17 510 16 567 14 544 WGHTD AVE 6.794 0 857 0 577 0 588 7.768 7.404 6 597 7.404 6 832 5.719

MONROE LOW 7 610 1094 0.761 0 677 9.122 8.708 7.775 8.708 8 050 6.736 HIGH 11.498 2 099 1.150 1 384 14.715 14 205 13 001 14 205 13 362 11 583 WGHTD AVE 9 307 1.438 0 917 0 958 11.450 10 989 9 929 10 989 10 244 8.716

NASSAU LOW 0.459 0040 0 046 0 009 0.450 0.417 0 361 0.417 0 376 0 314 HIGH 1 080 0.133 0.108 0 067 1 202 1.137 1 005 1.137 1 043 0 873 WGHTD AVE 0 836 0 093 0 081 0 040 0 894 0 841 0.740 0 841 0.768 0 643

OKALOOSA LOW 1 036 0094 0.104 0 023 1 058 0 974 0 819 0 974 0 860 0 691 HIGH 4.422 0 668 0.442 0 370 5 367 5.113 4 560 5.113 4.720 3 968 WGHTD AVE 2.776 0 343 0 275 0.164 3.178 2 996 2 616 2 996 2.724 2 231

OKEECHOBEE LOW 1 653 0.137 0.165 0 035 1 627 1.495 1 256 1.495 1 319 1 056 HIGH 1 807 0.150 0.181 0 040 1.786 1 643 1 376 1 643 1.447 1.152 WGHTD AVE 1.754 0.146 0.175 0 038 1.733 1 593 1 336 1 593 1.404 1.120

ORANGE LOW 0.780 0064 0 078 0 013 0.744 0 686 0 590 0 686 0 615 0 514 HIGH 1 200 0.102 0.120 0 026 1.190 1 097 0 930 1 097 0 974 0.788 WGHTD AVE 0 959 0 079 0 096 0 017 0 927 0 854 0.727 0 854 0.760 0 625

OSCEOLA LOW 0 911 0074 0 091 0 016 0 869 0 800 0 683 0 800 0.713 0 592 HIGH 1 213 0.101 0.121 0 024 1.182 1 088 0 922 1 088 0 965 0.786 WGHTD AVE 0 990 0 082 0 099 0 018 0 954 0 879 0.747 0 879 0.782 0 643

PALM BEACH LOW 2.146 0.180 0 215 0 054 2.152 1 984 1 666 1 984 1.752 1 387 HIGH 9 548 1 389 0 955 1 032 11.413 10 949 9 882 10 949 10.199 8 666 WGHTD AVE 4 885 0 536 0.426 0 324 5 360 5 067 4.442 5 067 4 621 3.796

PASCO LOW 0 940 0078 0 094 0 019 0 910 0 838 0.713 0 838 0.745 0 614 HIGH 2.719 0 327 0 272 0.192 3 084 2 921 2 580 2 921 2 678 2 226 WGHTD AVE 1 840 0.195 0.180 0 097 1 982 1 864 1 630 1 864 1 696 1.400

P NELLAS LOW 1 272 0.114 0.127 0 040 1 299 1 209 1 039 1 209 1 086 0 882 HIGH 4 329 0 631 0.433 0.424 5.139 4 917 4.426 4 917 4 571 3 881 WGHTD AVE 2 651 0 316 0 265 0.184 2 979 2 822 2.492 2 822 2 587 2.150

POLK LOW 0 832 0068 0 083 0 014 0.790 0.727 0 623 0.727 0 650 0 544 HIGH 1 069 0 087 0.107 0 019 1 021 0 940 0 800 0 940 0 836 0 691 WGHTD AVE 0 913 0 074 0 091 0 016 0 868 0.799 0 682 0.799 0.712 0 593

PUTNAM LOW 0 625 0052 0 063 0 011 0 608 0 560 0.477 0 560 0.499 0.411 HIGH 0 878 0 074 0 088 0 018 0 866 0.798 0 675 0.798 0.708 0 572 WGHTD AVE 0.775 0 065 0 077 0 016 0.763 0.704 0 597 0.704 0 625 0 506

SAINT JOHNS LOW 0 563 0 048 0 056 0 011 0 551 0 510 0.439 0 510 0.458 0 381 HIGH 1 895 0 231 0.190 0.124 2.141 2 023 1.776 2 023 1 846 1 527 WGHTD AVE 1.416 0.161 0.140 0 079 1 555 1.464 1 279 1.464 1 331 1 097

SAINT LUC E LOW 1.732 0.145 0.173 0 040 1.722 1 587 1 337 1 587 1.404 1.124 HIGH 5 284 0 666 0 528 0.417 6.110 5 803 5.129 5 803 5 326 4.403 WGHTD AVE 2 651 0 255 0 266 0.115 2 811 2 624 2 251 2 624 2 355 1 895

Form A-6: Output Ranges

SANTA ROSA LOW 1.175 0.106 0.118 0 028 1 212 1.118 0 943 1.118 0 990 0.790 HIGH 4 042 0 576 0.404 0 313 4 827 4 585 4 062 4 585 4 213 3 507 WGHTD AVE 2.762 0 342 0 262 0.161 3.142 2 962 2 585 2 962 2 693 2 203

SARASOTA LOW 1 241 0.108 0.124 0 034 1 229 1.137 0 974 1.137 1 017 0 838 HIGH 4.461 0 603 0.446 0 388 5 211 4 965 4.426 4 965 4 584 3 843 WGHTD AVE 2 918 0 322 0 275 0.175 3.189 3 005 2 624 3 005 2.732 2 242

SEM NOLE LOW 0 827 0068 0 083 0 014 0.792 0.730 0 625 0.730 0 652 0 542 HIGH 1 044 0 088 0.104 0 021 1 023 0 943 0 802 0 943 0 839 0 685 WGHTD AVE 0 891 0 074 0 089 0 016 0 861 0.793 0 677 0.793 0.707 0 583

SUMTER LOW 0.787 0066 0 079 0 015 0.765 0.707 0 607 0.707 0 633 0 525 HIGH 1 099 0 094 0.110 0 026 1 090 1 006 0 853 1 006 0 894 0.725 WGHTD AVE 0 902 0 076 0 091 0 020 0 886 0 819 0.700 0 819 0.732 0 602

SUWANEE LOW 0.420 0036 0 042 0 007 0.408 0 377 0 326 0 377 0 339 0 285 HIGH 0 594 0 052 0 059 0 013 0 589 0 545 0.467 0 545 0.487 0.403 WGHTD AVE 0.453 0 039 0 045 0 008 0.442 0.408 0 352 0.408 0 366 0 308

TAYLOR LOW 0 514 0046 0 051 0 010 0 509 0.471 0.406 0.471 0.423 0 353 HIGH 1 679 0 214 0.168 0 099 1 906 1.798 1 575 1.798 1 638 1 349 WGHTD AVE 0 604 0 056 0 060 0 015 0 610 0 566 0.488 0 566 0 509 0.422

UNION LOW 0 667 0057 0 067 0 014 0 660 0 608 0 515 0 608 0 540 0.439 HIGH 0 673 0 058 0 067 0 014 0 666 0 615 0 522 0 615 0 547 0.445 WGHTD AVE 0 668 0 057 0 067 0 014 0 661 0 609 0 516 0 609 0 540 0.439

VOLUSIA LOW 0 904 0077 0 090 0 018 0 893 0 823 0 697 0 823 0.731 0 591 HIGH 2 651 0 318 0 265 0.173 2 998 2 834 2.486 2 834 2 586 2.123 WGHTD AVE 1 693 0.179 0.166 0 079 1 818 1.705 1.480 1.705 1 543 1 261

WAKULLA LOW 0 828 0080 0 083 0 021 0 863 0 802 0 687 0 802 0.718 0 582 HIGH 1 882 0 251 0.188 0.113 2.172 2 051 1 802 2 051 1 873 1 553 WGHTD AVE 1 033 0.105 0.100 0 033 1 091 1 019 0 881 1 019 0 919 0.751

WALTON LOW 1.145 0.107 0.114 0 030 1.187 1 096 0 926 1 096 0 972 0.781 HIGH 3 276 0.447 0 328 0 224 3 840 3 633 3.194 3 633 3 319 2.742 WGHTD AVE 2.497 0 296 0 224 0.126 2.783 2 620 2 286 2 620 2 380 1 953

WASH NGTON LOW 0 968 0 089 0 097 0 021 0 991 0 913 0.773 0 913 0 810 0 654 HIGH 1 367 0.134 0.137 0 045 1.444 1 339 1.141 1 339 1.195 0 966 WGHTD AVE 1 024 0 094 0.103 0 024 1 053 0 972 0 823 0 972 0 863 0 696

STATEWIDE LOW 0 325 0028 0 032 0 005 0 312 0 289 0 251 0 289 0 261 0 224 HIGH 14.489 2 539 1.449 2 035 18 078 17 510 16.159 17 510 16 567 14 544 WGHTD AVE 3 226 0 344 0 271 0.195 3.465 3 273 2 874 3 273 2 988 2.468

Form A-6: Output Ranges

ALACHUA LOW 1 879 0.185 0.188 0 066 1 861 1 655 1 300 1 861 1 655 1 300 HIGH 2.752 0 283 0 275 0.118 2 815 2 518 1 981 2 815 2 518 1 981 WGHTD AVE 2.195 0 217 0 219 0 082 2.196 1 955 1 534 2.196 1 955 1 534

BAKER LOW 1.718 0.165 0.172 0 056 1 692 1 501 1.178 1 692 1 501 1.178 HIGH 1 901 0.186 0.190 0 067 1 888 1 675 1 312 1 888 1 675 1 312 WGHTD AVE 1 857 0.181 0.186 0 064 1 843 1 635 1 282 1 843 1 635 1 282

BAY LOW 3254 0374 0 325 0.154 3.424 3 074 2.443 3.424 3 074 2.443 HIGH 12.122 2.473 1 212 1 361 15 290 14.414 12 565 15 290 14.414 12 565 WGHTD AVE 5 941 0 905 0 574 0.452 6 807 6 265 5 211 6 807 6 265 5 211

BRADFORD LOW 2 256 0225 0 226 0 088 2 270 2 020 1 582 2 270 2 020 1 582 HIGH 2 385 0 239 0 238 0 094 2.407 2.145 1 680 2.407 2.145 1 680 WGHTD AVE 2 283 0 229 0 229 0 089 2 301 2 048 1 604 2 301 2 048 1 604

BREVARD LOW 3 341 0331 0 334 0.131 3 344 2 971 2 322 3 344 2 971 2 322 HIGH 14 373 2.719 1.437 1.765 17 576 16 608 14 522 17 576 16 608 14 522 WGHTD AVE 7.186 0 987 0 672 0 591 8 029 7 375 6 094 8 029 7 375 6 094

BROWARD LOW 8 927 1013 0 893 0 578 9 594 8.714 6 992 9 594 8.714 6 992 HIGH 30 698 6 339 3 070 4 655 38 055 36 377 32 569 38 055 36 377 32 569 WGHTD AVE 15 215 2 356 1.437 1 668 17 611 16.453 14 030 17 611 16.453 14 030

CALHOUN LOW 2 818 0305 0 282 0.112 2 910 2 599 2 044 2 910 2 599 2 044 HIGH 4 033 0 514 0.403 0 237 4 397 3 989 3 223 4 397 3 989 3 223 WGHTD AVE 2 988 0 338 0 300 0.132 3.130 2 806 2 222 3.130 2 806 2 222

CHARLOTTE LOW 6 801 0.790 0 680 0.409 7 252 6 563 5 265 7 252 6 563 5 265 HIGH 13 907 2 392 1 391 1 518 16 544 15.492 13 294 16 544 15.492 13 294 WGHTD AVE 10.150 1 544 1 028 0 927 11 626 10.764 9 034 11 626 10.764 9 034

CITRUS LOW 2 627 0269 0 263 0.112 2 645 2 362 1 876 2 645 2 362 1 876 HIGH 5 009 0 697 0 501 0.401 5 560 5.108 4 241 5 560 5.108 4 241 WGHTD AVE 3 543 0.417 0 353 0 210 3.729 3 373 2.728 3.729 3 373 2.728

CLAY LOW 1 617 0.155 0.162 0 052 1 573 1 393 1 098 1 573 1 393 1 098 HIGH 2 324 0 231 0 232 0 091 2 321 2 062 1 612 2 321 2 062 1 612 WGHTD AVE 2.115 0 209 0 211 0 079 2.111 1 877 1.473 2.111 1 877 1.473

COLLIER LOW 7 575 0812 0.758 0 393 8 008 7 209 5 669 8 008 7 209 5 669 HIGH 22 339 4.454 2 234 2 820 27 864 26 393 23.170 27 864 26 393 23.170 WGHTD AVE 13.776 2 217 1 354 1 341 16.158 15 031 12 688 16.158 15 031 12 688

COLUMBIA LOW 1 606 0.156 0.161 0 053 1 578 1 399 1 099 1 578 1 399 1 099 HIGH 2 012 0.198 0 201 0 072 2 013 1.791 1.401 2 013 1.791 1.401 WGHTD AVE 1.752 0.172 0.175 0 061 1.735 1 542 1 213 1.735 1 542 1 213

DESOTO LOW 4.467 0.433 0.447 0.175 4.459 3 963 3 084 4.459 3 963 3 084 HIGH 5 052 0 522 0 505 0 237 5.155 4 610 3 630 5.155 4 610 3 630 WGHTD AVE 4.483 0.436 0.448 0.177 4.479 3 981 3.100 4.479 3 981 3.100

DIXIE LOW 2 556 0286 0 256 0.120 2 633 2 366 1 894 2 633 2 366 1 894 HIGH 4 930 0.770 0.493 0.406 5 655 5 231 4.409 5 655 5 231 4.409 WGHTD AVE 2 615 0 298 0 262 0.127 2.706 2.435 1 954 2.706 2.435 1 954

Form A-6: Output Ranges

DUVAL LOW 1 355 0.128 0.136 0 041 1 306 1.159 0 924 1 306 1.159 0 924 HIGH 3 850 0 563 0 385 0 310 4 322 3 986 3 340 4 322 3 986 3 340 WGHTD AVE 1 812 0.192 0.179 0 076 1 821 1 628 1 302 1 821 1 628 1 302

ESCAMBIA LOW 3 332 0354 0 333 0.142 3.463 3 094 2.424 3.463 3 094 2.424 HIGH 11 029 2 097 1.103 1 208 13 607 12.781 11 047 13 607 12.781 11 047 WGHTD AVE 5 337 0.720 0 536 0 368 5 950 5.440 4.450 5 950 5.440 4.450

FLAGLER LOW 3 080 0316 0 308 0.132 3.146 2 812 2 209 3.146 2 812 2 209 HIGH 7 627 1 308 0.763 0.791 9 026 8.445 7 260 9 026 8.445 7 260 WGHTD AVE 5 219 0 828 0.433 0.472 5 977 5 523 4 639 5 977 5 523 4 639

FRANKL N LOW 6.165 1069 0 617 0 536 7 332 6 804 5.764 7 332 6 804 5.764 HIGH 10 386 2.141 1 039 1.131 13.127 12 372 10.796 13.127 12 372 10.796 WGHTD AVE 7 983 1.489 0.778 0.765 9.737 9.105 7 821 9.737 9.105 7 821

GADSEN LOW 1.732 0.176 0.173 0 052 1.723 1 528 1 202 1.723 1 528 1 202 HIGH 2 287 0 239 0 229 0 081 2 315 2 058 1 614 2 315 2 058 1 614 WGHTD AVE 1 921 0.199 0.191 0 063 1 924 1.711 1 349 1 924 1.711 1 349

GILCHRIST LOW 2 053 0216 0 205 0 084 2 069 1 851 1.473 2 069 1 851 1.473 HIGH 2 201 0 229 0 220 0 091 2 219 1 984 1 577 2 219 1 984 1 577 WGHTD AVE 2.154 0 225 0 215 0 089 2.172 1 942 1 544 2.172 1 942 1 544

GLADES LOW 6.481 0652 0 648 0 291 6 679 5 969 4 638 6 679 5 969 4 638 HIGH 6.481 0 652 0 648 0 291 6 679 5 969 4 638 6 679 5 969 4 638 WGHTD AVE 6.481 0 652 0 648 0 291 6 679 5 969 4 638 6 679 5 969 4 638

GULF LOW 3.777 0.478 0 378 0 217 4 091 3.707 2 993 4 091 3.707 2 993 HIGH 10 550 2 082 1 055 1.114 13.182 12 386 10.732 13.182 12 386 10.732 WGHTD AVE 6 279 1 058 0 596 0 532 7.400 6 866 5 810 7.400 6 866 5 810

HAMILTON LOW 1.117 0.106 0.112 0 029 1 067 0 943 0.744 1 067 0 943 0.744 * 1 515 0.148 0.151 0 048 1.487 1 318 1 036 1.487 1 318 1 036 WGHTD AVE 1 302 0.127 0.129 0 038 1 266 1.121 0 883 1 266 1.121 0 883

HARDEE LOW 3 655 0338 0 366 0.121 3 569 3.152 2.440 3 569 3.152 2.440 HIGH 3 960 0 371 0 396 0.144 3 891 3.442 2 662 3 891 3.442 2 662 WGHTD AVE 3 804 0 354 0 381 0.129 3.724 3 291 2 546 3.724 3 291 2 546

HENDRY LOW 6 340 0642 0 634 0 286 6 539 5 849 4 555 6 539 5 849 4 555 HIGH 7.492 0.769 0.749 0 363 7 827 7 022 5.477 7 827 7 022 5.477 WGHTD AVE 6 902 0.701 0 695 0 321 7.158 6.413 4 998 7.158 6.413 4 998

HERNANDO LOW 3 238 0342 0 324 0.158 3 312 2 974 2 372 3 312 2 974 2 372 HIGH 6 592 1 023 0 659 0 634 7 539 7 003 5 946 7 539 7 003 5 946 WGHTD AVE 3 940 0.464 0.400 0 238 4.162 3.773 3 057 4.162 3.773 3 057

HIGHLANDS LOW 3 965 0366 0 397 0.131 3 886 3.432 2 643 3 886 3.432 2 643 HIGH 5 201 0.499 0 520 0 200 5 228 4 644 3 583 5 228 4 644 3 583 WGHTD AVE 4 265 0 395 0.431 0.145 4 202 3.716 2 861 4 202 3.716 2 861

HILLSBOROUGH LOW 3.155 0 302 0 316 0.112 3.147 2.783 2.157 3.147 2.783 2.157 HIGH 8.706 1 364 0 871 0 862 10 038 9 344 7 938 10 038 9 344 7 938 WGHTD AVE 3 936 0.431 0 396 0 203 4 047 3 637 2 899 4 047 3 637 2 899

Form A-6: Output Ranges

HOLMES LOW 2 802 0296 0 280 0.104 2 860 2 541 1 983 2 860 2 541 1 983 HIGH 3 073 0 333 0 307 0.125 3.171 2 825 2 213 3.171 2 825 2 213 WGHTD AVE 2 884 0 307 0 290 0.111 2 957 2 629 2 054 2 957 2 629 2 054

INDIAN RIVER LOW 5.102 0 526 0 510 0 237 5 240 4 685 3 679 5 240 4 685 3 679 HIGH 13.733 2 292 1 373 1.499 16 283 15 244 13 064 16 283 15 244 13 064 WGHTD AVE 7.163 0 872 0.713 0.490 7.765 7 070 5.739 7.765 7 070 5.739

JACKSON LOW 1.447 0.137 0.145 0 032 1 378 1 210 0 948 1 378 1 210 0 948 HIGH 2 815 0 301 0 281 0.108 2 893 2 579 2 023 2 893 2 579 2 023 WGHTD AVE 2.159 0 217 0 217 0 066 2.150 1 903 1.483 2.150 1 903 1.483

JEFFERSON LOW 1 346 0.132 0.135 0 036 1 306 1.155 0 908 1 306 1.155 0 908 HIGH 1 526 0.153 0.153 0 044 1.499 1 327 1 043 1.499 1 327 1 043 WGHTD AVE 1 369 0.135 0.137 0 037 1 330 1.176 0 925 1 330 1.176 0 925

LAFAYETTE LOW 1.463 0.144 0.146 0 044 1.426 1 263 0 997 1.426 1 263 0 997 HIGH 1 815 0.188 0.181 0 067 1 816 1 616 1 276 1 816 1 616 1 276 WGHTD AVE 1 805 0.186 0.181 0 066 1 803 1 604 1 267 1 803 1 604 1 267

LAKE LOW 2 621 0244 0 262 0 082 2 573 2 264 1.753 2 573 2 264 1.753 HIGH 3.737 0 376 0 374 0.157 3 809 3 393 2 640 3 809 3 393 2 640 WGHTD AVE 2.793 0 262 0 287 0 094 2.736 2.416 1 875 2.736 2.416 1 875

LEE LOW 5850 0610 0 585 0 278 6 034 5.407 4 247 6 034 5.407 4 247 HIGH 18 213 3 361 1 821 2.115 22.103 20 809 18 048 22.103 20 809 18 048 WGHTD AVE 9.430 1 317 0 871 0.774 10 508 9 671 8 020 10 508 9 671 8 020

LEON LOW 1 515 0.152 0.152 0 044 1.492 1 324 1 045 1.492 1 324 1 045 HIGH 2 019 0 218 0 202 0 076 2 059 1 842 1.466 2 059 1 842 1.466 WGHTD AVE 1 818 0.191 0.182 0 063 1 830 1 633 1 297 1 830 1 633 1 297

LEVY LOW 2 388 0237 0 239 0 092 2 391 2.130 1 676 2 391 2.130 1 676 HIGH 7 521 1 349 0.752 0 814 8 969 8.403 7 267 8 969 8.403 7 267 WGHTD AVE 2 954 0 330 0 295 0.149 3 062 2.761 2 219 3 062 2.761 2 219

LIBERTY LOW 2.408 0261 0 241 0 095 2.469 2 205 1.746 2.469 2 205 1.746 HIGH 2 679 0 295 0 268 0.110 2.771 2.477 1 958 2.771 2.477 1 958 WGHTD AVE 2 577 0 282 0 258 0.105 2 657 2 375 1 878 2 657 2 375 1 878

MADISON LOW 1.128 0.106 0.113 0 027 1 076 0 949 0.747 1 076 0 949 0.747 HIGH 1 269 0.123 0.127 0 035 1 228 1 087 0 860 1 228 1 087 0 860 WGHTD AVE 1.189 0.113 0.119 0 030 1.139 1 006 0.794 1.139 1 006 0.794

MANATEE LOW 3 893 0390 0 389 0.162 3 914 3.492 2.741 3 914 3.492 2.741 HIGH 10.790 1 837 1 079 1.176 12.712 11 910 10 259 12.712 11 910 10 259 WGHTD AVE 6 977 0 970 0.726 0 578 7.767 7.157 5 965 7.767 7.157 5 965

MARION LOW 2.411 0239 0 241 0 091 2.400 2.136 1 685 2.400 2.136 1 685 HIGH 3.424 0 356 0 342 0.160 3 520 3.150 2.487 3 520 3.150 2.487 WGHTD AVE 2 853 0 289 0 287 0.121 2 884 2 572 2 026 2 884 2 572 2 026

MART N LOW 6.786 0.701 0 679 0 334 7 064 6 339 4 977 7 064 6 339 4 977 HIGH 17 338 3 037 1.734 2 065 20.799 19 567 16 935 20.799 19 567 16 935 WGHTD AVE 13.496 2.123 1 341 1 399 15.709 14 647 12.446 15.709 14 647 12.446

Form A-6: Output Ranges

MIAMI-DADE LOW 10 326 1 257 1 033 0.757 11.404 10.444 8 532 11.404 10.444 8 532 HIGH 36 319 8 060 3 632 5 847 45.762 43 988 39 879 45.762 43 988 39 879 WGHTD AVE 12.110 1 818 1.191 1 040 13 969 12 917 10.758 13 969 12 917 10.758

MONROE LOW 21 646 4.197 2.165 2 678 26 861 25.418 22 283 26 861 25.418 22 283 HIGH 29 898 7 015 2 990 4 331 39 206 37 582 33 859 39 206 37 582 33 859 WGHTD AVE 26 268 5 518 2 655 3 579 33.402 31 829 28 325 33.402 31 829 28 325

NASSAU LOW 1 517 0.152 0.152 0 054 1 500 1 334 1 057 1 500 1 334 1 057 HIGH 3.123 0.465 0 312 0 251 3.499 3 228 2.721 3.499 3 228 2.721 WGHTD AVE 1 890 0 216 0.192 0 094 1 947 1.755 1.422 1 947 1.755 1.422

OKALOOSA LOW 3 349 0364 0 335 0.141 3.479 3.106 2.430 3.479 3.106 2.430 HIGH 11 980 2.421 1.198 1 370 15 032 14.165 12 337 15 032 14.165 12 337 WGHTD AVE 6 237 0 975 0 608 0.492 7 201 6 639 5 535 7 201 6 639 5 535

OKEECHOBEE LOW 5 654 0 555 0 565 0 237 5.755 5.128 3 978 5.755 5.128 3 978 HIGH 6.141 0 610 0 614 0 269 6 294 5 618 4 367 6 294 5 618 4 367 WGHTD AVE 6 039 0 600 0 602 0 263 6.186 5 520 4 289 6.186 5 520 4 289

ORANGE LOW 2.706 0246 0 271 0 080 2 612 2 297 1.774 2 612 2 297 1.774 HIGH 3 978 0.402 0 398 0.168 4 050 3 612 2 825 4 050 3 612 2 825 WGHTD AVE 3 287 0 309 0 331 0.113 3 245 2 869 2 220 3 245 2 869 2 220

OSCEOLA LOW 3 264 0300 0 326 0.103 3.182 2 808 2.167 3.182 2 808 2.167 HIGH 4 232 0.407 0.423 0.160 4 231 3.758 2 919 4 231 3.758 2 919 WGHTD AVE 3 595 0 338 0 362 0.125 3 554 3.147 2.437 3 554 3.147 2.437

PALM BEACH LOW 7 377 0.776 0.738 0 389 7.745 6 969 5.494 7.745 6 969 5.494 HIGH 26 099 5 030 2 610 3 648 31 961 30 387 26 894 31 961 30 387 26 894 WGHTD AVE 14 512 2 242 1 328 1 531 16 804 15 664 13 279 16 804 15 664 13 279

PASCO LOW 3.151 0300 0 315 0.114 3.113 2.758 2.138 3.113 2.758 2.138 HIGH 7 835 1 281 0.784 0 809 9.101 8.486 7 251 9.101 8.486 7 251 WGHTD AVE 4 305 0 529 0.417 0 279 4 568 4.151 3 383 4 568 4.151 3 383

P NELLAS LOW 4 215 0.482 0.422 0 249 4.443 4 025 3 261 4.443 4 025 3 261 HIGH 11 957 2 245 1.196 1.463 14.437 13 633 11 952 14.437 13 633 11 952 WGHTD AVE 6 948 1 048 0.702 0 649 7 892 7 319 6.189 7 892 7 319 6.189

POLK LOW 2 994 0272 0 299 0 090 2 897 2 550 1 964 2 897 2 550 1 964 HIGH 3 811 0 352 0 381 0.125 3.730 3 294 2 541 3.730 3 294 2 541 WGHTD AVE 3 269 0 299 0 327 0.102 3.177 2 801 2.158 3.177 2 801 2.158

PUTNAM LOW 2.125 0203 0 213 0 071 2 097 1 858 1.445 2 097 1 858 1.445 HIGH 2 925 0 290 0 292 0.115 2 954 2 631 2 056 2 954 2 631 2 056 WGHTD AVE 2 616 0 258 0 261 0.101 2 633 2 344 1 833 2 633 2 344 1 833

SAINT JOHNS LOW 1 921 0.190 0.192 0 070 1 904 1 693 1 335 1 904 1 693 1 335 HIGH 5 563 0 885 0 556 0 513 6.419 5 959 5 041 6.419 5 959 5 041 WGHTD AVE 3 594 0.480 0 357 0 247 3 924 3 587 2 952 3 924 3 587 2 952

SAINT LUC E LOW 6 069 0 622 0 607 0 289 6 270 5 616 4.407 6 270 5 616 4.407 HIGH 15 640 2.729 1 564 1 816 18.732 17 601 15 200 18.732 17 601 15 200 WGHTD AVE 9 554 1 277 0 947 0.786 10 642 9.791 8 097 10 642 9.791 8 097

Form A-6: Output Ranges

LOSS COSTS PER $1,000 PERSONAL RESIDENTIAL - MOBILE HOMES $0 $0 $0 DEDUCTIBLE $0 DEDUCTIBLE $500 $1,000 $2,500 1% 2% 5% DEDUCTIBLE DEDUCTIBLE APPURTENANT ADDITIONAL DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE COUNTY LOSS COSTS STRUCTURE CONTENTS STRUCTURE LIVING EXPENSE TOTAL* TOTAL* TOTAL* TOTAL* TOTAL* TOTAL* SANTA ROSA LOW 3.729 0.418 0 373 0.183 3 944 3 543 2 803 3 944 3 543 2 803 HIGH 11.106 2.137 1.111 1 215 13.717 12 880 11.129 13.717 12 880 11.129 WGHTD AVE 6.417 1 031 0 644 0 556 7.483 6 911 5.780 7.483 6 911 5.780

SARASOTA LOW 4 237 0.437 0.424 0.195 4 285 3 825 3 011 4 285 3 825 3 011 HIGH 12 621 2 271 1 262 1.468 15.128 14 233 12 352 15.128 14 233 12 352 WGHTD AVE 8.430 1.199 0 861 0.729 9.474 8.743 7 297 9.474 8.743 7 297

SEM NOLE LOW 2 854 0262 0 285 0 088 2.774 2.444 1 889 2.774 2.444 1 889 HIGH 3 510 0 343 0 351 0.134 3 517 3.122 2.428 3 517 3.122 2.428 WGHTD AVE 3 087 0 291 0 311 0.104 3 043 2 691 2 085 3 043 2 691 2 085

SUMTER LOW 2 657 0257 0 266 0 098 2 628 2 329 1 821 2 628 2 329 1 821 HIGH 3 602 0 368 0 360 0.158 3 665 3 270 2 560 3 665 3 270 2 560 WGHTD AVE 3.105 0 311 0 314 0.130 3.120 2.778 2.178 3.120 2.778 2.178

SUWANEE LOW 1.410 0.137 0.141 0 041 1 372 1 214 0 954 1 372 1 214 0 954 HIGH 1 928 0.198 0.193 0 074 1 933 1.724 1 363 1 933 1.724 1 363 WGHTD AVE 1 511 0.149 0.151 0 048 1.481 1 313 1 033 1.481 1 313 1 033

TAYLOR LOW 1 692 0.175 0.169 0 057 1 682 1.497 1.189 1 682 1.497 1.189 HIGH 4.782 0.760 0.478 0 394 5 538 5.131 4 332 5 538 5.131 4 332 WGHTD AVE 2 278 0 274 0 230 0.115 2 388 2.160 1.757 2 388 2.160 1.757

UNION LOW 2.157 0214 0 216 0 080 2.165 1 927 1 507 2.165 1 927 1 507 HIGH 2.172 0 217 0 217 0 084 2.185 1 945 1 523 2.185 1 945 1 523 WGHTD AVE 2.159 0 215 0 216 0 081 2.167 1 929 1 509 2.167 1 929 1 509

VOLUSIA LOW 3 009 0301 0 301 0.117 3 043 2.712 2.121 3 043 2.712 2.121 HIGH 7.727 1 227 0.773 0.742 8 981 8 356 7 093 8 981 8 356 7 093 WGHTD AVE 5 515 0.786 0 561 0.432 6.181 5 683 4.718 6.181 5 683 4.718

WAKULLA LOW 2 594 0315 0 259 0.130 2.754 2.490 2 019 2.754 2.490 2 019 HIGH 5 321 0 908 0 532 0.452 6 279 5 821 4 929 6 279 5 821 4 929 WGHTD AVE 2.749 0 347 0 272 0.147 2 947 2 672 2.177 2 947 2 672 2.177

WALTON LOW 3 643 0.421 0 364 0.177 3 848 3.454 2.732 3 848 3.454 2.732 HIGH 9.171 1 643 0 917 0 895 11.105 10 360 8 837 11.105 10 360 8 837 WGHTD AVE 4 924 0 674 0.453 0 338 5.452 4 976 4 069 5.452 4 976 4 069

WASH NGTON LOW 3 030 0 332 0 303 0.126 3.140 2 803 2 203 3.140 2 803 2 203 HIGH 4.114 0 510 0.411 0 235 4.428 4 000 3 208 4.428 4 000 3 208 WGHTD AVE 3 240 0 365 0 322 0.144 3 385 3 030 2 391 3 385 3 030 2 391

STATEWIDE LOW 1.117 0.106 0.112 0 027 1 067 0 943 0.744 1 067 0 943 0.744 HIGH 36 319 8 060 3 632 5 847 45.762 43 988 39 879 45.762 43 988 39 879 WGHTD AVE 5 644 0.759 0 509 0.435 6 200 5 679 4 682 6 200 5 679 4 682

Form A-6: Output Ranges

ALACHUA LOW 0.116 0088 0.171 0.165 0.156 0.177 0.171 0.163 HIGH 0.187 0.163 0 291 0 278 0 261 0 302 0 291 0 274 WGHTD AVE 0.136 0.106 0 201 0.194 0.184 0 209 0 201 0.191

BAKER LOW 0.102 0072 0.144 0.139 0.132 0.150 0.144 0.137 HIGH 0.116 0 087 0.169 0.162 0.154 0.175 0.169 0.160 WGHTD AVE 0.115 0 087 0.167 0.161 0.153 0.174 0.167 0.159

BAY LOW 0250 0217 0 383 0 363 0 337 0.400 0 383 0 357 HIGH 2 294 3 617 4 099 3 903 3 534 4 229 4 099 3 824 WGHTD AVE 0 875 1 086 1.497 1.419 1 288 1 553 1.497 1 389

BRADFORD LOW 0.144 0.117 0 218 0 208 0.197 0 226 0 218 0 206 HIGH 0.155 0.128 0 235 0 225 0 213 0 244 0 235 0 222 WGHTD AVE 0.146 0.119 0 220 0 211 0 200 0 229 0 220 0 208

BREVARD LOW 0.193 0.157 0 291 0 280 0 264 0 301 0 291 0 276 HIGH 2 348 4.430 4.416 4 207 3 819 4 553 4.416 4.124 WGHTD AVE 0 618 0 837 1 090 1 036 0 949 1.130 1 090 1 016

BROWARD LOW 0 640 0.789 1.146 1 092 1 017 1.190 1.146 1 074 HIGH 5 936 13 385 11.164 10 690 9.744 11.463 11.164 10.494 WGHTD AVE 2 075 3.145 3 840 3 661 3 347 3 964 3 840 3 592

CALHOUN LOW 0.198 0.146 0 285 0 271 0 254 0 298 0 285 0 267 HIGH 0 355 0 361 0 581 0 549 0 503 0 608 0 581 0 538 WGHTD AVE 0 208 0.160 0 302 0 287 0 268 0 316 0 302 0 282

CHARLOTTE LOW 0 512 0 604 0 887 0 843 0.779 0 922 0 887 0 828 HIGH 1 971 3 563 3 668 3.489 3.169 3.789 3 668 3.419 WGHTD AVE 1.164 1.798 2 068 1 965 1.790 2.142 2 068 1 926

CITRUS LOW 0.171 0.157 0 270 0 258 0 242 0 279 0 270 0 254 HIGH 0 520 0.782 0 955 0 908 0 829 0 990 0 955 0 890 WGHTD AVE 0 298 0 374 0.475 0.453 0.419 0.492 0.475 0.445

CLAY LOW 0 092 0068 0.132 0.128 0.121 0.137 0.132 0.126 HIGH 0.147 0.124 0 225 0 216 0 203 0 233 0 225 0 213 WGHTD AVE 0.108 0 083 0.159 0.153 0.144 0.164 0.159 0.151

COLLIER LOW 0.499 0.469 0 812 0.774 0.726 0 846 0 812 0.762 HIGH 3 862 7 085 7.169 6 838 6 210 7 385 7.169 6.704 WGHTD AVE 2 084 3.461 3 640 3.469 3.166 3.758 3 640 3.403

COLUMBIA LOW 0 095 0070 0.137 0.132 0.125 0.142 0.137 0.130 HIGH 0.126 0.100 0.183 0.176 0.167 0.190 0.183 0.174 WGHTD AVE 0.111 0 084 0.160 0.154 0.145 0.166 0.160 0.152

DESOTO LOW 0 256 0217 0 390 0 376 0 357 0.404 0 390 0 372 HIGH 0 320 0 312 0 516 0.494 0.463 0 535 0 516 0.486 WGHTD AVE 0 257 0 219 0 393 0 379 0 359 0.407 0 393 0 374

DIXIE LOW 0.192 0202 0 304 0 290 0 270 0 315 0 304 0 286 HIGH 0 648 0 959 1.127 1 074 0 981 1.165 1.127 1 054 WGHTD AVE 0 203 0 210 0 322 0 308 0 285 0 334 0 322 0 302

Form A-6: Output Ranges

DUVAL LOW 0 074 0053 0.105 0.101 0 096 0.108 0.105 0.100 HIGH 0.453 0.705 0 812 0.772 0.704 0 839 0 812 0.757 WGHTD AVE 0.137 0.137 0 220 0 211 0.196 0 228 0 220 0 207

ESCAMBIA LOW 0 236 0.186 0 352 0 335 0 315 0 367 0 352 0 330 HIGH 1.794 2 869 3 297 3.129 2 824 3.410 3 297 3 062 WGHTD AVE 0 596 0.770 1 042 0 986 0 902 1 085 1 042 0 966

FLAGLER LOW 0.199 0.169 0 306 0 292 0 275 0 319 0 306 0 288 HIGH 1.108 1 926 2 041 1 944 1.768 2.108 2 041 1 906 WGHTD AVE 0.723 1.126 1 326 1 260 1.146 1 372 1 326 1 235

FRANKL N LOW 0 878 1208 1 551 1.472 1 332 1 606 1 551 1.441 HIGH 2 018 3 059 3 573 3.407 3 088 3 682 3 573 3 340 WGHTD AVE 1 608 2 320 3 067 2 922 2 648 3.164 3 067 2 863

GADSEN LOW 0.108 0067 0.143 0.138 0.131 0.149 0.143 0.136 HIGH 0.154 0.112 0 217 0 205 0.193 0 227 0 217 0 202 WGHTD AVE 0.130 0 089 0.178 0.170 0.160 0.185 0.178 0.167

GILCHRIST LOW 0.141 0.132 0 218 0 209 0.195 0 226 0 218 0 206 HIGH 0.150 0.137 0 233 0 223 0 209 0 242 0 233 0 220 WGHTD AVE 0.149 0.136 0 229 0 219 0 205 0 238 0 229 0 216

GLADES LOW 0.401 0350 0 631 0 605 0 572 0 656 0 631 0 597 HIGH 0.401 0 350 0 631 0 605 0 572 0 656 0 631 0 597 WGHTD AVE 0.401 0 350 0 631 0 605 0 572 0 656 0 631 0 597

GULF LOW 0 329 0341 0 537 0 508 0.466 0 560 0 537 0.498 HIGH 1 890 2 830 3 366 3 202 2 897 3.475 3 366 3.136 WGHTD AVE 1.168 1 579 2 066 1 964 1.780 2.136 2 066 1 924

HAMILTON LOW 0 060 0036 0 079 0 077 0 073 0 081 0 079 0 076 HIGH 0 091 0 065 0.128 0.124 0.117 0.133 0.128 0.122 WGHTD AVE 0 075 0 051 0.101 0 098 0 093 0.105 0.101 0 097

HARDEE LOW 0.192 0.140 0 278 0 269 0 257 0 286 0 278 0 266 HIGH 0 214 0.174 0 323 0 312 0 295 0 333 0 323 0 308 WGHTD AVE 0.198 0.148 0 289 0 280 0 268 0 298 0 289 0 278

HENDRY LOW 0 380 0330 0 596 0 571 0 540 0 620 0 596 0 564 HIGH 0.463 0.418 0.740 0.709 0 671 0.770 0.740 0 699 WGHTD AVE 0.410 0 356 0 642 0 615 0 582 0 668 0 642 0 607

HERNANDO LOW 0 221 0224 0 367 0 350 0 326 0 380 0 367 0 344 HIGH 0 846 1.468 1 556 1.481 1 350 1 609 1 556 1.452 WGHTD AVE 0 349 0.446 0 605 0 576 0 531 0 627 0 605 0 566

HIGHLANDS LOW 0 208 0.149 0 299 0 290 0 278 0 308 0 299 0 287 HIGH 0 296 0 234 0.445 0.429 0.409 0.461 0.445 0.424 WGHTD AVE 0 233 0.172 0 339 0 329 0 315 0 350 0 339 0 326

HILLSBOROUGH LOW 0.176 0.136 0 259 0 251 0 239 0 267 0 259 0 248 HIGH 1 078 1 912 1 993 1 894 1.725 2 063 1 993 1 856 WGHTD AVE 0 287 0 347 0.480 0.459 0.427 0.496 0.480 0.452

Form A-6: Output Ranges

HOLMES LOW 0.199 0.145 0 284 0 271 0 255 0 296 0 284 0 267 HIGH 0 227 0.178 0 334 0 318 0 297 0 348 0 334 0 313 WGHTD AVE 0 204 0.153 0 291 0 278 0 261 0 303 0 291 0 274

INDIAN RIVER LOW 0 325 0 301 0 521 0.499 0.469 0 541 0 521 0.492 HIGH 1 827 3 263 3.467 3 292 2 990 3 588 3.467 3 225 WGHTD AVE 0 976 1.425 1.776 1 684 1 538 1 844 1.776 1 650

JACKSON LOW 0 077 0040 0 095 0 093 0 090 0 098 0 095 0 093 HIGH 0 204 0.148 0 291 0 277 0 260 0 305 0 291 0 273 WGHTD AVE 0.130 0 080 0.171 0.165 0.157 0.177 0.171 0.163

JEFFERSON LOW 0 078 0 046 0.102 0 099 0 094 0.106 0.102 0 097 HIGH 0 094 0 059 0.125 0.120 0.113 0.129 0.125 0.118 WGHTD AVE 0 079 0 047 0.104 0.100 0 095 0.107 0.104 0 099

LAFAYETTE LOW 0 087 0057 0.119 0.114 0.108 0.123 0.119 0.112 HIGH 0.119 0 094 0.176 0.168 0.158 0.183 0.176 0.166 WGHTD AVE 0.119 0 094 0.175 0.168 0.157 0.182 0.175 0.165

LAKE LOW 0.142 0098 0 200 0.194 0.186 0 206 0 200 0.192 HIGH 0 237 0.198 0 364 0 349 0 330 0 378 0 364 0 344 WGHTD AVE 0.157 0.112 0 225 0 218 0 208 0 233 0 225 0 216

LEE LOW 0384 0349 0 615 0 588 0 553 0 639 0 615 0 579 HIGH 3 039 5 644 5 530 5 280 4 814 5 697 5 530 5.180 WGHTD AVE 1 081 1.467 1 924 1 832 1 677 1 991 1 924 1.797

LEON LOW 0 093 0057 0.123 0.118 0.111 0.128 0.123 0.116 HIGH 0.141 0.103 0 202 0.191 0.178 0 211 0 202 0.188 WGHTD AVE 0.103 0 066 0.141 0.135 0.127 0.147 0.141 0.133

LEVY LOW 0.147 0.119 0 221 0 212 0 200 0 229 0 221 0 209 HIGH 1.183 2 084 2.183 2 081 1 890 2 250 2.183 2 040 WGHTD AVE 0 283 0 293 0 373 0 356 0 330 0 387 0 373 0 350

LIBERTY LOW 0.166 0.125 0 240 0 228 0 214 0 251 0 240 0 225 HIGH 0.190 0.146 0 279 0 265 0 247 0 291 0 279 0 260 WGHTD AVE 0.183 0.139 0 259 0 246 0 230 0 271 0 259 0 242

MADISON LOW 0 060 0033 0 076 0 074 0 071 0 078 0 076 0 073 HIGH 0 072 0 045 0 096 0 092 0 088 0 099 0 096 0 091 WGHTD AVE 0 064 0 037 0 081 0 079 0 075 0 084 0 081 0 078

MANATEE LOW 0 233 0217 0 366 0 353 0 333 0 378 0 366 0 348 HIGH 1 549 2 901 2 849 2.713 2.471 2 942 2 849 2 660 WGHTD AVE 0 911 1 554 1 618 1 538 1.404 1 675 1 618 1 508

MARION LOW 0.145 0.113 0 219 0 211 0 200 0 226 0 219 0 208 HIGH 0 228 0 212 0 363 0 346 0 324 0 378 0 363 0 340 WGHTD AVE 0.170 0.151 0 268 0 257 0 242 0 278 0 268 0 254

MART N LOW 0.420 0393 0 681 0 652 0 615 0.708 0 681 0 643 HIGH 2 388 4 543 4 579 4 354 3 950 4.731 4 579 4 265 WGHTD AVE 1 320 2 288 2 539 2.412 2 201 2 629 2 539 2 364

Form A-6: Output Ranges

MIAMI-DADE LOW 0 815 1 081 1 518 1.441 1 332 1 580 1 518 1.414 HIGH 8.165 18 663 14 960 14 359 13.120 15 331 14 960 14.106 WGHTD AVE 4 043 7 944 7 635 7 299 6 654 7 854 7 635 7.162

MONROE LOW 3 529 6361 6 595 6 276 5 689 6 806 6 595 6.148 HIGH 6 802 12.792 12 321 11.795 10.727 12 648 12 321 11 574 WGHTD AVE 5.181 10.114 9 630 9.194 8 344 9 907 9 630 9 015

NASSAU LOW 0 093 0074 0.137 0.132 0.124 0.142 0.137 0.130 HIGH 0 389 0 607 0 682 0 650 0 593 0.705 0 682 0 638 WGHTD AVE 0 353 0 508 0 634 0 604 0 551 0 654 0 634 0 592

OKALOOSA LOW 0 241 0.191 0 358 0 342 0 321 0 373 0 358 0 337 HIGH 2.181 3 520 3 975 3.785 3.422 4.100 3 975 3.708 WGHTD AVE 1 217 1 821 2.190 2 075 1 874 2 268 2.190 2 031

OKEECHOBEE LOW 0 338 0 284 0 522 0 502 0.477 0 542 0 522 0.496 HIGH 0 375 0 325 0 587 0 563 0 534 0 609 0 587 0 556 WGHTD AVE 0 359 0 309 0 554 0 533 0 505 0 576 0 554 0 526

ORANGE LOW 0.142 0094 0.196 0.191 0.184 0 202 0.196 0.190 HIGH 0 251 0 213 0 389 0 372 0 350 0.404 0 389 0 366 WGHTD AVE 0.183 0.133 0 264 0 255 0 244 0 273 0 264 0 253

OSCEOLA LOW 0.168 0.117 0 238 0 232 0 222 0 245 0 238 0 230 HIGH 0 240 0.191 0 360 0 348 0 331 0 373 0 360 0 344 WGHTD AVE 0.187 0.136 0 271 0 263 0 251 0 280 0 271 0 260

PALM BEACH LOW 0.472 0.466 0.784 0.749 0.703 0 815 0.784 0.737 HIGH 4.450 9 577 8.421 8 041 7 320 8 669 8.421 7 887 WGHTD AVE 2 063 3 388 3.772 3 592 3 274 3 898 3.772 3 521

PASCO LOW 0.186 0.151 0 279 0 270 0 256 0 288 0 279 0 267 HIGH 1 036 1 846 1 970 1 874 1.702 2 035 1 970 1 836 WGHTD AVE 0 573 0 867 0 993 0 945 0 864 1 027 0 993 0 927

P NELLAS LOW 0 311 0366 0 538 0 511 0.472 0 559 0 538 0 501 HIGH 2 012 3 965 3.715 3 545 3 226 3 825 3.715 3.477 WGHTD AVE 0 952 1 579 1 882 1.790 1 629 1 945 1 882 1.755

POLK LOW 0.151 0.101 0 211 0 205 0.197 0 217 0 211 0 204 HIGH 0.199 0.142 0 285 0 277 0 266 0 294 0 285 0 275 WGHTD AVE 0.169 0.119 0 240 0 233 0 224 0 246 0 240 0 231

PUTNAM LOW 0.121 0089 0.174 0.169 0.161 0.180 0.174 0.167 HIGH 0.180 0.142 0 269 0 257 0 244 0 280 0 269 0 254 WGHTD AVE 0.156 0.123 0 234 0 225 0 213 0 244 0 234 0 222

SAINT JOHNS LOW 0.112 0 088 0.166 0.160 0.151 0.172 0.166 0.158 HIGH 0.711 1.158 1 307 1 245 1.134 1 350 1 307 1 221 WGHTD AVE 0.494 0.758 0 861 0 819 0.747 0 890 0 861 0 803

SAINT LUC E LOW 0 370 0 341 0 594 0 569 0 536 0 617 0 594 0 561 HIGH 2.125 3 938 4 058 3 858 3 500 4.195 4 058 3.779 WGHTD AVE 0.777 1 075 1.404 1 334 1 225 1.457 1.404 1 308

Form A-6: Output Ranges

LOSS COSTS PER $1,000 PERSONAL RESIDENTIAL - Renters -- FRAME $0 $0 DEDUCTIBLE $500 $1,000 $2,500 1% 2% 5% DEDUCTIBLE ADDITIONAL DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE COUNTY LOSS COSTS CONTENTS LIVING EXPENSE TOTAL* TOTAL* TOTAL* TOTAL* TOTAL* TOTAL* SANTA ROSA LOW 0 282 0 249 0.443 0.420 0 391 0.463 0.443 0.413 HIGH 1 872 2 992 3.425 3 257 2 948 3 537 3.425 3.191 WGHTD AVE 1 089 1 586 2 015 1 911 1.732 2 088 2 015 1 871

SARASOTA LOW 0 273 0291 0.439 0.422 0 397 0.454 0.439 0.417 HIGH 1 920 3 651 3 591 3.420 3.107 3.704 3 591 3 352 WGHTD AVE 1 096 1.795 1 961 1 863 1 698 2 029 1 961 1 826

SEM NOLE LOW 0.153 0.104 0 214 0 208 0 200 0 221 0 214 0 206 HIGH 0 210 0.167 0 315 0 303 0 287 0 327 0 315 0 299 WGHTD AVE 0.164 0.115 0 233 0 226 0 216 0 241 0 233 0 224

SUMTER LOW 0.155 0.124 0 232 0 224 0 212 0 240 0 232 0 221 HIGH 0 235 0 218 0 373 0 358 0 337 0 387 0 373 0 353 WGHTD AVE 0.182 0.155 0 271 0 260 0 245 0 280 0 271 0 257

SUWANEE LOW 0 082 0053 0.112 0.107 0.102 0.116 0.112 0.106 HIGH 0.127 0.106 0.191 0.183 0.171 0.198 0.191 0.180 WGHTD AVE 0 086 0 059 0.126 0.121 0.115 0.131 0.126 0.120

TAYLOR LOW 0.110 0079 0.155 0.148 0.139 0.162 0.155 0.146 HIGH 0 651 0 924 1.131 1 073 0 974 1.171 1.131 1 051 WGHTD AVE 0 226 0 241 0.444 0.422 0 386 0.461 0.444 0.414

UNION LOW 0.138 0.112 0 205 0.198 0.188 0 213 0 205 0.195 HIGH 0.139 0.114 0 209 0 201 0.190 0 217 0 209 0.198 WGHTD AVE 0.138 0.112 0 207 0.199 0.188 0 214 0 207 0.196

VOLUSIA LOW 0.186 0.145 0 276 0 266 0 252 0 286 0 276 0 262 HIGH 0 978 1 619 1 816 1.725 1 567 1 879 1 816 1 689 WGHTD AVE 0 638 0 963 1.150 1 093 0 996 1.192 1.150 1 071

WAKULLA LOW 0 213 0.193 0 337 0 317 0 289 0 352 0 337 0 311 HIGH 0.778 1 066 1 358 1 289 1.167 1.405 1 358 1 263 WGHTD AVE 0 253 0 225 0.400 0 377 0 343 0.417 0.400 0 369

WALTON LOW 0 287 0262 0.446 0.424 0 394 0.465 0.446 0.417 HIGH 1.415 2.109 2 517 2 388 2.161 2 605 2 517 2 338 WGHTD AVE 1.151 1 687 1 951 1 851 1 676 2 021 1 951 1 812

WASH NGTON LOW 0 227 0.178 0 335 0 318 0 297 0 350 0 335 0 313 HIGH 0 375 0.410 0 612 0 581 0 534 0 638 0 612 0 570 WGHTD AVE 0 236 0.181 0 365 0 346 0 322 0 381 0 365 0 340

STATEWIDE LOW 0 060 0033 0 076 0 074 0 071 0 078 0 076 0 073 HIGH 8.165 18 663 14 960 14 359 13.120 15 331 14 960 14.106 WGHTD AVE 0 517 0 671 0 883 0 841 0.770 0 914 0 883 0 825

Form A-6: Output Ranges

LOSS COSTS PER $1,000 PERSONAL RESIDENTIAL - Renters -- MASONRY 0% $0 DEDUCTIBLE $500 $1,000 $2,500 1% 2% 5% DEDUCTIBLE ADDITIONAL DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE COUNTY LOSS COSTS CONTENTS LIVING EXPENSE TOTAL* TOTAL* TOTAL* TOTAL* TOTAL* TOTAL*

ALACHUA LOW 0 081 0037 0 097 0 094 0 089 0.100 0 097 0 093 HIGH 0.130 0 068 0.160 0.154 0.145 0.166 0.160 0.152 WGHTD AVE 0 096 0 045 0.115 0.111 0.106 0.119 0.115 0.110

BAKER LOW 0 072 0031 0 082 0 080 0 076 0 085 0 082 0 079 HIGH 0 082 0 037 0 096 0 093 0 088 0 099 0 096 0 092 WGHTD AVE 0 081 0 037 0 094 0 091 0 086 0 097 0 094 0 090

BAY LOW 0.169 0089 0 208 0.198 0.185 0 218 0 208 0.195 HIGH 1 393 1 870 2 361 2 231 1 997 2.451 2 361 2.180 WGHTD AVE 0.721 0 652 1 059 0 997 0 897 1.103 1 059 0 975

BRADFORD LOW 0.101 0050 0.122 0.117 0.111 0.126 0.122 0.116 HIGH 0.108 0 054 0.131 0.126 0.120 0.136 0.131 0.125 WGHTD AVE 0.101 0 050 0.122 0.118 0.112 0.127 0.122 0.116

BREVARD LOW 0.134 0067 0.163 0.158 0.149 0.169 0.163 0.156 HIGH 1.469 2 372 2 623 2.482 2 233 2.719 2 623 2.427 WGHTD AVE 0.458 0.458 0.711 0 672 0 612 0.741 0.711 0 658

BROWARD LOW 0.427 0325 0 611 0 583 0 543 0 636 0 611 0 573 HIGH 3.739 7 641 6 917 6 582 5 945 7.134 6 917 6.446 WGHTD AVE 1 247 1 387 2 042 1 938 1.763 2.117 2 042 1 899

CALHOUN LOW 0.137 0060 0.158 0.151 0.143 0.164 0.158 0.149 HIGH 0 234 0.150 0 310 0 293 0 269 0 325 0 310 0 287 WGHTD AVE 0.143 0 066 0.167 0.160 0.150 0.174 0.167 0.157

CHARLOTTE LOW 0 342 0 255 0.476 0.452 0.418 0.497 0.476 0.444 HIGH 1 252 1 883 2.133 2 017 1 819 2 215 2.133 1 972 WGHTD AVE 0.731 0 922 1 205 1.139 1 031 1 254 1 205 1.114

CITRUS LOW 0.117 0066 0.148 0.142 0.133 0.153 0.148 0.140 HIGH 0 329 0 356 0 523 0.494 0.449 0 544 0 523 0.484 WGHTD AVE 0.176 0.133 0 241 0 229 0 212 0 250 0 241 0 225

CLAY LOW 0 065 0030 0 077 0 074 0 071 0 079 0 077 0 074 HIGH 0.102 0 051 0.125 0.120 0.113 0.129 0.125 0.118 WGHTD AVE 0 080 0 038 0 093 0 090 0 086 0 097 0 093 0 089

COLLIER LOW 0 341 0.188 0.434 0.416 0 392 0.452 0.434 0.410 HIGH 2 396 3.783 4 202 3 977 3 575 4 353 4 202 3 889 WGHTD AVE 1.434 2 063 2 321 2 200 1 988 2.408 2 321 2.153

COLUMBIA LOW 0 067 0030 0 078 0 076 0 072 0 080 0 078 0 075 HIGH 0 088 0 044 0.103 0.100 0 095 0.107 0.103 0 099 WGHTD AVE 0 079 0 036 0 091 0 088 0 084 0 094 0 091 0 087

DESOTO LOW 0.178 0093 0 220 0 213 0 202 0 227 0 220 0 211 HIGH 0 219 0.133 0 285 0 274 0 258 0 296 0 285 0 270 WGHTD AVE 0.187 0.103 0 228 0 221 0 210 0 236 0 228 0 219

DIXIE LOW 0.128 0091 0.171 0.163 0.152 0.177 0.171 0.161 HIGH 0.418 0 504 0 653 0 619 0 562 0 678 0 653 0 607 WGHTD AVE 0.133 0 095 0.179 0.171 0.158 0.186 0.179 0.168

Form A-6: Output Ranges

DUVAL LOW 0 053 0023 0 060 0 058 0 055 0 062 0 060 0 058 HIGH 0 292 0 357 0.465 0.440 0 399 0.483 0.465 0.430 WGHTD AVE 0 089 0 058 0.118 0.113 0.105 0.122 0.118 0.111

ESCAMBIA LOW 0.162 0075 0.192 0.185 0.175 0 200 0.192 0.182 HIGH 1 097 1.422 1 850 1.741 1 556 1 926 1 850 1.700 WGHTD AVE 0.412 0 360 0 601 0 567 0 516 0 628 0 601 0 555

FLAGLER LOW 0.137 0070 0.168 0.161 0.152 0.175 0.168 0.159 HIGH 0 690 0 989 1.197 1.132 1 020 1 242 1.197 1.108 WGHTD AVE 0.427 0.479 0 686 0 647 0 583 0.714 0 686 0 633

FRANKL N LOW 0 534 0566 0 853 0 803 0.717 0 890 0 853 0.783 HIGH 1 252 1 637 2.102 1 989 1.783 2.178 2.102 1 944 WGHTD AVE 0 943 1 098 1 589 1 502 1 345 1 649 1 589 1.467

GADSEN LOW 0 076 0029 0 083 0 081 0 077 0 086 0 083 0 080 HIGH 0.108 0 046 0.121 0.117 0.111 0.126 0.121 0.115 WGHTD AVE 0 092 0 037 0.101 0 097 0 092 0.105 0.101 0 096

GILCHRIST LOW 0 097 0057 0.122 0.117 0.109 0.126 0.122 0.115 HIGH 0.102 0 059 0.129 0.124 0.116 0.134 0.129 0.123 WGHTD AVE 0.102 0 059 0.129 0.124 0.116 0.134 0.129 0.122

GLADES LOW 0 279 0.144 0 346 0 334 0 317 0 359 0 346 0 330 HIGH 0 279 0.144 0 346 0 334 0 317 0 359 0 346 0 330 WGHTD AVE 0 279 0.144 0 346 0 334 0 317 0 359 0 346 0 330

GULF LOW 0 216 0.143 0 289 0 273 0 251 0 302 0 289 0 268 HIGH 1.168 1.463 1 932 1 823 1 632 2 007 1 932 1.781 WGHTD AVE 0 899 0 910 1 522 1.436 1 288 1 581 1 522 1.403

HAMILTON LOW 0 043 0016 0 047 0 046 0 044 0 048 0 047 0 045 HIGH 0 063 0 029 0 074 0 072 0 068 0 076 0 074 0 071 WGHTD AVE 0 056 0 024 0 061 0 060 0 057 0 063 0 061 0 059

HARDEE LOW 0.136 0061 0.160 0.156 0.149 0.164 0.160 0.154 HIGH 0.150 0 075 0.182 0.177 0.167 0.188 0.182 0.175 WGHTD AVE 0.140 0 065 0.165 0.161 0.154 0.170 0.165 0.160

HENDRY LOW 0 263 0.136 0 326 0 315 0 298 0 338 0 326 0 311 HIGH 0 320 0.171 0.403 0 389 0 369 0.418 0.403 0 384 WGHTD AVE 0 291 0.151 0 362 0 349 0 331 0 375 0 362 0 345

HERNANDO LOW 0.150 0093 0.198 0.189 0.176 0 205 0.198 0.186 HIGH 0 531 0.736 0 885 0 838 0.759 0 920 0 885 0 820 WGHTD AVE 0 231 0 203 0 333 0 316 0 291 0 347 0 333 0 310

HIGHLANDS LOW 0.148 0065 0.173 0.169 0.162 0.177 0.173 0.168 HIGH 0 208 0 098 0 250 0 243 0 232 0 258 0 250 0 240 WGHTD AVE 0.164 0 074 0.194 0.189 0.181 0.199 0.194 0.188

HILLSBOROUGH LOW 0.123 0 059 0.149 0.145 0.138 0.153 0.149 0.143 HIGH 0 683 0 973 1.140 1 078 0 977 1.185 1.140 1 054 WGHTD AVE 0.190 0.148 0 270 0 258 0 240 0 279 0 270 0 254

Form A-6: Output Ranges

HOLMES LOW 0.137 0060 0.158 0.153 0.144 0.165 0.158 0.151 HIGH 0.156 0 073 0.184 0.176 0.166 0.192 0.184 0.174 WGHTD AVE 0.138 0 060 0.159 0.153 0.145 0.165 0.159 0.151

INDIAN RIVER LOW 0 224 0.125 0 285 0 274 0 258 0 295 0 285 0 270 HIGH 1.148 1 644 1 963 1 852 1 670 2 044 1 963 1 810 WGHTD AVE 0 556 0 577 0 891 0 842 0.767 0 929 0 891 0 824

JACKSON LOW 0 055 0019 0 059 0 058 0 055 0 060 0 059 0 057 HIGH 0.141 0 061 0.161 0.155 0.146 0.168 0.161 0.153 WGHTD AVE 0 096 0 036 0.103 0.100 0 096 0.106 0.103 0 099

JEFFERSON LOW 0 055 0 021 0 061 0 059 0 056 0 062 0 061 0 058 HIGH 0 066 0 026 0 073 0 071 0 067 0 075 0 073 0 070 WGHTD AVE 0 056 0 021 0 061 0 060 0 057 0 063 0 061 0 059

LAFAYETTE LOW 0 061 0025 0 068 0 066 0 063 0 070 0 068 0 065 HIGH 0 082 0 039 0 098 0 094 0 088 0.101 0 098 0 093 WGHTD AVE 0 076 0 033 0 091 0 087 0 082 0 094 0 091 0 086

LAKE LOW 0.101 0043 0.117 0.114 0.109 0.120 0.117 0.113 HIGH 0.165 0 083 0 203 0.196 0.185 0 210 0 203 0.193 WGHTD AVE 0.114 0 050 0.131 0.128 0.122 0.135 0.131 0.127

LEE LOW 0263 0.143 0 334 0 321 0 303 0 347 0 334 0 317 HIGH 1 904 3 058 3 303 3.135 2 832 3.418 3 303 3 069 WGHTD AVE 0 656 0.721 0 985 0 933 0 850 1 024 0 985 0 914

LEON LOW 0 065 0025 0 071 0 069 0 065 0 073 0 071 0 068 HIGH 0 097 0 042 0.111 0.106 0.100 0.116 0.111 0.104 WGHTD AVE 0 073 0 028 0 081 0 078 0 074 0 085 0 081 0 077

LEVY LOW 0.102 0049 0.122 0.118 0.112 0.127 0.122 0.117 HIGH 0.738 1.105 1 273 1 204 1 084 1 319 1 273 1.178 WGHTD AVE 0.173 0.118 0 211 0 202 0.187 0 219 0 211 0.198

LIBERTY LOW 0.114 0051 0.132 0.127 0.119 0.138 0.132 0.125 HIGH 0.131 0 059 0.152 0.146 0.137 0.159 0.152 0.143 WGHTD AVE 0.126 0 058 0.151 0.144 0.136 0.158 0.151 0.142

MADISON LOW 0 043 0015 0 046 0 045 0 043 0 047 0 046 0 045 HIGH 0 051 0 020 0 056 0 054 0 052 0 058 0 056 0 054 WGHTD AVE 0 045 0 017 0 048 0 047 0 045 0 049 0 048 0 046

MANATEE LOW 0.161 0094 0 205 0.198 0.187 0 212 0 205 0.196 HIGH 0 976 1 543 1 672 1 584 1.432 1.735 1 672 1 550 WGHTD AVE 0 562 0.734 0 926 0 876 0.796 0 963 0 926 0 858

MARION LOW 0.101 0048 0.124 0.121 0.114 0.128 0.124 0.119 HIGH 0.157 0 089 0.197 0.188 0.177 0 205 0.197 0.186 WGHTD AVE 0.120 0 066 0.151 0.145 0.137 0.157 0.151 0.144

MART N LOW 0 290 0.159 0 368 0 354 0 335 0 382 0 368 0 349 HIGH 1.492 2 319 2 632 2.486 2 239 2.736 2 632 2.430 WGHTD AVE 0.787 0 978 1.410 1 333 1 210 1.468 1.410 1 304

Form A-6: Output Ranges

MIAMI-DADE LOW 0 537 0.434 0.797 0.755 0 699 0 832 0.797 0.741 HIGH 5 226 11.184 9 570 9.134 8 272 9 846 9 570 8 954 WGHTD AVE 2 570 3.789 4.450 4 228 3 822 4 599 4.450 4.139

MONROE LOW 2 206 3357 3 864 3 651 3 281 4 011 3 864 3 569 HIGH 4 318 7.434 7.750 7 371 6 633 7 992 7.750 7 215 WGHTD AVE 2 976 4 815 5 324 5 042 4 527 5 512 5 324 4 930

NASSAU LOW 0 065 0032 0 078 0 075 0 070 0 080 0 078 0 074 HIGH 0 254 0 328 0.403 0 382 0 347 0.418 0.403 0 374 WGHTD AVE 0 237 0 223 0 354 0 336 0 305 0 367 0 354 0 329

OKALOOSA LOW 0.165 0078 0.197 0.189 0.178 0 204 0.197 0.187 HIGH 1 316 1.782 2 265 2.136 1 907 2 351 2 265 2 086 WGHTD AVE 0 814 0 967 1 343 1 261 1.126 1.401 1 343 1 230

OKEECHOBEE LOW 0 236 0.118 0 289 0 280 0 267 0 299 0 289 0 277 HIGH 0 261 0.134 0 323 0 312 0 297 0 334 0 323 0 309 WGHTD AVE 0 249 0.127 0 308 0 297 0 283 0 318 0 308 0 294

ORANGE LOW 0.101 0042 0.116 0.114 0.109 0.119 0.116 0.113 HIGH 0.175 0 087 0 213 0 205 0.194 0 221 0 213 0 202 WGHTD AVE 0.129 0 057 0.151 0.147 0.141 0.155 0.151 0.146

OSCEOLA LOW 0.120 0052 0.139 0.136 0.130 0.142 0.139 0.135 HIGH 0.168 0 080 0 202 0.197 0.187 0 208 0 202 0.195 WGHTD AVE 0.133 0 060 0.156 0.152 0.146 0.160 0.156 0.151

PALM BEACH LOW 0 323 0.188 0.419 0.402 0 379 0.436 0.419 0 397 HIGH 2 832 5 397 5.119 4 860 4 393 5 294 5.119 4.758 WGHTD AVE 1 272 1.769 2.143 2 030 1 840 2 225 2.143 1 987

PASCO LOW 0.130 0065 0.158 0.154 0.146 0.163 0.158 0.152 HIGH 0 636 0 901 1.102 1 040 0 936 1.145 1.102 1 017 WGHTD AVE 0.410 0.484 0 693 0 655 0 593 0.721 0 693 0 641

P NELLAS LOW 0 206 0.147 0 283 0 268 0 248 0 295 0 283 0 263 HIGH 1 264 2.184 2 222 2.107 1 900 2 300 2 222 2 062 WGHTD AVE 0 632 0 817 1 090 1 031 0 931 1.133 1 090 1 008

POLK LOW 0.108 0046 0.124 0.122 0.117 0.127 0.124 0.121 HIGH 0.141 0 062 0.165 0.162 0.155 0.169 0.165 0.160 WGHTD AVE 0.119 0 052 0.139 0.136 0.130 0.142 0.139 0.135

PUTNAM LOW 0 086 0038 0.100 0 098 0 093 0.103 0.100 0 097 HIGH 0.126 0 060 0.150 0.145 0.138 0.156 0.150 0.144 WGHTD AVE 0.112 0 053 0.135 0.131 0.124 0.140 0.135 0.129

SAINT JOHNS LOW 0 079 0 038 0 094 0 091 0 086 0 097 0 094 0 090 HIGH 0.444 0 578 0.752 0.712 0 642 0.780 0.752 0 696 WGHTD AVE 0 329 0 375 0 525 0.497 0.449 0 546 0 525 0.486

SAINT LUC E LOW 0 255 0.139 0 323 0 310 0 294 0 335 0 323 0 307 HIGH 1 330 2 015 2 329 2.199 1 981 2.421 2 329 2.150 WGHTD AVE 0 556 0 540 0 912 0 863 0.787 0 951 0 912 0 845

Form A-6: Output Ranges

SANTA ROSA LOW 0.192 0 099 0 236 0 225 0 211 0 247 0 236 0 222 HIGH 1.138 1.499 1 938 1 828 1 636 2 014 1 938 1.786 WGHTD AVE 0 833 0.781 1 311 1 235 1.108 1 367 1 311 1 206

SARASOTA LOW 0.186 0.132 0 251 0 242 0 227 0 259 0 251 0 239 HIGH 1 205 1 928 2.109 1 995 1.797 2.188 2.109 1 951 WGHTD AVE 0.757 0 949 1.195 1.130 1 023 1 243 1.195 1.106

SEM NOLE LOW 0.109 0047 0.125 0.123 0.118 0.128 0.125 0.122 HIGH 0.147 0 071 0.177 0.171 0.162 0.183 0.177 0.169 WGHTD AVE 0.118 0 052 0.137 0.133 0.128 0.140 0.137 0.132

SUMTER LOW 0.109 0053 0.131 0.127 0.120 0.135 0.131 0.125 HIGH 0.162 0 091 0 206 0.198 0.187 0 213 0 206 0.196 WGHTD AVE 0.134 0 074 0.167 0.161 0.151 0.173 0.167 0.159

SUWANEE LOW 0 058 0023 0 065 0 063 0 060 0 067 0 065 0 062 HIGH 0 088 0 046 0.107 0.103 0 096 0.111 0.107 0.101 WGHTD AVE 0 061 0 026 0 070 0 068 0 065 0 073 0 070 0 067

TAYLOR LOW 0 076 0034 0 088 0 084 0 079 0 091 0 088 0 083 HIGH 0.413 0.465 0 646 0 609 0 550 0 672 0 646 0 596 WGHTD AVE 0.125 0.124 0.198 0.188 0.172 0 206 0.198 0.184

UNION LOW 0 096 0047 0.115 0.112 0.106 0.119 0.115 0.111 HIGH 0 097 0 048 0.117 0.114 0.108 0.121 0.117 0.112 WGHTD AVE 0 096 0 048 0.116 0.112 0.107 0.120 0.116 0.111

VOLUSIA LOW 0.130 0062 0.156 0.151 0.143 0.161 0.156 0.149 HIGH 0 620 0 807 1 039 0 981 0 886 1 081 1 039 0 959 WGHTD AVE 0 383 0 368 0 590 0 558 0 507 0 614 0 590 0 546

WAKULLA LOW 0.142 0078 0.178 0.168 0.154 0.188 0.178 0.164 HIGH 0.481 0 524 0.765 0.721 0 646 0.796 0.765 0.704 WGHTD AVE 0 274 0.127 0 357 0 336 0 304 0 373 0 357 0 329

WALTON LOW 0.193 0.110 0 244 0 233 0 218 0 254 0 244 0 230 HIGH 0 887 1 066 1.423 1 340 1 204 1.481 1.423 1 309 WGHTD AVE 0.762 0 689 1.172 1.102 0 987 1 222 1.172 1 076

WASH NGTON LOW 0.156 0 072 0.184 0.176 0.165 0.192 0.184 0.173 HIGH 0 248 0.182 0 333 0 316 0 291 0 348 0 333 0 310 WGHTD AVE 0.173 0 084 0.193 0.185 0.173 0 202 0.193 0.182

STATEWIDE LOW 0 043 0015 0 046 0 045 0 043 0 047 0 046 0 045 HIGH 5 226 11.184 9 570 9.134 8 272 9 846 9 570 8 954 WGHTD AVE 0.766 0 903 1.130 1 072 0 974 1.172 1.130 1 050

Form A-6: Output Ranges

LOSS COSTS PER $1,000 PERSONAL RESIDENTIAL - Condo Owners -- FRAME $0 $0 $0 DEDUCTIBLE $500 $1,000 $2,500 1% 2% 5% DEDUCTIBLE DEDUCTIBLE ADDITIONAL DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE COUNTY LOSS COSTS STRUCTURE CONTENTS LIVING EXPENSE TOTAL* TOTAL* TOTAL* TOTAL* TOTAL* TOTAL*

ALACHUA LOW 0 054 0.116 0 044 0.163 0.152 0.135 0.163 0.152 0.135 HIGH 0 083 0.187 0 081 0 271 0 250 0 219 0 271 0 250 0 219 WGHTD AVE 0 063 0.135 0 053 0.190 0.176 0.155 0.190 0.176 0.155

BAKER LOW 0 048 0.102 0 036 0.139 0.129 0.114 0.139 0.129 0.114 HIGH 0 054 0.116 0 044 0.162 0.150 0.132 0.162 0.150 0.132 WGHTD AVE 0 000 0 000 0 000 0 000 0 000 0 000 0 000 0 000 0 000

BAY LOW 0.100 0250 0.108 0 361 0 333 0 288 0 361 0 333 0 288 HIGH 0.486 2 294 1 809 3 886 3.716 3 375 3 886 3.716 3 375 WGHTD AVE 0 344 1 558 1 054 2 538 2.414 2.173 2 538 2.414 2.173

BRADFORD LOW 0 066 0.144 0 059 0 206 0.191 0.167 0 206 0.191 0.167 HIGH 0 070 0.155 0 064 0 221 0 205 0.180 0 221 0 205 0.180 WGHTD AVE 0 000 0 000 0 000 0 000 0 000 0 000 0 000 0 000 0 000

BREVARD LOW 0 090 0.193 0 078 0 275 0 256 0 225 0 275 0 256 0 225 HIGH 0 549 2 348 2 215 4.181 3 998 3 632 4.181 3 998 3 632 WGHTD AVE 0 230 0.704 0.498 1.187 1.118 0 993 1.187 1.118 0 993

BROWARD LOW 0 269 0640 0 394 1 038 0 964 0 846 1 038 0 964 0 846 HIGH 1 264 5 936 6 693 10.795 10.407 9 591 10.795 10.407 9 591 WGHTD AVE 0 566 2 096 2 015 3.709 3 530 3.194 3.709 3 530 3.194

CALHOUN LOW 0 083 0.198 0 073 0 273 0 251 0 218 0 273 0 251 0 218 HIGH 0.130 0 355 0.181 0 538 0.497 0.430 0 538 0.497 0.430 WGHTD AVE 0 086 0 208 0 000 0 292 0 269 0 232 0 292 0 269 0 232

CHARLOTTE LOW 0 205 0 512 0 302 0 814 0.757 0 662 0 814 0.757 0 662 HIGH 0 511 1 971 1.781 3.440 3 273 2 951 3.440 3 273 2 951 WGHTD AVE 0 381 1 538 1 221 2 543 2.412 2.164 2 543 2.412 2.164

CITRUS LOW 0 076 0.171 0 079 0 253 0 236 0 207 0 253 0 236 0 207 HIGH 0.170 0 520 0 391 0 884 0 833 0.741 0 884 0 833 0.741 WGHTD AVE 0.119 0 306 0.187 0 501 0.470 0.415 0 501 0.470 0.415

CLAY LOW 0 044 0092 0 034 0.128 0.120 0.106 0.128 0.120 0.106 HIGH 0 067 0.147 0 062 0 212 0.196 0.172 0 212 0.196 0.172 WGHTD AVE 0 050 0.105 0 040 0.148 0.138 0.122 0.148 0.138 0.122

COLLIER LOW 0 219 0.499 0 234 0.742 0 683 0 595 0.742 0 683 0 595 HIGH 0 860 3 862 3 543 6.794 6 507 5 929 6.794 6 507 5 929 WGHTD AVE 0 626 2 538 2 269 4.418 4 215 3 826 4.418 4 215 3 826

COLUMBIA LOW 0 045 0095 0 035 0.132 0.123 0.109 0.132 0.123 0.109 HIGH 0 058 0.126 0 050 0.175 0.162 0.142 0.175 0.162 0.142 WGHTD AVE 0 051 0.110 0 041 0.153 0.142 0.125 0.153 0.142 0.125

DESOTO LOW 0.121 0256 0.109 0 368 0 342 0 303 0 368 0 342 0 303 HIGH 0.142 0 320 0.156 0.481 0.447 0 393 0.481 0.447 0 393 WGHTD AVE 0.123 0 267 0.119 0 385 0 358 0 317 0 385 0 358 0 317

DIXIE LOW 0 076 0.192 0.101 0 289 0 270 0 239 0 289 0 270 0 239 HIGH 0.174 0 648 0.480 1 066 1 013 0 912 1 066 1 013 0 912 WGHTD AVE 0 000 0 000 0 000 0 000 0 000 0 000 0 000 0 000 0 000

Form A-6: Output Ranges

DUVAL LOW 0 036 0074 0 027 0.102 0 096 0 084 0.102 0 096 0 084 HIGH 0.133 0.453 0 353 0.761 0.721 0 646 0.761 0.721 0 646 WGHTD AVE 0 053 0.128 0 060 0.189 0.177 0.157 0.189 0.177 0.157

ESCAMBIA LOW 0.102 0236 0 093 0 332 0 305 0 265 0 332 0 305 0 265 HIGH 0.431 1.794 1.434 3 081 2 930 2 633 3 081 2 930 2 633 WGHTD AVE 0 226 0 813 0 519 1 283 1 203 1 062 1 283 1 203 1 062

FLAGLER LOW 0 089 0.199 0 084 0 287 0 265 0 231 0 287 0 265 0 231 HIGH 0 279 1.108 0 963 1 931 1 840 1 664 1 931 1 840 1 664 WGHTD AVE 0 230 0 866 0 686 1.484 1.408 1 265 1.484 1.408 1 265

FRANKL N LOW 0 223 0878 0 604 1.452 1 377 1 233 1.452 1 377 1 233 HIGH 0.414 2 018 1 530 3 393 3 250 2 962 3 393 3 250 2 962 WGHTD AVE 0 283 1 240 0 960 2 086 1 989 1.798 2 086 1 989 1.798

GADSEN LOW 0 048 0.108 0 034 0.142 0.131 0.115 0.142 0.131 0.115 HIGH 0 067 0.154 0 056 0 208 0.192 0.167 0 208 0.192 0.167 WGHTD AVE 0 048 0 000 0 000 0.146 0.135 0.118 0.146 0.135 0.118

GILCHRIST LOW 0 061 0.141 0 066 0 208 0.194 0.171 0 208 0.194 0.171 HIGH 0 065 0.150 0 068 0 220 0 205 0.180 0 220 0 205 0.180 WGHTD AVE 0 000 0.150 0 000 0 220 0 205 0.180 0 220 0 205 0.180

GLADES LOW 0.184 0.401 0.175 0 583 0 539 0.472 0 583 0 539 0.472 HIGH 0.184 0.401 0.175 0 583 0 539 0.472 0 583 0 539 0.472 WGHTD AVE 0.184 0.401 0 000 0 583 0 539 0.472 0 583 0 539 0.472

GULF LOW 0.121 0329 0.170 0 500 0.463 0.404 0 500 0.463 0.404 HIGH 0.414 1 890 1.415 3.173 3 028 2.740 3.173 3 028 2.740 WGHTD AVE 0.414 1 890 1.415 3.173 3 028 2.740 3.173 3 028 2.740

HAMILTON LOW 0 029 0060 0 018 0 079 0 074 0 066 0 079 0 074 0 066 HIGH 0 042 0 091 0 033 0.124 0.116 0.103 0.124 0.116 0.103 WGHTD AVE 0 000 0 000 0 000 0 000 0 000 0 000 0 000 0 000 0 000

HARDEE LOW 0 095 0.192 0 070 0 266 0 248 0 221 0 266 0 248 0 221 HIGH 0.104 0 214 0 087 0 305 0 285 0 252 0 305 0 285 0 252 WGHTD AVE 0 095 0.192 0 070 0 266 0 248 0 221 0 266 0 248 0 221

HENDRY LOW 0.174 0380 0.165 0 552 0 510 0.447 0 552 0 510 0.447 HIGH 0 211 0.463 0 209 0 679 0 626 0 549 0 679 0 626 0 549 WGHTD AVE 0.195 0.433 0.188 0 627 0 578 0 507 0 627 0 578 0 507

HERNANDO LOW 0 097 0221 0.112 0 338 0 315 0 275 0 338 0 315 0 275 HIGH 0 239 0 846 0.734 1.464 1 390 1 249 1.464 1 390 1 249 WGHTD AVE 0.139 0.442 0 356 0.715 0 673 0 599 0.715 0 673 0 599

HIGHLANDS LOW 0.103 0208 0 075 0 286 0 267 0 238 0 286 0 267 0 238 HIGH 0.141 0 296 0.117 0.418 0 387 0 342 0.418 0 387 0 342 WGHTD AVE 0.120 0 238 0 087 0 332 0 309 0 275 0 332 0 309 0 275

HILLSBOROUGH LOW 0 085 0.176 0 068 0 247 0 231 0 206 0 247 0 231 0 206 HIGH 0 308 1 078 0 956 1 877 1.779 1 595 1 877 1.779 1 595 WGHTD AVE 0.120 0 300 0.176 0.467 0.437 0 387 0.467 0.437 0 387

Form A-6: Output Ranges

HOLMES LOW 0 085 0.199 0 072 0 273 0 252 0 220 0 273 0 252 0 220 HIGH 0 095 0 227 0 089 0 317 0 293 0 255 0 317 0 293 0 255 WGHTD AVE 0 000 0 000 0 000 0 000 0 000 0 000 0 000 0 000 0 000

INDIAN RIVER LOW 0.146 0 325 0.151 0.482 0.447 0 393 0.482 0.447 0 393 HIGH 0 500 1 827 1 632 3 223 3 056 2.738 3 223 3 056 2.738 WGHTD AVE 0 321 1.143 0 802 1 856 1.747 1 551 1 856 1.747 1 551

JACKSON LOW 0 037 0077 0 020 0 098 0 092 0 083 0 098 0 092 0 083 HIGH 0 086 0 204 0 074 0 279 0 257 0 223 0 279 0 257 0 223 WGHTD AVE 0 059 0 000 0 000 0.171 0.159 0.141 0.171 0.159 0.141

JEFFERSON LOW 0 036 0 078 0 023 0.102 0 095 0 084 0.102 0 095 0 084 HIGH 0 042 0 094 0 029 0.123 0.115 0.101 0.123 0.115 0.101 WGHTD AVE 0 000 0 000 0 000 0 000 0 000 0 000 0 000 0 000 0 000

LAFAYETTE LOW 0 040 0087 0 029 0.117 0.108 0 095 0.117 0.108 0 095 HIGH 0 052 0.119 0 047 0.168 0.156 0.137 0.168 0.156 0.137 WGHTD AVE 0 052 0.119 0 000 0.168 0.156 0.137 0.168 0.156 0.137

LAKE LOW 0 070 0.142 0 049 0.194 0.181 0.162 0.194 0.181 0.162 HIGH 0.108 0 237 0 099 0 341 0 315 0 276 0 341 0 315 0 276 WGHTD AVE 0 077 0.157 0 055 0 216 0 202 0.180 0 216 0 202 0.180

LEE LOW 0.168 0384 0.174 0 567 0 524 0.459 0 567 0 524 0.459 HIGH 0 699 3 039 2 822 5 284 5 059 4 615 5 284 5 059 4 615 WGHTD AVE 0 370 1.452 1 090 2 351 2 228 2 001 2 351 2 228 2 001

LEON LOW 0 042 0093 0 029 0.122 0.113 0 099 0.122 0.113 0 099 HIGH 0 059 0.141 0 052 0.194 0.179 0.154 0.194 0.179 0.154 WGHTD AVE 0 047 0.106 0 034 0.141 0.130 0.113 0.141 0.130 0.113

LEVY LOW 0 068 0.147 0 060 0 209 0.194 0.170 0 209 0.194 0.170 HIGH 0 286 1.183 1 042 2 061 1 970 1.789 2 061 1 970 1.789 WGHTD AVE 0 286 1.183 1 042 2 061 1 970 1.789 2 061 1 970 1.789

LIBERTY LOW 0 070 0.166 0 062 0 230 0 212 0.184 0 230 0 212 0.184 HIGH 0 079 0.190 0 073 0 265 0 244 0 212 0 265 0 244 0 212 WGHTD AVE 0 000 0 000 0 000 0 000 0 000 0 000 0 000 0 000 0 000

MADISON LOW 0 029 0060 0 017 0 077 0 072 0 064 0 077 0 072 0 064 HIGH 0 034 0 072 0 022 0 095 0 089 0 078 0 095 0 089 0 078 WGHTD AVE 0 029 0 061 0 017 0 078 0 073 0 065 0 078 0 073 0 065

MANATEE LOW 0.108 0233 0.109 0 344 0 322 0 286 0 344 0 322 0 286 HIGH 0 396 1 549 1.450 2.710 2 582 2 336 2.710 2 582 2 336 WGHTD AVE 0 319 1.137 0 982 1 964 1 862 1 671 1 964 1 862 1 671

MARION LOW 0 068 0.145 0 056 0 209 0.195 0.172 0 209 0.195 0.172 HIGH 0.101 0 228 0.106 0 336 0 310 0 269 0 336 0 310 0 269 WGHTD AVE 0 080 0.175 0 078 0 257 0 239 0 210 0 257 0 239 0 210

MART N LOW 0.191 0.420 0.197 0 625 0 577 0 506 0 625 0 577 0 506 HIGH 0 630 2 388 2 272 4 279 4 070 3 665 4 279 4 070 3 665 WGHTD AVE 0.496 1.791 1 653 3.148 2 984 2 675 3.148 2 984 2 675

Form A-6: Output Ranges

MIAMI-DADE LOW 0 327 0 815 0 540 1 364 1 265 1.104 1 364 1 265 1.104 HIGH 1 574 8.165 9 332 14 643 14.173 13.157 14 643 14.173 13.157 WGHTD AVE 0 919 3 861 4.464 7.128 6 842 6 265 7.128 6 842 6 265

MONROE LOW 0 819 3529 3.181 6 216 5 934 5 373 6 216 5 934 5 373 HIGH 1 245 6 802 6 396 11 931 11 519 10 635 11 931 11 519 10 635 WGHTD AVE 1 045 5 504 4 861 9 567 9.198 8.429 9 567 9.198 8.429

NASSAU LOW 0 043 0093 0 037 0.131 0.122 0.108 0.131 0.122 0.108 HIGH 0.108 0 389 0 304 0 645 0 613 0 552 0 645 0 613 0 552 WGHTD AVE 0.108 0 388 0 302 0 643 0 611 0 550 0 643 0 611 0 550

OKALOOSA LOW 0.101 0241 0 096 0 338 0 312 0 273 0 338 0 312 0 273 HIGH 0.478 2.181 1.760 3.744 3 580 3 249 3.744 3 580 3 249 WGHTD AVE 0 320 1 246 0 879 2 086 1 974 1.759 2 086 1 974 1.759

OKEECHOBEE LOW 0.158 0 338 0.142 0.486 0.449 0 396 0.486 0.449 0 396 HIGH 0.174 0 375 0.163 0 544 0 502 0.442 0 544 0 502 0.442 WGHTD AVE 0.174 0 375 0.163 0 544 0 502 0.442 0 544 0 502 0.442

ORANGE LOW 0 071 0.142 0 047 0.191 0.179 0.160 0.191 0.179 0.160 HIGH 0.114 0 251 0.106 0 363 0 336 0 293 0 363 0 336 0 293 WGHTD AVE 0 088 0.181 0 066 0 250 0 233 0 207 0 250 0 233 0 207

OSCEOLA LOW 0 084 0.168 0 059 0 229 0 214 0.192 0 229 0 214 0.192 HIGH 0.115 0 240 0 095 0 339 0 315 0 279 0 339 0 315 0 279 WGHTD AVE 0 093 0.191 0 070 0 264 0 246 0 219 0 264 0 246 0 219

PALM BEACH LOW 0 211 0.472 0 233 0.715 0 660 0 577 0.715 0 660 0 577 HIGH 1 029 4.450 4.788 8 039 7.709 7 042 8 039 7.709 7 042 WGHTD AVE 0 536 2 007 1 806 3.488 3 312 2 984 3.488 3 312 2 984

PASCO LOW 0 089 0.186 0 076 0 264 0 247 0 219 0 264 0 247 0 219 HIGH 0 287 1 036 0 923 1 839 1.748 1 572 1 839 1.748 1 572 WGHTD AVE 0.155 0 559 0.400 0 889 0 840 0.751 0 889 0 840 0.751

P NELLAS LOW 0.127 0311 0.183 0.493 0.459 0.400 0.493 0.459 0.400 HIGH 0.462 2 012 1 982 3 563 3.415 3.117 3 563 3.415 3.117 WGHTD AVE 0 264 1 045 0 863 1.756 1 669 1 502 1.756 1 669 1 502

POLK LOW 0 076 0.151 0 051 0 204 0.192 0.172 0 204 0.192 0.172 HIGH 0 099 0.199 0 071 0 273 0 255 0 228 0 273 0 255 0 228 WGHTD AVE 0 084 0.168 0 058 0 229 0 215 0.193 0 229 0 215 0.193

PUTNAM LOW 0 058 0.121 0 044 0.167 0.156 0.138 0.167 0.156 0.138 HIGH 0 083 0.180 0 071 0 254 0 235 0 205 0 254 0 235 0 205 WGHTD AVE 0 069 0.148 0 055 0 206 0.191 0.168 0 206 0.191 0.168

SAINT JOHNS LOW 0 052 0.112 0 044 0.159 0.148 0.130 0.159 0.148 0.130 HIGH 0.195 0.711 0 579 1 224 1.163 1 048 1 224 1.163 1 048 WGHTD AVE 0.161 0 563 0.448 0 957 0 907 0 814 0 957 0 907 0 814

SAINT LUC E LOW 0.169 0 370 0.170 0 548 0 507 0.444 0 548 0 507 0.444 HIGH 0 563 2.125 1 969 3.791 3 603 3 241 3.791 3 603 3 241 WGHTD AVE 0 340 1.180 0 937 1 951 1 838 1 634 1 951 1 838 1 634

Form A-6: Output Ranges

SANTA ROSA LOW 0.117 0 282 0.125 0.412 0 379 0 328 0.412 0 379 0 328 HIGH 0.435 1 872 1.496 3 214 3 064 2.769 3 214 3 064 2.769 WGHTD AVE 0 351 1 366 1 009 2 298 2.174 1 939 2 298 2.174 1 939

SARASOTA LOW 0.119 0273 0.146 0.414 0 388 0 346 0.414 0 388 0 346 HIGH 0.474 1 920 1 825 3.406 3 252 2 948 3.406 3 252 2 948 WGHTD AVE 0 316 1 241 0 972 2 052 1 945 1.744 2 052 1 945 1.744

SEM NOLE LOW 0 076 0.153 0 052 0 207 0.194 0.173 0 207 0.194 0.173 HIGH 0 098 0 210 0 084 0 298 0 276 0 243 0 298 0 276 0 243 WGHTD AVE 0 079 0.160 0 056 0 219 0 204 0.181 0 219 0 204 0.181

SUMTER LOW 0 074 0.155 0 062 0 220 0 205 0.181 0 220 0 205 0.181 HIGH 0.105 0 235 0.109 0 347 0 323 0 285 0 347 0 323 0 285 WGHTD AVE 0 083 0.177 0 076 0 258 0 241 0 212 0 258 0 241 0 212

SUWANEE LOW 0 038 0082 0 027 0.110 0.102 0 090 0.110 0.102 0 090 HIGH 0 056 0.127 0 053 0.182 0.169 0.149 0.182 0.169 0.149 WGHTD AVE 0 055 0.105 0 048 0.163 0.152 0.134 0.163 0.152 0.134

TAYLOR LOW 0 048 0.110 0 039 0.151 0.140 0.122 0.151 0.140 0.122 HIGH 0.173 0 651 0.462 1 067 1 010 0 903 1 067 1 010 0 903 WGHTD AVE 0.173 0 651 0.462 1 067 1 010 0 903 1 067 1 010 0 903

UNION LOW 0 063 0.138 0 056 0.195 0.180 0.159 0.195 0.180 0.159 HIGH 0 063 0.139 0 057 0.198 0.183 0.161 0.198 0.183 0.161 WGHTD AVE 0 000 0 000 0 000 0 000 0 000 0 000 0 000 0 000 0 000

VOLUSIA LOW 0 086 0.186 0 072 0 262 0 243 0 214 0 262 0 243 0 214 HIGH 0 274 0 978 0 810 1 692 1 602 1.434 1 692 1 602 1.434 WGHTD AVE 0 203 0.713 0 507 1.180 1.114 0 992 1.180 1.114 0 992

WAKULLA LOW 0 081 0213 0 097 0 314 0 290 0 249 0 314 0 290 0 249 HIGH 0.196 0.778 0 533 1 276 1 211 1 086 1 276 1 211 1 086 WGHTD AVE 0.122 0 384 0 299 0 627 0 589 0 521 0 627 0 589 0 521

WALTON LOW 0.113 0287 0.131 0.419 0 388 0 339 0.419 0 388 0 339 HIGH 0 349 1.415 1 055 2 358 2 237 2 005 2 358 2 237 2 005 WGHTD AVE 0 340 1 372 1 000 2 280 2.161 1 933 2 280 2.161 1 933

WASH NGTON LOW 0 094 0 227 0 089 0 318 0 293 0 254 0 318 0 293 0 254 HIGH 0.138 0 375 0 205 0 571 0 531 0.465 0 571 0 531 0.465 WGHTD AVE 0.126 0 335 0.167 0 502 0.465 0.406 0 502 0.465 0.406

STATEWIDE LOW 0 029 0060 0 017 0 077 0 072 0 064 0 077 0 072 0 064 HIGH 1 574 8.165 9 332 14 643 14.173 13.157 14 643 14.173 13.157 WGHTD AVE 0 299 1.168 0 927 1 942 1 844 1 660 1 942 1 844 1 660

Form A-6: Output Ranges

ALACHUA LOW 0.040 0.081 0.019 0.102 0.096 0.085 0.102 0.096 0.085 HIGH 0.062 0.130 0.034 0.166 0.153 0.135 0.166 0.153 0.135 WGHTD AVE 0.047 0.095 0.022 0.118 0.110 0.098 0.118 0.110 0.098

BAKER LOW 0.036 0.072 0.016 0.089 0.083 0.073 0.089 0.083 0.073 HIGH 0.041 0.082 0.019 0.102 0.095 0.084 0.102 0.095 0.084 WGHTD AVE 0.041 0.000 0.000 0.102 0.095 0.084 0.102 0.095 0.084

BAY LOW 0.074 0.169 0.045 0.216 0.199 0.173 0.216 0.199 0.173 HIGH 0.349 1.393 0.935 2.271 2.150 1.916 2.271 2.150 1.916 WGHTD AVE 0.242 0.838 0.472 1.294 1.214 1.068 1.294 1.214 1.068

BRADFORD LOW 0.049 0.101 0.025 0.128 0.119 0.104 0.128 0.119 0.104 HIGH 0.052 0.108 0.027 0.137 0.127 0.112 0.137 0.127 0.112 WGHTD AVE 0.050 0.107 0.027 0.133 0.124 0.109 0.133 0.124 0.109

BREVARD LOW 0.067 0.134 0.033 0.171 0.160 0.141 0.171 0.160 0.141 HIGH 0.401 1.469 1.186 2.512 2.379 2.124 2.512 2.379 2.124 WGHTD AVE 0.226 0.674 0.411 1.072 1.003 0.880 1.072 1.003 0.880

BROWARD LOW 0.197 0.427 0.163 0.600 0.554 0.483 0.600 0.554 0.483 HIGH 0.920 3.739 3.821 6.703 6.412 5.818 6.703 6.412 5.818 WGHTD AVE 0.485 1.642 1.233 2.732 2.587 2.318 2.732 2.587 2.318

CALHOUN LOW 0.062 0.137 0.030 0.167 0.155 0.135 0.167 0.155 0.135 HIGH 0.095 0.234 0.075 0.312 0.287 0.247 0.312 0.287 0.247 WGHTD AVE 0.063 0.137 0.000 0.172 0.159 0.139 0.172 0.159 0.139

CHARLOTTE LOW 0.151 0.342 0.128 0.475 0.440 0.381 0.475 0.440 0.381 HIGH 0.373 1.252 0.942 2.059 1.941 1.723 2.059 1.941 1.723 WGHTD AVE 0.270 0.796 0.533 1.282 1.202 1.058 1.282 1.202 1.058

CITRUS LOW 0.057 0.117 0.033 0.153 0.142 0.124 0.153 0.142 0.124 HIGH 0.124 0.329 0.178 0.505 0.471 0.411 0.505 0.471 0.411 WGHTD AVE 0.086 0.202 0.087 0.291 0.271 0.236 0.291 0.271 0.236

CLAY LOW 0.033 0.065 0.015 0.082 0.077 0.068 0.082 0.077 0.068 HIGH 0.050 0.102 0.026 0.130 0.121 0.106 0.130 0.121 0.106 WGHTD AVE 0.034 0.068 0.016 0.085 0.080 0.070 0.085 0.080 0.070

COLLIER LOW 0.161 0.341 0.094 0.441 0.407 0.356 0.441 0.407 0.356 HIGH 0.621 2.396 1.892 4.036 3.828 3.425 4.036 3.828 3.425 WGHTD AVE 0.459 1.733 1.260 2.796 2.647 2.367 2.796 2.647 2.367

COLUMBIA LOW 0.034 0.067 0.015 0.083 0.078 0.069 0.083 0.078 0.069 HIGH 0.044 0.088 0.022 0.110 0.102 0.090 0.110 0.102 0.090 WGHTD AVE 0.038 0.077 0.018 0.096 0.089 0.079 0.096 0.089 0.079

DESOTO LOW 0.090 0.178 0.047 0.228 0.214 0.190 0.228 0.214 0.190 HIGH 0.106 0.219 0.067 0.291 0.270 0.237 0.291 0.270 0.237 WGHTD AVE 0.097 0.194 0.057 0.254 0.236 0.209 0.254 0.236 0.209

DIXIE LOW 0.056 0.128 0.046 0.174 0.163 0.144 0.174 0.163 0.144 HIGH 0.127 0.418 0.252 0.645 0.608 0.541 0.645 0.608 0.541 WGHTD AVE 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

Form A-6: Output Ranges

DUVAL LOW 0.027 0.053 0.012 0.065 0.061 0.054 0.065 0.061 0.054 HIGH 0.098 0.292 0.179 0.453 0.426 0.375 0.453 0.426 0.375 WGHTD AVE 0.048 0.130 0.047 0.177 0.166 0.146 0.177 0.166 0.146

ESCAMBIA LOW 0.075 0.162 0.038 0.202 0.187 0.163 0.202 0.187 0.163 HIGH 0.310 1.097 0.711 1.764 1.658 1.458 1.764 1.658 1.458 WGHTD AVE 0.208 0.628 0.326 0.954 0.887 0.770 0.954 0.887 0.770

FLAGLER LOW 0.066 0.137 0.035 0.175 0.162 0.142 0.175 0.162 0.142 HIGH 0.204 0.690 0.494 1.149 1.084 0.964 1.149 1.084 0.964 WGHTD AVE 0.177 0.586 0.380 0.946 0.890 0.786 0.946 0.890 0.786

FRANKL N LOW 0.161 0.534 0.283 0.820 0.768 0.672 0.820 0.768 0.672 HIGH 0.299 1.252 0.819 2.022 1.920 1.719 2.022 1.920 1.719 WGHTD AVE 0.220 1.115 0.624 1.653 1.567 1.397 1.653 1.567 1.397

GADSEN LOW 0.036 0.076 0.015 0.091 0.085 0.075 0.091 0.085 0.075 HIGH 0.050 0.108 0.023 0.130 0.121 0.107 0.130 0.121 0.107 WGHTD AVE 0.043 0.093 0.019 0.113 0.105 0.092 0.113 0.105 0.092

GILCHRIST LOW 0.045 0.097 0.029 0.127 0.118 0.104 0.127 0.118 0.104 HIGH 0.049 0.102 0.030 0.134 0.125 0.109 0.134 0.125 0.109 WGHTD AVE 0.045 0.097 0.029 0.127 0.118 0.104 0.127 0.118 0.104

GLADES LOW 0.136 0.279 0.072 0.356 0.330 0.291 0.356 0.330 0.291 HIGH 0.136 0.279 0.072 0.356 0.330 0.291 0.356 0.330 0.291 WGHTD AVE 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

GULF LOW 0.089 0.216 0.072 0.290 0.268 0.232 0.290 0.268 0.232 HIGH 0.299 1.168 0.731 1.861 1.759 1.561 1.861 1.759 1.561 WGHTD AVE 0.089 1.162 0.000 1.644 1.552 1.377 1.644 1.552 1.377

HAMILTON LOW 0.022 0.043 0.008 0.052 0.049 0.043 0.052 0.049 0.043 HIGH 0.032 0.063 0.014 0.079 0.074 0.066 0.079 0.074 0.066 WGHTD AVE 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

HARDEE LOW 0.071 0.136 0.031 0.170 0.160 0.142 0.170 0.160 0.142 HIGH 0.078 0.150 0.037 0.191 0.179 0.158 0.191 0.179 0.158 WGHTD AVE 0.072 0.000 0.000 0.173 0.163 0.146 0.173 0.163 0.146

HENDRY LOW 0.129 0.263 0.068 0.336 0.312 0.275 0.336 0.312 0.275 HIGH 0.156 0.320 0.086 0.411 0.381 0.336 0.411 0.381 0.336 WGHTD AVE 0.148 0.305 0.079 0.390 0.361 0.319 0.390 0.361 0.319

HERNANDO LOW 0.072 0.150 0.046 0.201 0.186 0.162 0.201 0.186 0.162 HIGH 0.173 0.531 0.368 0.858 0.807 0.714 0.858 0.807 0.714 WGHTD AVE 0.107 0.278 0.147 0.419 0.392 0.344 0.419 0.392 0.344

HIGHLANDS LOW 0.077 0.148 0.033 0.184 0.173 0.155 0.184 0.173 0.155 HIGH 0.105 0.208 0.049 0.261 0.244 0.217 0.261 0.244 0.217 WGHTD AVE 0.087 0.163 0.037 0.206 0.193 0.173 0.206 0.193 0.173

HILLSBOROUGH LOW 0.064 0.123 0.030 0.157 0.147 0.131 0.157 0.147 0.131 HIGH 0.224 0.683 0.487 1.104 1.036 0.915 1.104 1.036 0.915 WGHTD AVE 0.100 0.233 0.102 0.332 0.310 0.273 0.332 0.310 0.273

Form A-6: Output Ranges

HOLMES LOW 0.063 0.137 0.030 0.169 0.157 0.138 0.169 0.157 0.138 HIGH 0.070 0.156 0.036 0.194 0.180 0.157 0.194 0.180 0.157 WGHTD AVE 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

INDIAN RIVER LOW 0.108 0.224 0.063 0.292 0.271 0.238 0.292 0.271 0.238 HIGH 0.363 1.148 0.822 1.886 1.769 1.557 1.886 1.769 1.557 WGHTD AVE 0.287 0.876 0.541 1.378 1.288 1.129 1.378 1.288 1.129

JACKSON LOW 0.028 0.055 0.010 0.066 0.063 0.056 0.066 0.063 0.056 HIGH 0.064 0.141 0.030 0.172 0.159 0.139 0.172 0.159 0.139 WGHTD AVE 0.045 0.096 0.018 0.114 0.107 0.096 0.114 0.107 0.096

JEFFERSON LOW 0.027 0.055 0.011 0.067 0.063 0.056 0.067 0.063 0.056 HIGH 0.032 0.066 0.013 0.080 0.075 0.066 0.080 0.075 0.066 WGHTD AVE 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

LAFAYETTE LOW 0.030 0.061 0.012 0.074 0.070 0.061 0.074 0.070 0.061 HIGH 0.039 0.082 0.020 0.103 0.096 0.084 0.103 0.096 0.084 WGHTD AVE 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

LAKE LOW 0.053 0.101 0.022 0.126 0.118 0.106 0.126 0.118 0.106 HIGH 0.080 0.165 0.042 0.210 0.195 0.172 0.210 0.195 0.172 WGHTD AVE 0.058 0.112 0.025 0.139 0.131 0.117 0.139 0.131 0.117

LEE LOW 0.125 0.263 0.071 0.343 0.317 0.279 0.343 0.317 0.279 HIGH 0.508 1.904 1.529 3.210 3.049 2.741 3.210 3.049 2.741 WGHTD AVE 0.304 0.963 0.631 1.527 1.438 1.276 1.527 1.438 1.276

LEON LOW 0.031 0.065 0.012 0.078 0.073 0.064 0.078 0.073 0.064 HIGH 0.044 0.097 0.021 0.119 0.110 0.095 0.119 0.110 0.095 WGHTD AVE 0.034 0.072 0.014 0.086 0.080 0.070 0.086 0.080 0.070

LEVY LOW 0.051 0.102 0.025 0.129 0.120 0.105 0.129 0.120 0.105 HIGH 0.207 0.738 0.552 1.218 1.153 1.026 1.218 1.153 1.026 WGHTD AVE 0.098 0.411 0.088 0.507 0.478 0.422 0.507 0.478 0.422

LIBERTY LOW 0.052 0.114 0.025 0.141 0.130 0.114 0.141 0.130 0.114 HIGH 0.059 0.131 0.030 0.161 0.149 0.130 0.161 0.149 0.130 WGHTD AVE 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

MADISON LOW 0.022 0.043 0.008 0.051 0.048 0.043 0.051 0.048 0.043 HIGH 0.026 0.051 0.010 0.062 0.058 0.051 0.062 0.058 0.051 WGHTD AVE 0.026 0.051 0.010 0.062 0.058 0.051 0.062 0.058 0.051

MANATEE LOW 0.080 0.161 0.047 0.212 0.199 0.176 0.212 0.199 0.176 HIGH 0.288 0.976 0.772 1.630 1.540 1.374 1.630 1.540 1.374 WGHTD AVE 0.223 0.676 0.461 1.086 1.020 0.900 1.086 1.020 0.900

MARION LOW 0.050 0.101 0.024 0.131 0.122 0.107 0.131 0.122 0.107 HIGH 0.075 0.157 0.045 0.203 0.187 0.163 0.203 0.187 0.163 WGHTD AVE 0.057 0.117 0.033 0.154 0.143 0.125 0.154 0.143 0.125

MART N LOW 0.141 0.290 0.079 0.375 0.348 0.305 0.375 0.348 0.305 HIGH 0.457 1.492 1.160 2.521 2.372 2.097 2.521 2.372 2.097 WGHTD AVE 0.323 0.953 0.641 1.538 1.440 1.266 1.538 1.440 1.266

Form A-6: Output Ranges

MIAMI-DADE LOW 0.238 0.537 0.217 0.775 0.713 0.617 0.775 0.713 0.617 HIGH 1.151 5.226 5.592 9.366 9.006 8.252 9.366 9.006 8.252 WGHTD AVE 0.736 3.273 2.996 5.596 5.344 4.840 5.596 5.344 4.840

MONROE LOW 0.594 2.206 1.678 3.713 3.510 3.123 3.713 3.510 3.123 HIGH 0.912 4.318 3.717 7.488 7.175 6.524 7.488 7.175 6.524 WGHTD AVE 0.718 2.858 2.317 4.903 4.657 4.171 4.903 4.657 4.171

NASSAU LOW 0.032 0.065 0.016 0.082 0.077 0.067 0.082 0.077 0.067 HIGH 0.080 0.254 0.164 0.397 0.374 0.333 0.397 0.374 0.333 WGHTD AVE 0.080 0.254 0.164 0.396 0.374 0.332 0.396 0.374 0.332

OKALOOSA LOW 0.075 0.165 0.039 0.206 0.191 0.168 0.206 0.191 0.168 HIGH 0.343 1.316 0.891 2.160 2.043 1.813 2.160 2.043 1.813 WGHTD AVE 0.247 0.793 0.449 1.253 1.171 1.019 1.253 1.171 1.019

OKEECHOBEE LOW 0.118 0.236 0.059 0.300 0.279 0.247 0.300 0.279 0.247 HIGH 0.129 0.261 0.067 0.333 0.310 0.274 0.333 0.310 0.274 WGHTD AVE 0.129 0.260 0.067 0.332 0.309 0.273 0.332 0.309 0.273

ORANGE LOW 0.054 0.101 0.021 0.125 0.118 0.106 0.125 0.118 0.106 HIGH 0.085 0.175 0.044 0.222 0.206 0.180 0.222 0.206 0.180 WGHTD AVE 0.066 0.128 0.028 0.159 0.149 0.133 0.159 0.149 0.133

OSCEOLA LOW 0.063 0.120 0.026 0.148 0.140 0.125 0.148 0.140 0.125 HIGH 0.086 0.168 0.040 0.212 0.198 0.176 0.212 0.198 0.176 WGHTD AVE 0.069 0.133 0.030 0.165 0.155 0.138 0.165 0.155 0.138

PALM BEACH LOW 0.156 0.323 0.094 0.425 0.392 0.343 0.425 0.392 0.343 HIGH 0.750 2.832 2.699 4.954 4.713 4.245 4.954 4.713 4.245 WGHTD AVE 0.439 1.461 1.096 2.407 2.271 2.020 2.407 2.271 2.020

PASCO LOW 0.066 0.130 0.033 0.166 0.156 0.138 0.166 0.156 0.138 HIGH 0.208 0.636 0.451 1.054 0.990 0.871 1.054 0.990 0.871 WGHTD AVE 0.161 0.483 0.294 0.760 0.712 0.624 0.760 0.712 0.624

P NELLAS LOW 0.093 0.206 0.074 0.284 0.262 0.225 0.284 0.262 0.225 HIGH 0.336 1.264 1.092 2.157 2.049 1.841 2.157 2.049 1.841 WGHTD AVE 0.220 0.715 0.501 1.172 1.104 0.978 1.172 1.104 0.978

POLK LOW 0.057 0.108 0.023 0.133 0.126 0.113 0.133 0.126 0.113 HIGH 0.074 0.141 0.031 0.176 0.165 0.148 0.176 0.165 0.148 WGHTD AVE 0.064 0.120 0.026 0.150 0.141 0.127 0.150 0.141 0.127

PUTNAM LOW 0.044 0.086 0.019 0.106 0.100 0.089 0.106 0.100 0.089 HIGH 0.062 0.126 0.030 0.158 0.146 0.129 0.158 0.146 0.129 WGHTD AVE 0.053 0.101 0.023 0.129 0.120 0.106 0.129 0.120 0.106

SAINT JOHNS LOW 0.039 0.079 0.019 0.099 0.093 0.082 0.099 0.093 0.082 HIGH 0.143 0.444 0.289 0.720 0.678 0.600 0.720 0.678 0.600 WGHTD AVE 0.124 0.370 0.229 0.588 0.552 0.487 0.588 0.552 0.487

SAINT LUC E LOW 0.125 0.255 0.070 0.331 0.307 0.269 0.331 0.307 0.269 HIGH 0.409 1.330 1.008 2.226 2.094 1.850 2.226 2.094 1.850 WGHTD AVE 0.297 0.932 0.572 1.468 1.375 1.208 1.468 1.375 1.208

Form A-6: Output Ranges

LOSS COSTS PER $1,000 PERSONAL RESIDENTIAL - Condo Owners -- MASONRY $0 $0 $0 DEDUCTIBLE $500 $1,000 $2,500 1% 2% 5% DEDUCTIBLE DEDUCTIBLE ADDITIONAL DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE DEDUCTIBLE COUNTY LOSS COSTS STRUCTURE CONTENTS LIVING EXPENSE TOTAL* TOTAL* TOTAL* TOTAL* TOTAL* TOTAL* SANTA ROSA LOW 0.086 0.192 0.049 0.245 0.225 0.195 0.245 0.225 0.195 HIGH 0.313 1.138 0.750 1.843 1.738 1.537 1.843 1.738 1.537 WGHTD AVE 0.273 0.925 0.573 1.464 1.373 1.205 1.464 1.373 1.205

SARASOTA LOW 0.089 0.186 0.066 0.255 0.239 0.212 0.255 0.239 0.212 HIGH 0.345 1.205 0.964 2.035 1.924 1.715 2.035 1.924 1.715 WGHTD AVE 0.252 0.837 0.554 1.318 1.239 1.096 1.318 1.239 1.096

SEM NOLE LOW 0.057 0.109 0.023 0.134 0.127 0.113 0.134 0.127 0.113 HIGH 0.073 0.147 0.035 0.186 0.173 0.153 0.186 0.173 0.153 WGHTD AVE 0.059 0.113 0.024 0.140 0.131 0.117 0.140 0.131 0.117

SUMTER LOW 0.055 0.109 0.026 0.138 0.129 0.114 0.138 0.129 0.114 HIGH 0.078 0.162 0.046 0.211 0.197 0.173 0.211 0.197 0.173 WGHTD AVE 0.061 0.131 0.036 0.166 0.155 0.137 0.166 0.155 0.137

SUWANEE LOW 0.029 0.058 0.012 0.071 0.066 0.058 0.071 0.066 0.058 HIGH 0.042 0.088 0.023 0.112 0.104 0.091 0.112 0.104 0.091 WGHTD AVE 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

TAYLOR LOW 0.036 0.076 0.017 0.094 0.087 0.076 0.094 0.087 0.076 HIGH 0.126 0.413 0.232 0.634 0.595 0.525 0.634 0.595 0.525 WGHTD AVE 0.126 0.413 0.232 0.634 0.595 0.525 0.634 0.595 0.525

UNION LOW 0.047 0.096 0.024 0.121 0.113 0.100 0.121 0.113 0.100 HIGH 0.047 0.097 0.024 0.123 0.114 0.101 0.123 0.114 0.101 WGHTD AVE 0.047 0.097 0.024 0.123 0.114 0.101 0.123 0.114 0.101

VOLUSIA LOW 0.064 0.130 0.031 0.164 0.153 0.135 0.164 0.153 0.135 HIGH 0.201 0.620 0.404 0.998 0.936 0.824 0.998 0.936 0.824 WGHTD AVE 0.170 0.509 0.304 0.802 0.751 0.659 0.802 0.751 0.659

WAKULLA LOW 0.060 0.142 0.039 0.184 0.168 0.143 0.184 0.168 0.143 HIGH 0.142 0.481 0.262 0.738 0.693 0.609 0.738 0.693 0.609 WGHTD AVE 0.142 0.481 0.262 0.738 0.693 0.609 0.738 0.693 0.609

WALTON LOW 0.083 0.193 0.055 0.250 0.231 0.203 0.250 0.231 0.203 HIGH 0.252 0.887 0.533 1.383 1.300 1.146 1.383 1.300 1.146 WGHTD AVE 0.244 0.817 0.463 1.269 1.188 1.038 1.269 1.188 1.038

WASH NGTON LOW 0.070 0.156 0.036 0.194 0.179 0.156 0.194 0.179 0.156 HIGH 0.101 0.248 0.091 0.337 0.313 0.272 0.337 0.313 0.272 WGHTD AVE 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

STATEWIDE LOW 0.022 0.043 0.008 0.051 0.048 0.043 0.051 0.048 0.043 HIGH 1.151 5.226 5.592 9.366 9.006 8.252 9.366 9.006 8.252 WGHTD AVE 0.383 1.492 1.052 2.372 2.246 2.010 2.372 2.246 2.010

Form A-7: Percentage Change in Output Ranges

Form A-7: Percentage Change In Output Ranges

A. Provide the percentage change in the weighted average loss costs using the 2002 Florida Hurricane Catastrophe Fund’s aggregate personal residential exposure data found in the file named “hlpm2002.exe”, as of August 1, 2003, from the output ranges from the prior year submission for the following: • statewide (overall percentage change), • by region, as defined in Figure 4 – North, Central and South, • by coastal and inland counties, as defined in Figure 5.

The percentage change in the weighted average loss costs is provided in Form A-7 on the page 187.

Form A-7 is provided on CD-ROM in both an Excel and PDF format.

Form A-7: Percentage Change in Output Ranges

Figure 40: State of Florida by North/Central/South Regions

North

Central

South

Figure 41: State of Florida by Coastal/Inland Counties

Inland

Coastal

Form A-7: Percentage Change in Output Ranges

Form A-7: Percentage Change In Output Ranges

$0 Deductible

Additional $500 $1,000 $2,500 1% 2% 5% Appurtenant Living Deductible Deductible Deductible Deductible Deductible Deductible Structure Contents Structure Expense Total Total Total Total Total Total Frame Coastal 17.0% -27.2% 17.6% 39.3% 6.5% 6.6% 6.6% 6.6% 6.6% 6.3% Owners Inland 17.0% -19.5% 17.0% 36.2% 8.1% 8.0% 7.2% 8.0% 7.6% 5.2% North 13.9% -25.0% 14.0% 25.8% 5.5% 5.5% 5.2% 5.5% 5.3% 4.6% Central 16.7% -27.3% 16.8% 37.1% 6.2% 6.3% 6.0% 6.3% 6.2% 4.9% South 19.1% -27.1% 20.3% 46.5% 7.6% 7.8% 8.1% 7.8% 8.0% 8.0% Statewide 17.0% -26.6% 17.4% 39.6% 6.6% 6.7% 6.7% 6.7% 6.7% 6.2% Masonry Coastal 20.9% -26.5% 20.6% 52.0% 9.5% 9.9% 10.3% 9.9% 10.2% 10.1% Owners Inland 18.2% -17.7% 18.2% 40.7% 8.0% 7.9% 6.5% 7.9% 7.0% 4.1% North 13.1% -24.6% 13.2% 23.6% 4.7% 4.6% 4.3% 4.6% 4.4% 3.6% Central 17.4% -26.0% 17.5% 38.0% 6.9% 6.9% 6.4% 6.9% 6.6% 5.2% South 21.9% -26.3% 21.8% 55.4% 10.3% 10.7% 11.3% 10.7% 11.1% 11.3% Statewide 20.8% -26.2% 20.3% 51.6% 9.5% 9.7% 10.1% 9.7% 10.0% 9.8% Mobile Coastal 17.6% -27.2% 17.6% 41.8% 6.3% 6.4% 6.4% 6.3% 6.4% 6.4% Homes Inland 16.9% -24.5% 16.8% 40.6% 9.6% 10.1% 10.0% 9.6% 10.1% 10.0% North 13.5% -25.3% 13.5% 25.7% 5.2% 5.1% 4.6% 5.2% 5.1% 4.6% Central 17.0% -26.8% 17.0% 41.4% 7.4% 7.7% 7.5% 7.4% 7.7% 7.5% South 18.9% -27.0% 19.5% 43.5% 6.6% 6.8% 6.9% 6.6% 6.8% 6.9% Statewide 17.5% -26.8% 17.5% 41.8% 6.9% 7.1% 7.0% 6.9% 7.1% 7.0% Frame Coastal N/A -0.8% N/A 36.0% 9.0% 9.5% 10.5% 8.7% 9.0% 9.7% Renters Inland N/A 2.2% N/A 38.2% 17.5% 17.9% 18.9% 17.1% 17.5% 18.1% North N/A 0.7% N/A 24.7% 8.1% 8.4% 9.2% 7.9% 8.1% 8.5% Central N/A -2.9% N/A 35.3% 9.5% 10.0% 11.1% 9.2% 9.5% 10.2% South N/A 1.3% N/A 52.1% 11.8% 12.4% 13.6% 11.4% 11.8% 12.6% Statewide N/A -0.5% N/A 36.2% 9.8% 10.3% 11.2% 9.5% 9.8% 10.4% Masonry Coastal N/A 1.0% N/A 54.7% 12.3% 12.9% 14.3% 11.8% 12.3% 13.2% Renters Inland N/A 1.9% N/A 38.6% 13.5% 13.9% 14.8% 13.1% 13.5% 14.2% North N/A 0.0% N/A 17.9% 6.1% 6.4% 7.1% 6.0% 6.1% 6.5% Central N/A -2.9% N/A 35.5% 8.2% 8.6% 9.6% 7.9% 8.2% 8.8% South N/A 2.1% N/A 60.2% 13.9% 14.7% 16.2% 13.4% 13.9% 15.0% Statewide N/A 1.0% N/A 54.5% 12.3% 12.9% 14.4% 11.8% 12.3% 13.2% Frame Coastal 19.6% 0.6% N/A 46.7% 10.3% 10.5% 11.1% 10.3% 10.5% 11.1% Condos Inland 18.8% 1.5% N/A 39.2% 10.6% 10.6% 11.4% 10.6% 10.6% 11.4% North 14.0% 2.4% N/A 24.4% 7.7% 7.7% 8.0% 7.7% 7.7% 8.0% Central 17.1% -2.9% N/A 37.1% 6.8% 6.8% 7.2% 6.8% 6.8% 7.2% South 21.6% 1.5% N/A 52.8% 12.0% 12.2% 13.1% 12.0% 12.2% 13.1% Statewide 19.6% 0.7% N/A 46.6% 10.3% 10.5% 11.2% 10.3% 10.5% 11.2% Masonry Coastal 22.0% 2.0% N/A 58.4% 13.2% 13.5% 14.5% 13.2% 13.5% 14.5% Condos Inland 19.1% 1.5% N/A 39.3% 7.8% 7.7% 8.4% 7.8% 7.7% 8.4% North 13.8% 1.0% N/A 21.5% 6.3% 6.2% 6.2% 6.3% 6.2% 6.2% Central 17.7% -3.0% N/A 39.1% 6.7% 6.6% 6.9% 6.7% 6.6% 6.9% South 22.7% 2.5% N/A 60.7% 13.9% 14.2% 15.3% 13.9% 14.2% 15.3% Statewide 22.1% 2.0% N/A 58.3% 13.2% 13.5% 14.5% 13.2% 13.5% 14.5%

Form A-8: Percentage Change in Output Ranges by County

Provide color-coded maps by county reflecting the percentage changes in the weighted average 2% deductible loss costs for frame owners, masonry owners, mobile homes, frame renters, masonry renters, frame condos, and masonry condos from the output ranges using the 2002 Florida Hurricane Catastrophe Fund’s aggregate personal residential exposure data found in the file named “hlpm2002.exe”, as of August 1, 2003.

Counties with a negative percentage change (reduction in loss costs) would be indicated with shades of blue; counties with a positive percentage change (increase in loss costs) would be indicated with shades of red, and counties with no percentage change would be white. The larger the percentage change in the county, the more intense the color-shade.

The maps below show the percentage change in weighted average 2% deductible loss costs by county.

Figure 42: Percentage Change in Weighted Average Loss Costs by County – Frame Owners

Form A-8: Percentage Change in Output Ranges by County

Figure 43: Percentage Change in Weighted Average Loss Costs by County – Masonry Owners

Figure 44: Percentage Change in Weighted Average Loss Costs by County – Mobile Homes

Form A-8: Percentage Change in Output Ranges by County

Figure 45: Percentage Change in Weighted Average Loss Costs by County – Frame Renters

Figure 46: Percentage Change in Weighted Average Loss Costs by County – Masonry Renters

Form A-8: Percentage Change in Output Ranges by County

Figure 47: Percentage Change in Weighted Average Loss Costs by County – Frame Condos

Figure 48: Percentage Change in Weighted Average Loss Costs by County – Masonry Condos

Standard S-1: Modeled Results and Goodness-of-Fit

2006 Statistical Standards

S-1 Modeled Results and Goodness-of-Fit

A. The use of historical data in developing the model shall be supported by rigorous methods published in currently accepted scientific literature.

Historical data have been used to develop the probability distributions for key model variables such as annual hurricane frequency, landfall location, central pressure, radius of maximum winds, forward speed, and track direction. Data for the period 1900-2006 are used to model annual landfall frequency, landfall location, and hurricane intensity. Data for the period 1900-2003 are used for other variables. Spatial smoothing and meteorological adjustments have been used to overcome spatial gaps and other limitations caused by the relative scarcity of the historical data.

The probability distributions used for individual input variables include Negative Binomial for annual landfall frequency, Weibull for central pressure and Lognormal for forward speed. The parameters of these distributions have been estimated using the maximum likelihood method. The adequacy of the fit has been examined using established procedures such as the Kolmogorov-Smirnov and the Shapiro-Wilk tests. Graphical comparisons using Q-Q plots and other procedures have also been performed to confirm the agreement between the historical data and the fitted probability distributions.

B. Modeled and historical results shall reflect agreement using currently accepted scientific and statistical methods.

Agreement between modeled and historical hurricane characteristics is confirmed using widely accepted scientific and statistical methods. The simulated values have been carefully examined and determined to be reasonable based both on statistical and meteorological grounds.

Disclosures

1. Identify the form of the probability distributions used for each function or variable, if applicable. Identify statistical techniques used for the estimates and the specific goodness-of-fit tests applied. Describe whether the p-values associated with the fitted distributions provide a reasonable agreement with the historical data.

Standard S-1: Modeled Results and Goodness-of-Fit

Annual Frequency of Occurrence Storm frequency is modeled using a Negative Binomial distribution fitted to the number of annual landfalls during the period 1900-2006. The parameters of this probability distribution are estimated using the maximum likelihood method. The adequacy of the fit is examined graphically and tested using Pearson’s chi-square goodness-of-fit test. The value of the test statistic is small, and its p-value of 0.33 indicates no lack of fit.

Landfall Location The probability distribution for landfall location is based on the number of historical landfalls per 50- mile coastal segment during the period 1900-2006. Due to the relative scarcity of historical data at this spatial resolution, the estimation involves smoothing of the historical frequencies as well as meteorological adjustments to arrive at credible landfall probabilities. The checks performed on the final landfall distribution include graphical and numerical comparisons of historical and simulated landfall frequencies as well as Chi-square goodness-of-fit tests. The associated p-values indicate no lack of fit.

Central Pressure The probability distribution for central pressure is a Weibull distribution with the shape and scale parameters estimated for each 100-mile coastline segment. The maximum likelihood method is used for the parameter estimation. A second calculation combines the data for six larger regions and computes Weibull parameter estimates for each of these regions. The final probability distribution used for each segment is a linear combination of the segment and regional Weibull distributions. The parameters of the Weibull distribution are estimated for each one hundred nautical mile segment and several regions, using the maximum likelihood estimation method.

The adequacy of the segment and regional Weibull distributions is tested using the Kolmogorov-Smirnov goodness-of-fit test. The empirical and fitted probability distributions are also compared using quantile-quantile plots and other graphical methods. In addition, the historical and simulated central pressure distributions are compared graphically for each 100-mile coastal segment. The various checks performed, along with associated p-values, confirm that the fitted distributions provide a reasonable approximation for central pressure.

Radius of Maximum Winds For each simulated hurricane, the radius of maximum winds is simulated from a regression model that relates the radius to central pressure and latitude. The error term in this model is assumed to follow a Normal distribution. The parameters are estimated using least squares and standard residual checks are performed to determine the adequacy of the fitted model. The resulting values are bounded based on central pressure to produce a final distribution for the radius. The consistency between historical and simulated values is demonstrated using scatter diagrams, as well as segment-by-segment comparisons of observed and simulated values.

Forward Speed

Standard S-1: Modeled Results and Goodness-of-Fit

Forward speed is generated from a Lognormal distribution with parameters estimated for each 100-mile segment. The parameters are estimated using the maximum likelihood method by computing the mean and variance of the log-transformed data. The adequacy of the fit is again tested using the Shapiro-Wilk goodness-of-fit test and by comparing the empirical and fitted cumulative distribution functions. Quantile-quantile plots were also constructed. In addition, the historical and simulated values are compared graphically for each 100-mile coastal segment. The checks performed, including the examination of p- values, indicate no lack of fit and suggest that the lognormal distribution provides a reasonable probability distribution for forward speed.

Storm Heading at Landfall Landfall angle is measured clockwise (+) or counterclockwise (-) with 0 representing due North. Separate distributions for storm heading at landfall are estimated for segments of coastline that are variable in length. The length of each segment is governed by the coastal orientation of that segment. Storm heading is generated from a combined Normal distribution. The maximum likelihood estimation procedure and the Shapiro-Wilk goodness-of-fit tests are used. The resulting distributions are bounded based on the historical record and meteorological expertise. The p-values associated with the tests performed show a reasonable agreement between historical and modeled values.

Storm Tracks AIR’s storm track generation procedure is based on the historical storm tracks available in the HURDAT database. The track information in this database is available at 6-hour time intervals. A time series analysis was performed to determine appropriate models for the dependence present in key model variables from one time period to the next. This included an examination of the autocorrelation function of the original and differenced data corresponding to each model variable. This analysis showed that a random walk with drift is appropriate for the track direction. A first-order autoregressive model is appropriate for forward speed, while a second-order autoregressive model is required to adequately represent central pressure along the track. To capture the spatial variability in the storm characteristics across the Atlantic basin, the parameters in these models were estimated by binning the data into grids that captured the spatial variation. Diagnostic checks on the model included grid-by-grid comparisons of the historical and simulated storm frequencies and intensity distributions across the basin.

Physical Damage The vulnerability functions developed by AIR are based on published structural engineering research, wind engineering principles, damage surveys conducted by wind engineering experts, and analysis of actual loss data. Over the years, AIR has compiled an extensive database of claims data from clients with large portfolios for historical hurricanes affecting various regions along the coast. Validation has been performed by comparing simulated and actual loss data by state, county, ZIP Code and by line of business.

Standard S-1: Modeled Results and Goodness-of-Fit

2. Provide the source and the number of years of the historical data set used to develop probability distributions for specific hurricane characteristics. If any modifications have been made to the data set, describe them in detail and their appropriateness.

Table 17: Sources of Historical Data Source Years of Data

Monthly Weather Review 1900-present NWS-23 1900-1976 NWS-38 1900-1984 Neumann, Charles J., “Tropical Cyclones of the North Atlantic Ocean, 1871-1998 1871-1998.” NCDC, NOAA National Hurricane Center Preliminary Reports for Specific 1977-2005 Hurricanes* Hurricane Best Track Files (HURDAT) available from 1900-2006 http://www.nhc.noaa.gov/tracks1851to2006 atl.txt

Extended Best Track File (EBTRK), available from 1988-1999 NESDIS/CIRA/RAMM, Colorado State University, Fort Collins, CO * Supplemental data added to report by NHC upon request by AIR.

3. Describe the nature and results of the tests performed to validate the wind speeds generated.

The wind speeds generated by the AIR hurricane model have been compared with observed wind speed data obtained from National Hurricane Center (NHC) reports and Monthly Weather Review publications. Also, when available, the wind speeds are compared to wind speed contour maps, such as the one illustrated below, produced by the NOAA’s Hurricane Research Division, and other research publications.

Figure 49: Hurricane Andrew Wind Speed Contour Map

Source: Powell and Houston, 1996

Standard S-1: Modeled Results and Goodness-of-Fit

The AIR hurricane windfield model has been validated using available wind speed data for landfalling hurricanes since Hurricane Alicia in 1983. The following map shows the wind speed validation results for Hurricane Andrew. The numbers shown on the map are actual reported wind speeds. The shaded areas represent AIR simulated wind speeds.

Figure 50: Validation of Wind Speeds for Hurricane Andrew

The work performed as part of the Florida Commission filing has also confirmed that the wind speed calculation works as intended. Using sample input characteristics on central pressure, radius of maximum winds, forward speed, and far field pressure provided by the Commission in 2003, AIR generated hourly wind speed measurements over a rectangular grid covering 65 locations in Southern Florida. The results generated in this study, including the subsequent sensitivity and uncertainty analysis, helped confirm the validity of the calculation.

4. Provide the date of loss of the insurance company data available for validation and verification of the model.

AIR has actual insurance company loss data for the following storms: Hurricanes Hugo (1989), Bob (1991), Andrew (1992), Erin (1995), Opal (1995), Bertha (1996), Fran (1996), Bonnie (1998), Earl (1998), Frances (1998), Georges (1998), Floyd (1999), Irene (1999), and Georges (2001), Charley (2004), Ivan (2004), Frances (2004), and Jeanne (2004).

5. Provide an assessment of uncertainty in loss costs for output ranges using confidence intervals or other accepted scientific characterizations of uncertainty.

Standard S-1: Modeled Results and Goodness-of-Fit

Basic summary statistics and confidence intervals have been computed, and the uncertainty in the output ranges has been found to be consistent with the underlying distribution of hazard.

6. Provide graphical comparisons of modeled and historical data and goodness- of-fit tests. Examples include hurricane frequencies, tracks, intensities, and physical damage.

Annual Landfall Frequency The figure below compares the empirical annual landfall distribution and the fitted distribution. As can be seen, the agreement between the two distributions is quite close and shows no evidence of lack of fit. The calculated value of the Chi-square statistic is 1.56, which for 3 degrees of freedom gives a p-value of 0.33.

Figure 51: Empirical Annual Landfall and Fitted Distributions

0.35

0.30 Historical 0.25 Modeled 0.20

0.15 Frequency 0.10

0.05

0.00 012345678+ Number of Landfalls

Hurricane Tracks The following graph compares the tracks of historical and simulated hurricanes making landfall in a 50-mile coastal segment in South East Florida. The overall behavior of the historical and simulated tracks is clearly quite similar.

Standard S-1: Modeled Results and Goodness-of-Fit

Figure 52: Historical and Simulated Hurricanes Landfalling in SE Florida

AIR Hist 07 Simulated

Intensities The graph below compares the historical and simulated central pressure distributions for a 100-mile segment in South East Florida. The simulated frequencies are drawn from Weibull distributions fitted to data pooled from several coastal segments. Goodness-of-fit summaries are included to illustrate some of the statistical tests that were performed for this variable.

Standard S-1: Modeled Results and Goodness-of-Fit

Figure 53: Historical and Simulated Central Pressure Distributions

CP - Segment 15 0.06 Historical 0.05 Modeled 0.04

0.03

0.02 Relative Frequency Relative 0.01

0 Cat1 Cat2 Cat3 Cat4 Cat5 Category

Standard S-1: Modeled Results and Goodness-of-Fit

Physical Damage The graph included below shows historical and simulated damage ratios versus wind speed for Coverage A based on ZIP Code level data. Each observation refers to an individual ZIP Code. The agreement between the historical and simulated damage ratios is reasonable. This is confirmed by a paired two-sample t-test on the means, which has a p-value of 0.35.

Figure 54: Sample Damage Ratio Comparison

Sample Damage Ratio Comparison 1

0.1

0.01

0.001

0.0001

Damage Ratio 0.00001 Actual 0.000001 Simulated 0.0000001 40 50 60 70 80 90 100 110 Wind Speed (mph)

7. Provide a completed Form S-1, Probability of Florida Landfalling Hurricanes per Year.

A completed Form S-1 is provided on page 215.

8. Provide a completed Form S-2, Probable Maximum Loss.

A completed Form S-2 is provided on page 216.

Standard S-2: Sensitivity Analysis for Model Output

S-2 Sensitivity Analysis for Model Output

The modeler shall have assessed the sensitivity of temporal and spatial outputs with respect to the simultaneous variation of input variables using currently accepted scientific and statistical methods and have taken appropriate action.

Over the past fifteen years, AIR scientists and engineers have done extensive sensitivity testing on all aspects of the hurricane model. This has involved testing alternative probability distributions for key input variables, as well as changing the parameter values of these probability distributions. One-parameter-at-a-time as well as multi-parameter studies have been conducted. Much of this work has focused on the sensitivity of individual model variables on the estimated losses by state, county, and ZIP Code. However, the sensitivity of temporal and spatial wind speeds generated by the model has also been investigated.

Disclosures

1. Provide a detailed explanation of the sensitivity analyses that have been performed on the model above and beyond those completed for the original submission of Form S-5 and provide specific results. (Requirement for modeling companies that have previously provided the Commission with Form S-5. This Disclosure can be satisfied with an updated Form S-5 that incorporates changes to the model since the previous submission of the Form.)

The sensitivity analyses performed on the model include one-parameter-at-a-time studies as well as multi-parameter studies where several input variables are varied simultaneously. In most of these studies, the output variable of interest has been the insured losses evaluated at the state, county, and ZIP Code level. However, the changes in wind speed resulting from changes in the input variables have also been studied. The approaches used include changing the probability distributions used for key input variables and changing the parameters associated with these probability distributions. Model output from base- case simulations has also been examined to identify sensitive input variables.

AIR uses a Weibull distribution to model central pressure. However, the impact on the losses of using a Lognormal distribution has also been tested. For radius of maximum winds, the Lognormal has been considered as an alternative to the Normal distribution for the noise term in the regression model for Rmax. The impact of varying the bounds placed on the radius of maximum wind for very intense storms has also been evaluated. On the vulnerability side, the sensitivity studies performed include changing the damage functions used in the loss estimation.

AIR uses a Negative Binomial distribution to model annual landfall frequency with the parameters estimated using historical data from1900 to present. Due to the limited historical record, sampling error is present in the parameter estimates. AIR has examined the impact of the sampling error by perturbing the mean of the annual landfall distribution and examining the impact on losses.

Standard S-2: Sensitivity Analysis for Model Output

In a multi-parameter study focusing on sampling error, AIR included three input variables: annual landfall frequency, central pressure, and radius of maximum wind. The level of damage was included as a fourth factor in the study. The study utilized a 24 factorial design (see, for example, Box, Hunter, and Hunter, 1978∗) in which the mean levels of each of the four variables were perturbed and placed at two new settings, high or low, resulting in 16 experimental combinations. The high and low mean values for each input variable were generated by changing the mean values of the probability distributions for these variables by one standard error, either up or down. Main effects and interaction effects were computed using ZIP Code level loss costs in Florida as output variable. Among the input variables included in this study, changing central pressure was identified as having the largest impact on losses.

During the past few years, AIR has examined the relationship between the annual landfall frequency and different climate signals, such as sea surface temperatures (SST) in the North Atlantic. Many scientists believe that an increase in SST will lead to an increase in the frequency as well as intensity of landfalling hurricanes. Estimates of the effect of SST on hurricane frequency can be obtained by comparing the number of landfalling hurricanes during years when SST has been warmer than average to the landfall frequency during the entire historical period 1900-2006. The resulting estimates have large standard errors indicating a high degree of uncertainty in the impact of the SST on hurricane frequency and intensity. However, sensitivity tests can be conducted by assuming different levels of impact and examining the changes in the resulting loss costs. These studies confirm previous findings that the losses are sensitive to changes in frequency and the intensity of landfalling storms. Changes in the frequency of major hurricanes, in particular, have a significant impact on the estimated losses.

2. Provide a description of the statistical methods used to perform the sensitivity analysis.

The statistical methods used by AIR include graphical tools such as histograms, empirical density plots and cdf plots, as well as basic summary statistics such as the mean, median, standard deviation, percentiles, etc. In the multi-parameter study described above, the analysis of main effects and interaction effects is equivalent to looking at standardized regression coefficients associated with the input variables.

3. Identify the most sensitive aspect of the model and the basis for making this determination. Provide a full discussion of the degree to which these sensitivities affect output results and illustrate with an example.

Central pressure is one of the most sensitive model parameters. Relatively small changes in this model variable can have a large impact on the resulting output. For example, the four-

∗ Box, G.E.P., W.G. Hunter, and J.S. Hunter. (1978). Statistics for Experimenters: An Introduction to Design, Data Analysis, and Model Building. John Wiley and Sons, New York.

Standard S-2: Sensitivity Analysis for Model Output parameter sensitivity study described above showed that decreasing the central pressure by one standard error resulted in an 18 percent increase in ZIP Code level losses.

4. Describe how other aspects of the model may have a significant impact on the sensitivities in output results and the basis for making this determination.

Other sensitive parameters include the annual landfall frequency, locational frequency, and the radius of maximum winds. The correlation between radius of maximum winds and central pressure is also important as an increase in the radius for very intense storms can have a significant impact on the losses. These findings are based on studies conducted by AIR.

5. Describe actions taken in light of the sensitivity analyses performed.

Over the past fifteen years, AIR scientists have tested numerous distributions for each of the model variables. The probability distributions used in the current model were chosen based on such tests combined with information available in the scientific literature. Also, as a result of its recent studies of the relationship between SST and hurricane frequency, AIR has made a near term alternative stochastic catalog available to its clients. However, all results in this submission are based on the AIR Atlantic Tropical Cyclone Model V9.0.

6. Provide a completed Form S-5, Hypothetical Events for Sensitivity and Uncertainty Analysis (requirement for models submitted by modeling organizations which have not previously provided the Commission with this analysis).

AIR provided a completed Form S-5 (formerly Form F) in compliance with the 2002 Standards.

Standard S-3: Uncertainty Analysis for Model Output

S-3 Uncertainty Analysis for Model Output

The modeler shall have performed an uncertainty analysis on the temporal and spatial outputs of the model using currently accepted scientific and statistical methods and have taken appropriate action. The analysis shall identify and quantify the extent that input variables impact the uncertainty in model output as the input variables are simultaneously varied.

An uncertainty analysis focusing on temporal and spatial wind speed measurements associated with a set of hypothetical storms was performed as part of the Form F analysis conducted in 2003. A study of the resulting losses estimates was part of the analysis. The study included four input variables and the contribution of each variable to the variation in the model output was quantified. This study was later extended to include a more detailed analysis of one of the input variables used in the 2003 analysis.

AIR routinely performs uncertainty analyses in that the variation in the insured losses resulting from simultaneous variation of input variables such as annual storm frequency, storm intensity, radius of maximum winds and forward speed is studied after each simulation run. This involves examining the cumulative distribution functions and various summary statistics associated with the estimated losses.

Disclosures

1. Provide a detailed explanation of the uncertainty analyses that have been performed on the model above and beyond those completed for the original submission of Form S-5 and provide specific results. (Requirement for modeling companies that have previously provided the Commission with Form S-5. This Disclosure can be satisfied with an updated Form S-5 that incorporates changes to the model since the previous submission of the Form.)

An uncertainty analysis focusing on the temporal and spatial behavior of the wind field component of the AIR model was conducted as part of the Form F analysis in 2003. The input variables included central pressure, forward speed, radius of maximum winds, and the gradient wind speed conversion factor described in NWS 23, p.235. The contribution of the different input variables to the uncertainty in the output at different time intervals and at different locations was analyzed following the guidelines provided by the Professional Team. A study of the uncertainty in the estimated losses associated with each event was also performed as part of this study. The results showed that central pressure and the gradient wind speed conversion factor are major contributors to the uncertainty in the output.

This study was expanded to a more detailed examination of the gradient wind speed conversion factor in 2004. Rather than using a fixed value of 0.9 for this factor, random perturbations were added to the factor by making drawings from a N(0,σ2) distribution with different choices for σ. The results showed that relatively small perturbations of this

Standard S-3: Uncertainty Analysis for Model Output factor had a significant impact on the uncertainty in the output. For example, already for σ = 0.02, the expected percentage reduction in the variance of the loss costs attributable to this factor was 29 percent for Category 3 hurricanes and 25 percent for Category 5 hurricanes. The corresponding values for the radius of maximum winds were 34 percent and 67 percent, respectively. However, the contribution of the conversion factor increased relative to that of other input variables as the value of σ increased.

Standard model runs performed by AIR are followed by an uncertainty analysis in the sense that variation in the losses generated by simultaneous variation of the model input variables is always subject to analysis. This involves a study of the simulated cdf’s along with summary statistics such as mean, median, standard deviation, and various percentiles of the loss distribution. Confidence intervals are computed to evaluate the uncertainty in percentiles corresponding to high return period losses. Uncertainty studies of this type are performed on a routine basis and are an essential part of research associated with model changes. Changes related to variables or model components that make a large contribution to the uncertainty in the output are subject to special scrutiny.

2. Provide a description of the statistical methods used to perform the uncertainty analysis.

Statistical methods used to perform the uncertainty analysis include graphical tools such as histograms and empirical cdf plots, and summary statistics such as the mean, median, percentiles, standard deviation, coefficient of variation, etc. Confidence intervals are constructed to assess the sampling variability in estimated values. Variance ratios are used to assess the contribution of individual input variables to the uncertainty in the output.

3. Identify the major contributors to the uncertainty in model outputs and the basis for making this determination. Provide a full discussion of the degree to which these uncertainties affect output results and illustrate with an example.

The 2004 uncertainty analysis conducted by AIR showed that when the conversion factor is held fixed at 0.9, or is perturbed very little (σ=0.01), radius of maximum winds and central pressure are the key contributors uncertainty in loss costs. Radius of maximum winds is a major contributor for very intense hurricanes, while central pressure is more influential for weaker hurricanes. To illustrate, for Category 5 hurricanes, the expected percentage reduction in the variance of the loss cost attributable the radius was 85 percent compared to 26 percent attributable to central pressure. For Category 1 hurricanes, the corresponding values were 7 percent and 96 percent, respectively.

4. Describe how other aspects of the model may have a significant impact on the uncertainties in output results and the basis for making this determination.

Standard S-3: Uncertainty Analysis for Model Output

Our recent work on the relationship between SST and the frequency and intensity of landfalling storms confirm earlier findings that frequency and intensity have a significant impact on the uncertainty in modeled losses.

5. Describe actions taken in light of the uncertainty analyses performed.

Input variables that are major contributors to uncertainty in model output require special scrutiny at the model building stage or when changes are made to the model. AIR has carefully tested the probability distributions used for such variables. For example, a Weibull distribution is used to represent central pressure but alternatives such as the Lognormal distribution have also been tested. The methodology used to simulated radius of maximum winds was recently revised to more fully account for the correlation present between radius, central pressure and latitude. In developing the new methodology, special attention was given to the modeling of the radius for intensive hurricanes where the radius is a major contributor to the uncertainty in the output.

While AIR has studied uncertainty associated with the wind speed conversion factor, additional research will be conducted before the findings are implemented in the model. AIR will also continue to study the relationship between hurricane landfall frequency and climate signals such as the SST to determine if any adjustments to the probability distributions used for annual frequency and intensity are called for.

6. For models submitted by modeling organizations, which have not previously provided this analysis to the Commission, Form S-5 was disclosed under Standard S-2 and will be used in the verification of Standard S-3.

AIR provided a completed Form S-5 (formerly Form F) in compliance with the 2002 Standards.

Standard S-4: County Level Aggregation

S-4 County Level Aggregation

At the county level of aggregation, the contribution to the error in loss costs estimates attributable to the sampling process shall be negligible.

The error in the county level loss costs induced by the sampling process can be quantified by computing standard errors and confidence intervals for the county level loss costs. These calculations have been performed for all counties in Florida. The results indicate the sampling error is negligible for the 50,000-year simulation used to generate the loss costs. Graphs demonstrating the rate of convergence at the county-level have also been produced.

Disclosure

1. Describe the sampling plan used to obtain the average annual loss costs and output ranges. For a direct Monte Carlo simulation, indicate steps taken to determine sample size. For an importance sampling design, describe the underpinnings of the design.

AIR uses constrained Monte Carlo simulation to obtain the average annual loss costs and output ranges. This sample size is based on a calculation placing limits on the standard errors of the output ranges. AIR uses 50,000 simulation years in its FCHLPM filing.

Standard S-5: Replication of Known Hurricane Losses

S-5 Replication of Known Hurricane Losses

The model shall estimate incurred losses in an unbiased manner on a sufficient body of past hurricane events from more than one company, including the most current data available to the modeler. This Standard applies separately to personal residential and, to the extent data are available, to mobile homes. Personal residential experience may be used to replicate structure-only and contents-only losses. The replications shall be produced on an objective body of loss data by county or an appropriate level of geographic detail.

Losses generated by the model’s simulation of past hurricane events reasonably replicate actual incurred losses from those events. This is true for both personal residential of various construction types and for mobile homes, as well as for various coverages. County- level comparisons also show reasonable agreement between modeled and incurred losses.

Disclosures

1. Describe the nature and results of the analyses performed to validate the loss projections generated by the model.

The tables on the following pages show how AIR’s simulated damage ratios compare, in total and by coverage and construction, to the damage ratios of specific client companies for each of eight recent storms: Hurricanes Andrew, Erin, Opal, Bertha, Fran, Bonnie, Earl and Georges. Note that the losses in the tables below have been scaled to protect the identity of the companies.

Standard S-5: Replication of Known Hurricane Losses

Table 18: Actual vs Modeled Losses for Eight Storms and Five Companies Actual Modeled Event Company Loss ($) Loss ($) Andrew A 671,714,829 754,052,572 Andrew C 30,703,489 50,642,736 Andrew Total 702,418,318 804,695,308 Erin B 2,750,567 1,571,546 Erin C 11,424,974 10,956,147 Erin Total 14,175,541 12,527,693 Opal B 11,549,817 6,928,629 Opal C 13,888,128 21,467,522 Opal Total 25,437,945 28,396,151 Bertha A 6,907,485 9,121,813 Bertha B 359,049 307,058 Bertha D 7,382,452 9,663,577 Bertha Total 14,648,986 19,092,448 Fran A 37,074,096 24,792,939 Fran B 4,315,466 1,294,633 Fran D 40,030,203 22,264,297 Fran Total 81,419,765 48,351,869 Bonnie A 3,853,767 16,110,546 Bonnie B 1,797,270 646,469 Bonnie D 7,399,300 21,660,891 Bonnie Total 13,050,337 38,417,906 Earl A 845,260 785,816 Earl D 1,270,044 1,039,987 Earl Total 2,115,304 1,825,803 Georges A 15,608,866 36,017,520 Georges B 2,412,406 2,489,032 Georges D 24,656,941 26,608,314 Georges Total 42,678,213 65,114,866

Standard S-5: Replication of Known Hurricane Losses

Table 19: Actual vs Modeled Losses by Coverage for Eight Storms and Three Companies Actual Modeled Coverage Event Company Loss ($) Loss ($) A Andrew A 454,343,263 557,829,604 A Andrew C 22,769,311 36,745,422 A Erin C 10,391,627 9,694,233 A Opal C 11,715,650 18,347,947 A Bertha A 6,321,284 8,400,884 A Fran A 33,988,382 22,371,310 A Bonnie A 3,444,985 14,274,068 A Bonnie D 4,743,281 19,191,374 A Earl A 777,842 715,991 A Earl D 1,044,269 950,453 A Georges D 12,079,314 22,918,629 Total A 561,619,208 711,439,915 C Andrew A 172,963,076 112,142,217 C Andrew C 6,110,879 10,114,502 C Erin C 796,617 1,162,442 C Opal C 1,586,888 2,546,478 C Bertha A 520,864 672,566 C Fran A 2,673,494 2,058,956 C Bonnie A 342,944 1,535,987 C Bonnie D 2,348,525 2,154,308 C Earl A 60,126 64,604 C Earl D 201,468 85,920 C Georges D 11,205,212 3,121,473 Total C 198,810,093 135,659,453 D Andrew A 44,408,490 84,080,752 D Bertha A 65,337 48,363 D Fran A 412,220 362,673 D Andrew C 1,823,299 3,782,812 D Erin C 236,730 99,471 D Opal C 596,497 573,096 D Bonnie A 65,838 300,492 D Bonnie D 307,494 315,209 D Earl A 7,292 5,222 D Earl D 24,307 3,613 D Georges D 1,372,415 568,212 Total 49,319,919 90,139,915

Standard S-5: Replication of Known Hurricane Losses

Table 20: Actual vs Modeled Losses by Construction Type for Eight Storms and Four Companies Actual Modeled Construction Event Company Loss ($) Loss ($) Frame Andrew A 24,144,527 24,307,191 Frame Andrew C 8,873,505 15,745,727 Frame Erin C 6,841,680 7,350,876 Frame Opal C 9,716,747 18,446,736 Frame Bertha A 5,881,874 7,650,464 Frame Bertha D 6,459,504 6,926,573 Frame Fran A 32,106,008 20,223,123 Frame Fran D 30,736,826 15,234,232 Frame Bonnie A 2,538,634 9,033,928 Frame Earl A 396,584 388,172 Frame Georges A 6,098,852 13,530,113 Frame Bonnie B 1,113,542 510,732 Frame Bertha B 357,230 304,245 Frame Erin B 2,645,378 1,359,471 Frame Fran B 4,296,270 1,282,915 Frame Georges B 1,121,998 1,098,863 Frame Opal B 11,164,951 6,482,574 Frame Georges D 11,025,016 11,122,068 Total Frame 165,519,126 160,998,003 Masonry Andrew A 647,570,302 729,745,382 Masonry Andrew C 19,890,067 34,217,629 Masonry Erin C 2,948,767 2,281,441 Masonry Opal C 1,259,168 1,582,311 Masonry Bertha A 984,279 891,824 Masonry Fran A 4,850,222 2,539,313 Masonry Bonnie B 110,017 28,130 Masonry Georges B 279,793 95,675 Masonry Bonnie A 339,804 1,836,378 Masonry Earl A 139,378 135,542 Masonry Georges A 1,621,248 2,592,118 Masonry Earl D 226,924 202,899 Masonry Georges D 5,593,275 4,541,378 Total Masonry 685,813,244 780,690,020 Mobile Homes Erin B 105,189 212,075 Mobile Homes Bertha B 1,819 2,813 Mobile Homes Fran B 19,196 11,717 Mobile Homes Fran D 244,184 216,644 Mobile Homes Opal B 384,866 446,055 Mobile Homes Opal C 910,779 1,438,475 Mobile Homes Andrew C 478,078 679,380 Mobile Homes Erin C 661,802 911,645 Total Mobile Homes 2,805,913 3,918,804

Standard S-5: Replication of Known Hurricane Losses

Table 21: Actual vs Modeled Losses by County for One Company -- Hurricane Bonnie

Actual Modeled Loss/ Loss/ County Exposure ($) Loss ($) Exposure Exposure ($) Loss ($) Exposure Brunswick 186,735,667 834,911 0.004471 186,735,667 2,009,344 0.010760 Duplin 11,595,616 12,106 0.001044 11,595,616 13,008 0.001122 Lenoir 29,759,624 2,622 0.000088 29,759,624 15,608 0.000524 Pender 539,330,046 896,194 0.001662 539,330,046 1,710,913 0.003172 Onslow 41,844,577 126,749 0.003029 41,844,577 217,502 0.005198 Total 809,265,530 1,872,582 0.002314 809,265,530 3,966,375 0.004901

2. Provide a completed Form S-3, Five Validation Comparisons.

A completed Form S-3 is provided on page 217.

Standard S-6: Comparison of Projected Hurricane Loss Costs

S-6 Comparison of Projected Hurricane Loss Costs

The difference, due to uncertainty, between historical and modeled annual average statewide loss costs shall be reasonable, given the body of data, by established statistical expectations and norms.

The average annual statewide loss cost produced using the 2002 FHCF exposure data and the Base Storm Set extended to include the 2006 hurricane season is $1.670 billion. The statewide loss cost produced on an average industry basis is $2.369 billion. The difference between these two numbers is statistically reasonable.

Disclosures

1. Describe the nature and results of the tests performed to validate the expected loss projections generated. If a set of simulated hurricanes or simulation trials was used to determine these loss projections, specify the convergence tests that were used and the results. Specify the number of hurricanes or trials that were used.

Under the simulation technique used by the AIR hurricane model, the number of iterations required for the sample mean to converge within P% of the true mean μ is given by: 2 ⎛ Sz ⎞ n = ⎜ α 2/ ⎟ ⎝ Pμ ⎠

For the purpose of ratemaking in Florida, 50,000 model iterations are run. Standard errors and convergence charts showing the change in losses for increasing iterations have been used to validate convergence.

Confidence intervals constructed for the difference between historical and modeled average annual loss costs show that the difference between the two sets of loss costs is statistically reasonable. The validation of projected loss costs also includes comparisons of other summary statistics such as the historical and simulated losses at various return periods.

As described elsewhere in this document, AIR has carefully validated all model components including meteorological input variables such as annual frequency, landfall location, and storm intensity. Loss cost maps have been inspected for smoothness and consistency and found to reasonably reflect differences in landfall rates and storm characteristics for different parts of Florida.

2. Identify differences, if any, in how the model produces loss costs for specific historical events versus loss costs for events in the stochastic hurricane set.

Standard S-6: Comparison of Projected Hurricane Loss Costs

The methodology for producing loss costs for historical events is the same as that used for generating loss costs for events in the stochastic catalog.

3. Provide a completed Form S-4, Average Annual Zero Deductible Statewide Loss Costs.

A completed form S-4 is provided on page 222.

Form S-1: Probability of Hurricanes per Year

Form S-1: Probability of Florida Landfalling Hurricanes per Year

Complete the table below showing the probability of landfalling Florida hurricanes per year. Modeled probability should be rounded to four decimal places. The historical probabilities below have been derived from the Commission’s Official Hurricane Set. If the National Hurricane Center’s HURDAT or other hurricanes in addition to the Official Hurricane Set as specified in Standard M-1 are used by the modeler, then the historical probabilities should be modified accordingly. If the National Hurricane Center’s HURDAT is used, provide the HURDAT revision date.

Model Results Probability of Florida Landfalling Hurricanes per Year

Number Of Hurricanes Historical Modeled Per Year Probability* Probability 0 0.5981 0.5462 1 0.2523 0.3343 2 0.1215 0.0939 3 0.0280 0.0208 4 0.0000 0.0040 5 0.0000 0.0009 6 0.0000 0.0001 7 0.0000 0.0000 8 0.0000 0.0000 9 0.0000 0.0000 10 or more 0.0000 0.0000

* Revised to include the 2006 season.

Form S-2: Probable Maximum Loss (PML)

Provide projections of the insured loss for various probability levels using the hypothetical data set provided in the file named “FormA1Input06.xls.” Provide the total average annual loss for the PML distribution. If the methodology of your model does not allow you to produce a viable answer, please state so and why.

Part A

Return Time Probability of Estimated (years) Exceedance Loss

Top Event 0.002% 126,765,174 10,000 0.01% 113,163,163 5,000 0.02% 99,392,473 2,000 0.05% 88,616,293 1,000 0.10% 79,258,641 500 0.20% 65,553,392 250 0.40% 51,697,756 100 1.00% 36,299,892 50 2.00% 25,081,887 20 5.00% 12,884,775 10 10.00% 6,357,630 5 20.00% 2,328,959

Part B

Mean (Total Average Annual Loss) 2,442,171 Median 86,114 Standard Deviation 7,101,815 Interquartile Range 1,479,121 Sample Size 50,000 years of simulated events

Form S-3: Five Validation Comparisons

A. Provide five validation comparisons of actual exposures and loss to modeled exposures and loss. These comparisons must be provided by line of insurance, construction type, policy coverage, county or other level of similar detail in addition to total losses. Include loss as a percent of total exposure. Total exposure represents the total amount of insured values (all coverages combined) in the area affected by the hurricane. This would include exposures for policies that did not have a loss. If this is not available, use exposures for only those policies that had a loss. Specify which was used. Also, specify the name of the hurricane event compared.

Comparison #1 Hurricane = Erin Exposure = Total Note: All the exposures and losses have been multiplied by a single constant to disguise the identity of the client. Actual Modeled Construction Loss/ Loss/ Exposure ($) Loss ($) Exposure Exposure ($) Loss ($) Exposure Frame 10,819,207,281 6,841,680 0.000632 10,819,207,281 7,350,876 0.000679 Masonry 9,580,238,782 2,948,767 0.000308 9,580,238,782 2,281,441 0.000238 Mobile Homes 510,973,002 661,802 0.001295 510,973,002 911,645 0.001784 Total 20,910,419,065 10,452,249 0.000500 20,910,419,065 10,543,962 0.000504

Comparison #2 Hurricane = Andrew Exposure = Total Note: All the exposures and losses have been multiplied by a single constant to disguise the identity of the client. Actual Modeled Construction Loss/ Loss/ Exposure ($) Loss ($) Exposure Exposure ($) Loss ($) Exposure Frame 1,862,768,694 24,144,527 0.012962 1,862,768,694 24,307,191 0.013049 Masonry 16,086,058,206 647,570,302 0.040257 16,086,058,206 729,745,382 0.045365 Total 17,948,826,900 671,714,829 0.037424 17,948,826,900 754,052,573 0.042011

Form S-3: Five Validation Comparisons

Comparison #3 Hurricane = Opal Exposure = Total Note: All the exposures and losses have been multiplied by a single constant to disguise the identity of the client. Actual Modeled Coverage Loss/ Loss/ Exposure ($) Loss ($) Exposure Exposure ($) Loss ($) Exposure Coverage A 10,189,880,881 11,715,650 0.001150 10,189,880,881 18,347,947 0.001801 Coverage C 6,581,863,446 1,586,888 0.000241 6,581,863,446 2,546,478 0.000387 Coverage D 1,868,657,981 596,497 0.000319 1,868,657,981 573,096 0.000307 Total 18,640,402,308 13,899,035 0.000746 18,640,402,308 21,467,521 0.001152

Comparison #4 Hurricane = Bertha Exposure = Total Note: All the exposures and losses have been multiplied by a single constant to disguise the identity of the client. Actual Modeled Coverage Loss/ Loss/ Exposure ($) Loss ($) Exposure Exposure ($) Loss ($) Exposure Coverage A&B 17,037,964,896 6,321,284 0.000371 17,037,964,896 8,400,884 0.000493 Coverage C 8,769,230,846 520,864 0.000059 8,769,230,846 672,566 0.000077 Coverage D 3,350,605,940 65,337 0.000020 3,350,605,940 48,363 0.000014 Total 29,157,801,682 6,907,485 0.000237 29,157,801,682 9,121,813 0.000313

Comparison #5 Hurricane = Bonnie Exposure = Total Note: All the exposures and losses have been multiplied by a single constant to disguise the identity of the client. Actual Modeled County Loss/ Loss/ Exposure ($) Loss ($) Exposure Exposure ($) Loss ($) Exposure Brunswick 186,735,667 834,911 0.004471 186,735,667 2,009,344 0.010760 Duplin 11,595,616 12,106 0.001044 11,595,616 13,008 0.001122 Lenoir 29,759,624 2,622 0.000088 29,759,624 15,608 0.000524 Pender 539,330,046 896,194 0.001662 539,330,046 1,710,913 0.003172 Onslow 41,844,577 126,749 0.003029 41,844,577 217,502 0.005198 Total 809,265,530 1,872,582 0.002314 809,265,530 3,966,375 0.004901

Form S-3: Five Validation Comparisons

B. Provide scatter plot(s) of modeled vs. historical losses for each of the five validation comparisons. (Plot the historical losses on the x-axis and the modeled losses on the y-axis.)

Rather than using directly a specific published hurricane wind field, the winds underlying the modeled loss cost calculations must be produced by the model being evaluated and should be the wind field most emulated by the model.

Figure 55: Scatter Plot of Comparison #1

12,000 10,000 8,000 6,000 Historical 4,000 Modeled 2,000 Loss ($ Thousands) ($ Loss - - 10,000,000 20,000,000 30,000,000 Exposure ($ Thousands)

Figure 56: Scatter Plot of Comparison #2

800,000 700,000 Historical 600,000 500,000 Modeled 400,000 300,000 200,000 100,000 - Loss ($ Thousands) ($ Loss - 5,000,000 10,000,000 15,000,000 20,000,000

Exposure ($ Thousands)

Form S-3: Five Validation Comparisons

Figure 57: Scatter Plot of Comparison #3

25,000

20,000

15,000

10,000 Historical

5,000 Modeled

Loss ($ Thousands) ($ Loss - - 5,000,000 10,000,000 15,000,000 20,000,000

Exposure ($ Thousands)

Figure 58: Scatter Plot of Comparison #4

10,000

8,000

6,000

4,000 Historical

2,000 Modeled

Loss ($ Thousands) ($ Loss - 10,000,000 20,000,000 30,000,000 40,000,000

Exposure ($ Thousands)

Form S-3: Five Validation Comparisons

Figure 59: Scatter Plot of Comparison #5

5,000

4,000 Historical

3,000 Modeled

2,000

1,000

Loss ($ Thousands) ($ Loss - - 200,000 400,000 600,000 800,000 1,000,000

Exposure ($ Thousands)

Form S-4: Average Annual Zero Deductible Statewide Loss Costs

Form S-4: Average Annual Zero Deductible Statewide Loss Costs— Historical versus Modeled

A. Provide the average annual zero deductible statewide loss costs produced using the list of hurricanes in the Base Hurricane Storm Set based on the 2002 Florida Hurricane Catastrophe Fund’s aggregate personal residential exposure data, as of August 1, 2003 (hlpm2002.exe).

Average Annual Zero Deductible Statewide Loss Costs

Time Period Historical Hurricanes Produced by Model Current Year 1.670 2.369 Previous Year 1.501 2.075 Second Prior 1.494 1.955 Percentage Change Current Year/Previous Year 11.3% 14.2% Percentage Change Current Year/Second Prior 11.8% 21.2%

B. Provide a comparison with the statewide loss costs produced by the model on an average industry basis.

The average annual zero deductible statewide loss cost produced using the list of hurricanes in the Official Storm Set including the 2006 hurricane season and based on the 2002 FHCF aggregate personal residential exposure data is $1.670 billion (μH). The statewide loss cost produced on an average industry basis is $2.369 billion (μS).

C. Provide the 95% confidence interval on the differences between the mean of the historical and modeled loss.

The 95% confidence interval on the difference between these two numbers is (-1.49 billion <= (μH - μS) <= +0.10 billion).

2006 Computer Standards Standard C-1: Documentation

2006 Computer Standards

C-1 Documentation

A. The modeler shall maintain a primary document binder, containing a complete set of documents specifying the model structure, detailed software description, and functionality. Development of each section shall be indicative of accepted software engineering practices.

A primary document binder containing fully documented sections for each computer standard is available for inspection by the Professional Team. This includes complete user documentation as well as consistent documentation for all computer software relevant to the submission. The documentation supporting CLASIC/2 contained in the primary document binder is reviewed and updated each time there is a model version number change, which is typically once a year. An additional binder is included to provide the reader with a full table of contents for the documents contained in binders C-1 through C- 7.

B. All computer software (i.e., user interface, scientific, engineering, actuarial, data preparation and validation) relevant to the modeler’s submission shall be consistently documented and dated.

All computer software, data files, and databases are fully documented and such documentation shall be available for review by the Professional team.

C. Documentation shall be created separately from the source code.

Documentation is provided via in-line detailed comments and external higher level documentation that will be able to be available for review by the Professional Team.

Standard C-2: Requirements

C-2 Requirements

The modeler shall maintain a complete set of requirements for each software component as well as for each database or data file accessed by a component.

CLASIC/2™ documentation specifying model requirements and computer implementation for each software component is available for verification by the Professional Team. Also included in the C-2 Binder is a document that explains the changes from the existing version of CLASIC/2 to the next version of CLASIC/2.

The CLASIC/2™ requirements specification documents describe functional, user interface, data format, security, performance and other requirements of CLASIC/2™ for the hurricane peril; it also describes design constraints.

Disclosure

1. Provide a description of the documentation for interface, human factors, functionality, documentation, data, human and material resources, security, and quality assurance.

The documentation for interface, human factors, and functionality is provided by the Clasic/2 Users Manual which is a hard copy paper based manual. This documentation is also provided via help in the application by accessing the help menu.

The data files and database documentation is provided in a set of manuals, which describes each of the schema definitions as well as a high level explanation of the database. All data file documentation is maintained within source control and carefully managed throughout the development process.

Security procedures as well as our security policy are explained in separate documents.

The Quality Assurance procedures are explained in a separate document.

Standard C-3: Model Architecture and Component Design

C-3 Model Architecture and Component Design

The modeler shall maintain and document (1) detailed control and data flow diagrams, and interface specifications for each software component, and (2) schema definitions for each database and data file. Documentation shall be to the level of components that make significant contributions to the model output.

A component design document, included in the primary document binder, contains detailed control and data flow diagrams and interface specifications that illustrate the component design of the CLASIC/2™ software system and the architecture of the AIR hurricane model, its components and sub-components, including the interface to the CLASIC/2 API. The component design document includes detailed control and data flow diagrams for event generation, and class diagrams for all other portions of the hurricane model and CLASIC/2™ system. This document is available to the Professional Team during its site visit.

Model component custodians are identified in the primary document binder. These individuals will be available to explain the functional behavior of any component referenced in the component design document and to respond to questions concerning changes in code, documentation, or data for that component.

Each data file and database that is used by the model has its schema documented in an external document that is part of the document binder.

Standard C-4: Implementation

C-4 Implementation

A. The modeler shall maintain a complete procedure of coding guidelines consistent with accepted software engineering practices.

AIR maintains a complete set of software engineering practices for coding and documentation guidelines that are followed by the software developers.

B. The modeler shall maintain a complete procedure used in creating, deriving, or procuring and verifying databases or data files accessed by components.

AIR maintains a procedure for procuring and creating data files and databases. AIR will demonstrate its process for verifying data files and databases accessed by the modeling components.

C. All components shall be traceable, through explicit component identification in the flow diagrams, down to the code level.

AIR has developed documentation that provides component identification from documentation diagrams which are fully traceable down to the code level.

D. The modeler shall maintain a table of all software components affecting loss costs, with the following table columns: (1) Component name, (2) Number of lines of code, minus blank and comment lines; and (3) Number of explanatory comment lines.

AIR has a table of components in the primary document binder that contains each of the Component names, the Number of lines of code, Comments, and Blank lines.

E. Each component shall be sufficiently and consistently commented so that a software engineer unfamiliar with the code shall be able to comprehend the component logic at a reasonable level of abstraction.

AIR has developed documentation that is clearly written and that can by used by new software engineers to gain an understanding of the software being reviewed.

Disclosure

1. Specify the hardware, operating system, other software, and all computer languages required to use the model.

Standard C-4: Implementation

System Requirements CLASIC/2 can be configured in a variety of different ways. The table below illustrates the minimum hardware configurations.

Minimum Hardware Configuration Configuration A Configuration B Configuration Workstation + Workstation Database Server Component Database Server Memory 2 GB 1 GB 1 GB CPU Dual 2.0 GHz 2.0 GHz 2.0 GHz Disk Space 100 GB 20 GB 80 GB

Required Software Database Server Windows 2000 Server, Windows 2003 Server SQL Server 2000 Standard Edition Service Pack 3a

Workstation Windows 2000 or Windows XP SP1 or later Internet Explorer 5.5 or later

Computer Languages: C++ FORTRAN Visual Basic

Other Software: Stingray Objective Grid by Rouge Wave Software Map Objects by ESRI SEGUE SILKTEST AIR Tasks MICROSOFT Visual Source Safe

Standard C-5: Verification

C-5 Verification

A. General

For each component, the modeler shall maintain procedures for verification, such as code inspections, reviews, calculation crosschecks, and walkthroughs, sufficient to demonstrate code correctness.

AIR software engineers employ a variety of verification procedures to check code correctness. These procedures include code-level debugging, component-level unit testing, verifying newly developed code against a stable reference version, and running diagnostic software tools to detect runtime problems.

In addition, other verification mechanisms are used to test the correctness of key variables that might be subject to modification. These mechanisms include code tracing, intermediate output printing and error logging. Examples of these verification procedures, including code inspections, reviews, calculation crosschecks, walk-throughs, and the use of logical assertions and exception-handling mechanisms in the code, are described with the documentation, can be shown to the Professional Team during its on-site visit.

B. Component testing

1. The modeler shall use testing software to assist in documenting and analyzing all components.

2. Unit tests shall be performed and documented for each component.

3. Regression tests shall be performed and documented on incremental builds.

4. Aggregation tests shall be performed and documented to ensure the correctness of all model components. Sufficient testing shall be performed to ensure that all components have been executed at least once.

All members of AIR’s software development group are responsible for insuring the quality of their work through unit testing. Unit tests are performed for individual software components, independent of all other components, and are documented. A complete set of CLASIC/2 test plans structured according test classification, is provided in the C-5 binder.

In addition, the software quality assurance (SQA) department is charged with responsibility of formal testing procedures, through all successive phases of development from design, coding, initial testing, and regression testing. Specifically, the SQA department performs “black-box”, interface, and integration testing, supported by SilkTest 6.5 automated testing software, and MUTS (Model Unit Testing System).

Standard C-5: Verification

Materials that demonstrate SQA processes will be available to the Professional Team for review during their on-site visit.

C. Data Testing

1. The modeler shall use testing software to assist in documenting and analyzing all databases and data files accessed by components.

2. The modeler shall perform and document integrity, consistency, and correctness checks on all databases and data files accessed by the components.

The software quality assurance (SQA) department is charged with responsibility of formal testing procedures, through all successive phases of development from design, coding, initial testing, and regression testing. Specifically, the SQA department performs “black- box”, interface, and integration testing, supported by SilkTest 6.5 automated testing software, and MUTS (Model Unit Testing System).

AIR has a Verification Utility program that checks the existence, consistency, and correctness on all databases and data files. This program will verify each database and data file matches a known version of the specific database or data file by generating an MD5 digital signature for each database and data file. After a database or file is installed, or at anytime, a database’s or data file’s MD5 signature can be compared with the original generated value to determine if a file is still valid for use.

Disclosures

1. State whether the model produces the same loss costs if it runs the same information more than once without changing the seed of the random number generator.

Modeled loss costs do not change if identical exposure information is analyzed more than once, without changing the seed of the random number generator.

2. Provide an overview of the component testing procedures.

Model component testing procedures are divided into three braod sections. These include procedures to 1) ensure that the event generation, local intensity, and damage estimation modules are functioning correctly in each component and as a whole, 2) perform reasonability checks on the loss results, hazard pattern analysis, and document and quantify model changes, and 3) check various other model functionalities.

Standard C-6: Model Maintenance and Revision

C-6 Model Maintenance and Revision

A. The modeler shall maintain a clearly written policy for model revision, including verification and validation of revised components, databases, and data files.

AIR maintains a clearly documented policy for model revision with respect to methodology and data. AIR employs a verification mechanism consisting of manual comparisons of its data files and databases used in the modeling process and a computer verification process that consists of comparing program, configuration, data file and database MD5 cryptographic service values against their known valid values.

B. A revision to any portion of the model that results in a change in any Florida residential hurricane loss cost shall result in a new model version number.

Since 1987, AIR has implemented a clearly documented policy for model revision with respect to methodology and data. Any enhancement to the model that results in a change in hurricane loss costs also results in a new model version number. The model is updated each year. At a minimum each year, the ZIP Code information is updated to take into account the most recent data. Specifically, the ZIP Code Centroids are updated, and using this new information, the ZIP Code site characteristics are updated. These characteristics include elevation, surface roughness, and distance from the coastline. The historical meteorological information is periodically updated to reflect new events (or lack thereof).

Other enhancements may be made to the model based on ongoing research undertaken by AIR scientists and engineers. For example, post disaster damage surveys and collected loss data from actual events may improve our understanding of the effectiveness of building codes in specific areas.

C. The modeler shall use tracking software to identify all errors, as well as modifications to code, data, and documentation.

AIR’s software development group employs source revision and control software for all software product development and documentation, including CLASIC/2™.

Disclosure

1. Identify procedures used to maintain code, data, and documentation.

AIR employs Microsoft Visual Source Safe, an accepted and effective software product for version control. In addition, AIR employs consistent and documented methods for data and documentation control for all software product development and QA scripts. The installation date, current version number and date of most recent changes are documented

Standard C-6: Model Maintenance and Revision for the individual components in the system decomposition. Samples of the software version control using VSS are provided in the C-6 binder.

We are prepared to demonstrate these systems and methodologies during the Professional Team’s on-site visit.

Standard C-7: Security

C-7 Security (*Significant Revision)

The modeler shall have implemented and fully documented security procedures for: (1) secure access to individual computers where the software components or data can be created or modified, (2) secure operation of the model by clients, if relevant, to ensure that the correct software operation cannot be compromised, (3) anti-virus software installation for all machines where all components and data are being accessed, and (4) secure access to documentation, software, and data in the event of a catastrophe.

AIR has implemented security procedures for access to code, data, and documentation that are in accordance with standard industry practices.

AIR employs a number of physical and electronic security measures to protect all code, data and documentation against both internal and external potential sources of damage, and against deliberate and inadvertent, unauthorized changes.

AIR’s security policy document describes these measures in greater detail.

Disclosure

1. Describe methods used to ensure the security and integrity of the code, data, and documentation.

AIR’s network and the files related to CLASIC/2 and the U.S. hurricane model are physically protected in a number of ways. AIR is located in a multi-story office building with restricted access controlled by professional security staff 24 hours a day. Additionally, the network file servers are located in a secure server room. Access to the server room is granted by electronic badge verification and is limited to essential personnel only.

A thorough policy of virus protection covers all network equipment. In addition to other virus protection measures, a full network virus scan is performed weekly.

AIR’s network back-up policy covers all files stored on the network, including all documents, source and data files related to CLASIC and the U.S. hurricane model. Incremental back-ups are performed daily during the workweek, and a full back-up is performed at the end of the workweek. Back-up tapes are kept permanently in secure offsite storage.

Standard C-7: Security

Secure operation of the model by a client is controlled by the CLASIC/2 software checking version numbers of all software and data files against numbers that are compiled into the code. Prior to running the model, the software checks to see that all data and software that is associated with running the analysis is operating at the same version.

Attachment A: Project Information Assumption Form

Project Information Assumption Form

Project Information & Assumptions Form

Contact Information

Subscriber: Florida Commission on Contract Number: CLXXXX Hurricane Loss Projection Methodology Contact Person: Donna Sirmons Lead AIR Contact: Rob Newbold E-mail Address: sirmons [email protected] E-mail Address: [email protected] Telephone: (850) 413-1349 Telephone: (617) 267-6645 Fax Number: (850) 413-1344 Fax Number: (617) 267-8284

Project Status Exposure Summary Sent: February 14, 2007 Initial Analysis Report Due: February 28, 2007 Follow-up Perils to be Modeled: One (1)

Perils & Models

Peril 1: HURRICANE Simulation Years: 50,000 Implementation: CLASIC/2

Model Name: Atlantic Tropical cyclone Model 9.0 Version: 9 0 0

Reports & Deliverables Format of Deliverable: Electronic File Bound Report

Standard Reports: Distribution of Potential Catastrophe Losses Loss Cost and Pure Premium State Other: Average Annual Losses State Other:

Large Loss Scenarios Scenario Value 120 Return Period 2 50 3 100 Rank 4 500 5 700

Customized Reports: CLFtm Submission Pack (UNICEDER/2, Company Loss Files)

Other: Output Ranges

Attachment A: Project Information Assumption Form

Original Data File Information

Received: November 1, 2006 Data In Force date: N/A Logged: November 1, 2006 Original File Name(s): N/A Data Media:N/A Working File Name(s): OR_06Zips_ZeroDed OR 06Zips MultiDed

File Format: UNICEDE/px Access Database Text Other: Contrived Data per Output Range Specifications

Resolution of Data Provided: Individual Risk Aggregated to Zipcode Aggregated (Other) :

Level of Location Detail: Geo-code County Territory

Street/City 9-Digit Zip 5-Digit Zip Mixed

Level of Analysis Resolution: Geo-code 9-Digit Zip 5-Digit Zip Mixed

Original Data File Summary Number of Total Total Number of Records Insured Value Total Replacement Value Deductible Value Risks Various Assumptions - See 6,426 $615,060,000 $61,506,000Page 4 6,426

Excluded Records Summary Field Total Invalid / Excluded Reason for Description Recs Risks Insured Value Exclusion There are no excluded records

Total Excluded 00 $0 Net Exposures to be Modeled 6,426 6,426 $615,060,000

Attachment A: Project Information Assumption Form

Construction/Occupancy Information and Data Mapping Client Line Of Business MODEL Occupancy Code (OC) Name Construction Insured Value Code Risks Construction Code (CC) Condo Unit Owners N/A Frame $68,850,000 918 OC : 306: Apartment / Condominium CC : 101: Wood Frame Condo Unit Owners N/A Masonry $68,850,000 918 OC : 306: Apartment / Condominium CC : 111: Masonry Owners N/A Frame $165,240,000 918 OC : 301: General Residential CC : 101: Wood Frame Owners N/A Masonry $165,240,000 918 OC : 301: General Residential CC : 111: Masonry Mobile Home N/A Mobile Home $82,620,000 918 OC : 301: General Residential Owners CC : 191: Mobile Homes Renters N/A Frame $32,130,000 918 OC : 306: Apartment / Condominium CC : 101: Wood Frame Renters N/A Masonry $32,130,000 918 OC : 306: Apartment / Condominium CC : 111: Masonry

Total Insured Value to be Modeled: $ 615,060,000 6,426

Attachment A: Project Information Assumption Form

Line of Business & Coverage Summary A B C D LOB Limits Building Other Structures Contents Loss of Use Code Apply Rep Limit Ded Rep Limit Ded Rep Limit Ded Rep/d* Limit Ded

Condo C L 10% CovC BA, BP $0 $0 $0 L $50,000 BA, BP $150 40% CovC BA, BP

Owners C L $100,000 BA, BP L 10% Cov A BA, BP L 50% CovA BA, BP $150 20% CovA BA, BP

MH C L $50,000 BA, BP L 10% Cov A BA, BP L 50% CovA BA, BP $150 20% CovA BA, BP

Renters C $0 $0 $0 $0 $0 $0 L $25,000 BA, BP $150 40% CovC BA, BP

Note: Analyses are conducted at deductible levels of $0, $500, $1,000, $2,500, 1%, 2%, and 5%.

* Loss of Use Replacement (Rep/d) is a per diem value.

CLASIC/2 Key: Limit Application Code Deductible Application Code (Ded) N = None N = None C = Applies by Coverage SA = Combined Flat S = Applies to the sum of all coverages SP = Combined Percent of Coverage SL = Combined Percent of Loss Limit Value (Limit) CA = By coverage Flat P = As Provided CP = By Coverage Percent Other as noted BA = Combined excluding time Flat BP = Combined excluding time Percent Replacement Value (Rep) MA = Mini-policy Flat P = As Provided MP = Mini-Policy Percent. L = Equal Limit Other as noted Other as noted

Attachment A: Project Information Assumption Form

Geo-code Record Summary

Records Records Matched Records Records Records Matched at At Relaxed Matched at Matched at Remapped to Book Names Address Address Zipcode County AIR Zip OR_06Zips_ZeroDed, 0 0 6,426 0 0 OR_06Zips_MultiDed Total Number of Records (each book): 6,426

Total Number of Risks (each book): 6,426

Modeling Assumptions & Criteria

Aggregation of Input Data: Modeled as Provided Aggregated By:

Geographical Resolution of Results: Location Postal code County State Coverage Event

Analysis Save Results: Contract Contract/Summary Layer Country

Analysis Specifications: Storm Surge (Flooding, default is 10% of separately modeled surge loss)

Demand Surge (Post Event Inflation)

Loss Adjustment Expenses Factor:

Physical Properties:

Zipcode Average Individual Site

Output Aggregation Fields for Reporting: Territory State County Postal code Coverage Line of Business Event Location

Other Reporting: Low, High, and Weighted Average Loss Cost by County calculated in post-processing

Attachment A: Project Information Assumption Form

Location Detail Characteristics

Characteristic Provided Modeled

Age Height Floor of Interest Building Condition Proximity Exposure Tree Exposure Small Debris Source Large Missile Source Terrain Roughness Average Height of Adjacent Buildings Building Orientation Building Shape Torsion Elements Soft Story Structural Irregularity Special Earthquake Resistant Systems Retrofit Measures Roof Geometry Roof Pitch Roof Covering Roof Deck Roof Covering Attachment Roof Deck Attachment Roof Anchorage Year Roof Built Wall Type Wall Siding Glass Type Glass Percent Window Protection Exterior Doors Building Foundation Connection Foundation Type Internal Partition Walls Wall Attached Structures Appurtenant Structures Roof Attached Structures

Note: Age is used for Earthquake and Florida Hurricane analysis. For Florida Hurricane analysis, selecting year built =2002 (or later) specifies that the building has been consrtucted to 2001 Florida Building Code standards.

Attachment A: Project Information Assumption Form

Exposure Notes & Customized Assumptions

Attachments & Exhibits

Insured Value Summary By LOB: County State Building (A) Building (B) Building (A+B) Contents (C) Time (D) Total

Replacement Value Summary By LOB: County State Building (A) Building (B) Building (A+B) Contents (C) Time (D) Total

Deductible Summary by LOB: County State

Building (A) Building (B) Building (A+B) Contents (C) Time (D) Total

Premium Summary: County State

Deductible By Coverage: County State

Insured Value By Coverage: County State

Construction Summary

Exposure Summary and Modeling Assumption Approval

Subscriber Signature: Date: Print Name: Title:

Attachment A: Project Information Assumption Form

************************************************************************ AOP Result Set Name Custom ... ************************************************************************ Description Custom ... Event set: 50K US AP (2007) - Standard * Stochastic Y * YearStart 1 * YearEnd 50000 Perils included: * EQ Shake - * EQ Fire Following - * Wildfire - * EQ Sprinkler Leakage - * Flood - * Tropical Cyclone Y [Model 23 v:3.06.1223 y:50000 e:17455] [Model 27 v:9.00.117 y:50000 e:165720] * Storm Surge - * Severe Thunderstorm - * Winterstorm - * Terrorism - Results Level Event Total Coverage Resolution All combined Net Loss defined as Net of deductibles and applicable reinsurance Apply options: * SaveResults_Contract Y * SaveResults_Summary Y * SaveResults_Area_Summary - * SaveResults_Layer - * SaveResults_Coverage Y * SaveResults_Injury - * SaveResults_Claims_Count - * SaveResults_EPCurve - * BinStart 0.00 * BinDelta 1000.00 * EPStart 0.00000 * EPEnd 0.99998 * EPDelta 0.00002 * ApplyRe_QS - * ApplyRe_SS - * ApplyRe_PR - * ApplyRe_FAC - * ApplyAvg_Properties Y * ApplyDemandSurge Y(Aggregate) * ApplyUncertainty - * Save_Ground-up - * Save_Gross - * Save_Net - * Save_Percentile1 0.00 * Save_Percentile2 0.00 * Correlation - * Intra-Policy 0.00 * Inter-Policy 0.00 * Area 0.00 * Subarea 0.00 * Postcode 0.00 * ApplyExposureFilter - * Submit Status - * ApplyAsOfDate - * As of Date - * ApplyTargetFilter - * BookList - * ApplyEventFilter - * Rules -

Attachment A: Project Information Assumption Form

* Area - * Events - * ApplyGlobalOverrides - * Global Overrides - * ApplyMAOLTerms - * MAOL Term 1 - * MAOL Term 2 - * MAOL Term 3 - * MAOL Term 4 -

Attachment B: Model Evaluation and Curriculum Vitae of Joseph Minor, Ph.D., P.E.

A Peer Review of

The AIR Worldwide Corporation Hurricane Model

Joseph E. Minor, P.E. February 23, 2007

INTRODUCTION

AIR Worldwide Corporation (AIR) retained Dr. Joseph E. Minor, P.E. to conduct a peer review of vulnerability standards within the AIR Hurricane Model. Dr. Minor was briefed on the AIR Hurricane Model on February 9, 2007 in the AIR offices in Boston, Massachusetts. Dr. Minor’s training as a structural engineer with experience in wind engineering provides a basis for his analysis of Sections V-1 (Derivation of Vulnerability Functions), V-2 (Mitigations Measures), and V-3 (Mitigation Measures – Mean Damage Ratio Trade Secret List Item) of the 2006 AIR submission to the Florida Commission on Hurricane Loss Projection Methodology (the Florida Commission).

REVIEWER COMMENTS

V-1 Derivation of Vulnerability Functions

The AIR Hurricane Model is based on historical data and expert opinion. Hurricanes occurring in 2004 and 2005 have significantly enhanced the pool of historical data and have offered opportunities for focused site inspections that enhance expert opinion. Moreover, these recent historical data acquisitions contain new information on the effects of building code changes and mitigation measures that will further advance the effigy of vulnerability functions.

B. The method of derivation of the vulnerability functions shall be theoretically sound.

The vulnerability functions in the AIR Hurricane Model are derived using historical data and expert opinion, augmented by site inspections and information from newly published literature. Recent (2004-2005) hurricane events continue to produce new, relevant, and significant loss records. Reviews of current engineering publications and the conduct of objective-focused site inspections provide the AIR team with a capacity for

Attachment B: Model Evaluation and Curriculum Vitae of Joseph Minor, Ph.D., P.E. effective application of expert opinion. This approach is theoretically sound when considerations of historical data and expert opinion are carefully balanced and the vulnerability functions are cautiously calibrated and independently validated.

C. Any modification factors/functions to the vulnerability functions or structural characteristics and their corresponding effects shall be clearly defined and be theoretically sound.

Two modification factors applied to vulnerability functions pertain to building code effectiveness and to mitigation measures. Loss data from recent hurricanes allow empirical assessments of these two factors. These data provide confidence in the modification factors employed in the model. Procedures leading to the derivation of these factors are theoretically sound.

D. Construction type and construction characteristics shall be used in the derivation and application of vulnerability functions.

The AIR Hurricane Model includes construction type and construction characteristics in vulnerability functions for 20 different residential classifications.

E. In the derivation and application of vulnerability functions, assumptions concerning building code revisions and building code enforcement shall be reasonable and be theoretically sound.

Assessments of loss data have permitted a refinement of modification factors to vulnerability functions that address building code revisions and building code enforcement. Previously the model addressed building code revisions and, implicitly, building code enforcement in four regions in Florida. This approach has been retained for buildings built before enactment of the 2001 Florida Building Code. For buildings built to requirements of the 2001 Florida Building Code six separate vulnerability functions have been developed. The model does not explicitly consider building code enforcement. These vulnerability functions recognize the unique characteristics of the design wind speeds, terrain exposure, and windborne debris as defined in the 2001 Florida Building Code. These refinements to the vulnerability functions are reasonable and theoretically sound.

F. Vulnerability functions shall be separately derived for building structures, mobile homes, appurtenant structures, contents, and additional living expense.

The model computes, separately, damages for building structures, mobile homes, appurtenant structures, contents, and additional living expenses.

G. The minimum wind speed that generates damage shall be reasonable.

The model considers damage that may occur at hurricane related wind speeds as low as 40 mph. Use of vulnerability functions that include this wind speed assists in the assessment of loss caused by slow moving storms.

Attachment B: Model Evaluation and Curriculum Vitae of Joseph Minor, Ph.D., P.E.

Disclosures

1. Provide a flow chart documenting the process by which the vulnerability functions are derived and implemented.

The flow chart provided in the AIR submission (Figure 1) illustrates the interaction between historical data and expert opinion that form a basis for the derivation of vulnerability functions. The roles of historical data in forming the vulnerability functions, calibrating them, and validating them are clearly outlined.

Insurance claims data used in the development of vulnerability functions have been provided by several AIR clients with insured properties both within and outside of Florida. A wide range of variability in the data reflects differences in client company records and methods of reporting. However, there is a sufficient quantity of data to permit calibration of the vulnerability functions as well as execution of independent (“blind”) validations. The quality of client supplied data is improving with new loss information from recent storms. These new data are allowing assessments of building code changes and mitigation measures, as well.

3. Summarize site inspections, including the source, and a brief description of the resulting use of these data in development, validation, or verification of vulnerability functions.

Site inspections conducted in recent years by AIR field teams have significantly enhanced the expert opinion component of the vulnerability function derivation process. AIR researchers have developed an excellent grasp of building failure mechanisms and the importance of engineering input to the performance of buildings in wind storms. Site inspections by AIR are objective focused, systematic, and statistically based.

4. Describe the research used in the development of the model’s vulnerability functions.

The library of research publications and client-supplied data employed by AIR in the derivation and refinement of vulnerability functions are both extensive and current. In addition, members of the staff attend conferences and seminars where the effects of wind storms on buildings are discussed and analyzed. Research activities also include field investigations and extracting the maximum amount of information from loss records. The relating of claims information with dates and locations of building code change and mitigation measures is an example of this type of research.

Attachment B: Model Evaluation and Curriculum Vitae of Joseph Minor, Ph.D., P.E.

The 20 basic construction types that characterize residential, apartments and condominiums, and manufactured housing (mobile homes) provide a good basis for analysis. The failure mechanisms described for these construction types are reasonable characterizations of failure modes in hurricane winds. The roles of engineering attention in defining structural performance of different construction types in wind storms are properly described. Further, it is understood that these roles are changing as new building codes force more engineering attention upon the construction of residences, apartments, and condominiums.

6. Identify the one-minute average sustained wind speed at which the model begins to estimate damage.

The used of a 1-minute average, 40 mph wind speed in both slow and fast moving storms captures important aspects of hurricane-winds damaging phenomena.

7. Describe how the duration of wind speeds at a particular location over the life of a hurricane is considered.

The process wherein cumulative damage is calculated at specific points for all time increments when the wind speeds are above 40 mph (1-minute average) is effective in capturing the effects of wind speed duration over the life of a hurricane. More time increments are involved when storms are slow moving; hence, more damage accumulates within each wind speed range as loss computations are made.

8. Provide a completed Form V-1, One Hypothetical Event.

The completed Form V-1 has been reviewed. Comments are recorded below.

V-2 Mitigation Measures

• Roof strength • Roof covering performance • Roof-to-wall strength • Wall-to-floor-to-foundation strength • Opening protection • Window, door, and skylight strength.

Attachment B: Model Evaluation and Curriculum Vitae of Joseph Minor, Ph.D., P.E.

A separate risk analysis module was developed by AIR to address the effects of mitigation measures. This module is necessarily based on expert opinion at this time because of the limited amount of historical data that apply specifically to 22 the building features that have been identified as having a significant impact on building losses. The module supports any combination of multiple building features and produces appropriate modifications to the vulnerability functions. Modifications to vulnerability functions vary with wind speed to reflect the relative effectiveness of a building feature when subjected to different wind speeds. The building vulnerability component of the AIR model explicitly addresses new construction built in accordance with the Florida Building Code, 2001.

B. Application of mitigation measures shall be reasonable both individually and in combination.

Reviews of percentage changes in losses as related to each mitigation measure indicate that the risk analysis module captures changes to building vulnerability when certain building features are present. Results of projections using multiple measures appear reasonable. The relatively large percent changes in losses for some mitigation measures at wind speeds comparable to design wind speeds are particularly revealing.

Disclosures

1. Provide a completed Form V-2, Mitigation Measures – Range of Changes in Damage

The ranges of changes in damage presented in the completed Form V-2 have been reviewed.

2. Provide a description of the mitigation measures used by the model that are not listed in Form V-2.

The mitigation measures listed in Table 2 are comprehensive and reasonable reflections of factors that affect building performance in windstorms.

Form V-1: One Hypothetical Event

Attachment B: Model Evaluation and Curriculum Vitae of Joseph Minor, Ph.D., P.E. completing Part A, Estimated Damage for each individual wind speed range is the sum of loss to all buildings in the ZIP Codes subjected to that individual wind speed range. Subject Exposure is all exposures in the ZIP Codes subjected to that individual wind speed range. For completing Part B, Estimated Damage is the sum of the loss to all buildings of a specific type (wood frame, masonry, or mobile home) in all of the wind speed ranges. Subject Exposure is all exposures of that specific type in all of the ZIP Codes.

One base structure for each of the construction types should be placed at the population center of the ZIP Codes.

The damage ratios presented by AIR as Part A and Part B reflect the effects of storm speed and appear reasonable, both in magnitude and in the range of damage ratios for fast and slow moving storms.

The AIR response to this request provides justifications of the assumptions that were necessary in order to apply their vulnerability functions to the defined problem.

Attachment B: Model Evaluation and Curriculum Vitae of Joseph Minor, Ph.D., P.E.

C. Provide a plot of the Form V-1 Part A data.

The plot of the Form V-1 Part A data presented in Figure 8 reflect clearly the effects of storm speed and the range of expected damage ratios. Differences are almost double at low wind speeds and are significant at higher wind speeds. These results are consistent with reviewer observations and experience.

Form V-2: Mitigation Measures – Range of Changes in Damage

The percent changes in damage for the various mitigation measures presented on Form V-2 are both reasonable and significant. Variations in these percentages as a function of wind speed further characterize mitigation effects and provide a basis for assessing their significance. Modification factors presented in Table 3 are also reasonable and have significant implications.

There were no further assumptions in the completion of Form V-2.

Reviewer did not have access to the CD ROM. Reviewer assessments are based on the hard copy of Form V-2 included in the submission.

Attachment B: Model Evaluation and Curriculum Vitae of Joseph Minor, Ph.D., P.E.

Standard glass windows No door covers No door covers No skylight covers No skylight covers Constructed in 1980 Constructed in 1980 Mitigated Frame Structure: Mitigated Masonry Structure: Rated shingles (110mph) Rated shingles (110mph) 8d nails, deck to roof members 8d nails, deck to roof members Truss straps at roof Truss straps at roof Plywood Shutters Plywood Shutters Place the reference structure at the population centroid for ZIP Code 33921 located in Lee County

Table 3 contains a wealth of information that relates modification factors (to vulnerability functions) and construction features to wind speed. The five-windspeed format is informative in that trends can be revealing. Because of the extensive nature of the list of construction features it is difficult to make general assessments. However, it is revealing that some of the major features reflect significant changes to modification factors at higher wind speeds while more minor features have significance at lower or medium wind speeds. The magnitudes and trends in the modification factors appear reasonable.

Form V-3: Mitigation Measures – Mean Damage Ratio Trade Secret List Item

C. Provide a graphical representation of the vulnerability curves for the reference structure and the fully mitigated structure.

Attachment B: Model Evaluation and Curriculum Vitae of Joseph Minor, Ph.D., P.E.

No shutters Standard glass windows Standard glass windows No door covers No door covers No skylight covers No skylight covers Constructed in 1980 Constructed in 1980

The presentation is effective in its descriptions of mitigation factors as a function of wind speed. It may be concluded that the effects of storms with wind speeds approaching or exceeding code specified values can be mitigated to significant degrees with selected mitigation measures employed singly or in combination. The reported effects of mitigation measures appear reasonable and relate well to reviewer experiences.

CONCLUSION

The information provided by AIR in response to Sections V-1 (Derivation of Vulnerability Functions), V-2 (Mitigation Measures), and V-3 (Mitigation Measures – Mean Damage Ratio Trade Secret List Item) of the Florida Commission standards is thorough and complete. Components of the AIR Hurricane Model that define building vulnerability and project loss reflect a balanced blending of historical data and expert opinion. The vulnerability functions, loss projections, and assessments of mitigation measures represent reasonable assessments of building behavior in hurricanes.

Joseph E. Minor, P.E. February 23, 2007

Attachment B: Model Evaluation and Curriculum Vitae of Joseph Minor, Ph.D., P.E.

JOSEPH E. MINOR, P.E.

Consulting Engineer P.O. Box 603, Rockport, Texas 78381 Telephone and Facsimile (361) 729-0656 e-mail [email protected]

Dr. Minor is recognized internationally in the fields of wind engineering, window glass design practice, and natural hazards research. He has supervised investigations into more than sixty damaging windstorm events and has directed the utilization of information gained in a wide ranging program of research and engineering practice. Special areas of expertise include wind-structure interaction phenomena, effects of tornadoes and hurricanes on buildings, performance of window glass and curtain wall systems, building code provisions for wind loads, the economics of wind resistant construction, and the impact of natural hazards on socio-economic systems.

A special expertise has been developed in the science and engineering of window glass. Dr. Minor has conducted an extensive program of research, testing, and applications activities involving window glass behavior under the effects of wind pressures, blast loading, and impacts from windborne debris and hailstones. He has experience with stress analysis of glass plates and the testing of window glass in pressure, blast and impact environments.

Dr. Minor is active on building code committees, industrial advisory boards, and professional society committees, and serves as a consultant to governmental agencies, trade associations, and private organizations. He has experience as an expert witness in legal matters. He has lectured nationally and internationally on topics related to the integration of wind engineering research into professional practice, and participates regularly in short courses and seminars related to the practice of wind engineering and window glass design practice. He was appointed a Fulbright Senior Scholar in Australia in 1978 where he conducted an assessment of research and practice in Australian natural hazards management. He served five years as Chairman of the Department of Civil Engineering at the University of Missouri-Rolla. He served on the Windstorm Building Code Advisory Committee to the Texas Department of Insurance, administrator of the Texas Windstorm Insurance Association’s Building Code for Windstorm Resistant Construction. Currently he is a visiting professor at Texas A&M University - Kingsville.

EDUCATION: B.S. (Civil Engineering), Texas A&M University, 1959; M. Engin. (Civil Engineering), Texas A&M University, 1960; Ph.D. (Civil Engineering), Texas Tech University, 1974.

PROFESSIONAL CHRONOLOGY: Senior Engineer Assistant, Texas Highway Depart- ment (1959); Graduate Student and National Defense Fellow, Texas A&M University (1959-60); Company Grade Officer, U.S. Army Corps of Engineers (1960-62); Senior Research Engineer, Southwest Research Institute (1962-69); P.W. Horn Professor and Director of Institute for Disaster Research and Glass Research and Testing Laboratory, Texas Tech University (1969-1988); Thomas H. Reese Professor (1988-95), Chairman, Department of Civil Engineering (1988-1993), and Research Professor, Graduate Center for Materials Research (1995- ), University of Missouri-Rolla; Consulting Engineer in Private Practice (1995- ); Visiting Professor, Texas A&M University – Kingsville (2003- ).

Attachment B: Model Evaluation and Curriculum Vitae of Joseph Minor, Ph.D., P.E.

MEMBERSHIPS: American Society of Civil Engineers (Fellow), National Society of Professional Engineers (Member), ASTM, Inc. (Member), NCSBCS, Inc. (Member), Institute for Business and Home Safety (Associate), Licensed Professional Engineer (Texas; Missouri; Florida).

PUBLICATIONS: Refereed Journals and chapters in books – 65 Refereed conference Proceedings – 60; Technical reports and other publications – 100+

HONORS: Phi Eta Sigma, Tau Beta Pi, Phi Kappa Phi, Chi Epsilon (Honor Member, Texas Tech Chapter); Distinguished Service Award, Defense Electric Power Administration, 1977; Fulbright Senior Scholar, Australia, 1978; Halliburton Excellence in Research Award, Texas Tech Univ., 1983; Collegiate Research Award (Engineering), Texas Tech Univ., 1983; Paul Whitfield Horn Professor, Texas Tech University, 1984; Halliburton Lecturer, Texas Tech Univ., 1986; Distinguished Visiting Lecturer, New Mexico State University, 1987; Thomas H. Reese Professor, University of Missouri-Rolla, 1988; Distinguished Engineer, College of Engineering, Texas Tech University, 1989; Academy of Civil Engineers, Texas Tech University, 1989; Distinguished Service Award, National Hurricane Conference, 1999; Award of Honor, Texas Section, ASCE, 2003.

SERVICE: V-P for Technical Affairs, Texas Section, ASCE, 1980-81; Publications Sec., Committee on Wind Effects, ASCE, 1982-84; President, Texas Section, ASCE, 1984-85; President, Texas Tech Chapter, Sigma Xi, 1985-86; Faculty Advisor, Texas Beta, Tau Beta Pi, 1983-86; Sec.-Treas., Wind Engineering Research Council, 1985-87; President, Rolla Chapter, MSPE, 1991-92; President, Insulating Glass Certification Council, 1985-88; Chairman, Wind Committee for HAZUS, FEMA/NIBS, 1998- ; Member, Advisory Committee on Residential Building Code, Texas Department of Insurance, 1999- .

PERSONAL: Age: 65 (b. Corpus Christi, TX); Wife: Ann E. Minor (b. Hamilton, TX); Son: Joseph E. Minor, Jr. (Ph.D. Caltech); Daughter: S. Diane Calvin (M.S. Duke Univ.); three grandchildren; excellent health, veteran, fisherman.

SELECTED PUBLICATIONS ON HURRICANES

Minor, J.E., “Formal Engineering of Residential Buildings,” Journal of Architectural Engineering, ASCE, Volume 8, No. 2 (June 2002), 55-59.

Minor, J.E., "Windborne Debris and the Building Envelope," Journal of Wind Engineering and Industrial Aerodynamics, 53 (1994) 207-227.

Minor, J.E., "Construction Tradition for Housing Determines Disaster Potential from Severe Tropical Cyclones," Journal of Wind Engineering and Industrial Aerodynamics, 1983, Vol. 14, pp. 55-66.

Minor, J.E., "There's No Way to Stop Debris in a Hurricane," Civil Engineering, Vol. 54, No. 11, November 1984, p. 20.

Behr, R. A. and Minor, J. E., "A Survey of Glazing System Behavior in Multi-Story Buildings During Hurricane Andrew," The Structural Design of Tall Buildings, Vol. 3, 1994, pp. 143-161.

JEM 02/20/04

Attachment C: Model Evaluation and Curriculum Vitae of Narges Pourghasemi, M.S.

A Peer Review of Model 21 Implementation within CLASIC/2TM

Narges Pourghasemi – Independent Auditor February 5, 2007

Attachment C: Model Evaluation and Curriculum Vitae of Narges Pourghasemi, M.S.

Overview

This document summarizes the peer review I conducted for AIR Worldwide Corporation’s implementation of their Atlantic Tropical Cyclone Model within the CLASIC/2 software. The intent of the review is to ensure that the software complies with the software standards and requirements established by the Florida Commission on Hurricane Loss Projection Methodology, as well as current industry-standard software engineering practices. This review covers AIR Worldwide’s Demand Surge development process and its output range for compliance with the new standard.

Findings of the Peer Review

The findings are grouped according to the seven computer standards established by the Florida Commission on Hurricane Loss Projection Methodology as of Fall 2006. During the peer review, I interviewed the Modeler personnel or their proxies from Research and Modeling Group, Software Development Group, Consulting and Software Services Group, Quality Assurance, Information Services group and Product Managers, who demonstrated their knowledge of the pertinent subjects.

C-1 Documentation

AIR Worldwide’s documentation for the Atlantic Tropical Cyclone Model (internally referred to as Model 21 hurricane model) is grouped as follows:

1.1 – Primary Document Binder AIR Worldwide has created binders that document CLASIC/2 and its software engineering practices. A separate binder contains a master table of contents that lists all the documents in binders C-1 through C-7. I have reviewed the C-1 though C-7 Primary Document Binders as related to the implementation of Atlantic Tropical Cyclone Model within CLASIC/2. These binders include complete user and consistent computer software documentation relevant to the submission. A Document Control Information page is provided in each document that details the author, reviewer, and the revision history.

These documents are created separately from the source code and are located and easily accessible in the Software Development group office space.

1.2– User Documentation AIR Worldwide’s user documentation includes: • CLASIC/2 Installation and System Configuration • CLASIC/2 Update Instructions • CLASIC/2 Release Notes • CLASIC/2 User’s Guide explains how to use the software and its utilities.

Attachment C: Model Evaluation and Curriculum Vitae of Narges Pourghasemi, M.S.

o AIR ImportExpress for CLASIC/2 User’s o UNICEDE/px Data Exchange Format Preparer’s Guide (for Primary Insurers) o UNICEDE/fx Data Exchange Format Preparer’s Guide (for Facultative Insurers) o Managing CLASIC/2 Database with AIRDBAdmin User’s Guide for database maintenance utility • CLASIC/2 Reference Guide

1.3– Technical Documentation The AIR Tropical Cyclone Model for the U.S. Gulf and East Coasts document presents a comprehensive overview of the hurricane model.

AIR Worldwide maintains internal technical documentation in the following four major categories: Software Internals Databases and Data Files Coding and Documentation Guidelines Test Documents

1.3.1 – Software Internals For each major CLASIC/2 component, there is a technical document providing a technical overview and defining the components, flow diagrams, and relevant classes or source file names.

1.3.2 - Databases and Data Files For each database or data file used in CLASIC/2, there is a technical document discussing the following aspects of the data: requirements, overall description, sample information, construction, validation, and relation to other data.

1.3.3 - Test Documents For each major software component, there is a technical document discussing what must be tested and how the test is implemented.

Attachment C: Model Evaluation and Curriculum Vitae of Narges Pourghasemi, M.S.

Standard C-2 – Requirements

The CLASIC/2 requirements in the Primary Document Binder address the following major areas: • Functional Requirements • System Requirements • External Interface Requirements • CLASIC/2 Database Requirements • Insurer and Re-insurer User Model

The HTML-based online documentation provides the CLASIC/2 software requirements. The C-2 Primary Document Binder also contains a document that defines the changes from the existing version of CLASIC/2 (Version 8.5) to the spring 2007 release of CLASIC/2 (Version 9.0). The requirement and specifications are tracking tools as well.

2.1 Functional Requirements The Primary Document Binder contains the AIR Tropical Cyclone Model for the U.S. Gulf and East Coasts document, which explains the purpose and formulation of the hurricane model.

This Primary Document Binder provides requirements for US Hurricane Catalog, Physical Properties, ZIP All software code and data files used during model development. The data is the input to the Hurricane model. The requirements specify the format of the loss estimate output data files and summarize the business rules and processes that are applied to create the output data.

CLASIC/2 is constructed in a two-stage process. In the initial stage, scientists with hurricane modeling expertise implement software that performs hurricane simulations and damage calculations. At this stage, the emphasis is hurricane model scientific accuracy in damage cost calculation. Upon completion of this task, the scientists release the hurricane modeling software and simulation data to the software development team.

The hurricane model software and simulation data serve as functional specifications for hurricane simulation and damage cost calculation in CLASIC/2. The software development team is responsible for using the latest software technology to re-implement the hurricane model into CLASIC/2 with emphasis on maintainability and performance. The software development team incorporates additional functionality into CLASIC/2 that enables users to perform hurricane insurance and re-insurance risk calculations.

2.2 Insurer and Re-insurer User Model The Primary Document Binder describes the Contract Logic Model, which allows corporate customers, primary insurers, and re-insurers to enter insurance contract exposures and calculate risk potential due to hurricanes. Multiple methods for entering insurance contract exposures and presenting risk potential outputs via CLASIC/2 are specified by either defining a new set of data or through new options in the CLASIC/2

Attachment C: Model Evaluation and Curriculum Vitae of Narges Pourghasemi, M.S.

User Interface. Also, the major features for the spring 2007 release of CLASIC/2 are included.

2.3 External Interface Requirements This section covers various CLASIC/2 software requirements for interfacing with users and other systems:

2.3.1 User Interface Requirements The document CLASIC/2 User Interface – System Requirements and Specifications specifies the user interface requirements for Enterprise Detail, Business Unit Details, Book Details, Company Details, and Policy Details screens. The document specifies the User Interface components on the screen and identifies the screens to be implemented. For each screen, the guide describes the data fields, the supported user actions, and the expected responses.

2.3.2 Bulk Insurance Policy Data Transfer Requirements The UNICEDE/px and UNICEDE/fx Preparer’s Guides, which are used by corporate customers, primary insurers and re-insurers, provide detailed information on the data format for bulk data input and output. The former spreadsheet-based tool operated on fixed format. AIR Worldwide implemented a tool that allows flexibility in data formatting and increases data validation capability through selection menus or options.

2.4 System Requirements General hardware, software, and system requirements, including data sources, analysis engine and security for running CLASIC/2, have been specified.

2.5 Database Requirements The CLASIC/2 AIR Area Code, Exposure, and Loss database requirements are specified, including a summary of the data in each database and the functions (procedures) to be defined on each database.

Standard C-3 – Model Architecture and Component Design

The CLASIC/2 software development team uses Microsoft Component Object Model (COM) and object oriented paradigm industry standards for its system design. The technical documentation includes architectural diagrams, model frameworks, class definitions, and flow charts. In particular, the Model 21 Updated Classes document identifies the classes and structures used to implement Model 21 in CLASIC/2. The Analysis Sequence Diagram illustrates how objects of various classes work together to translate a user’s request to actual computation of the exposure risk. The Primary Document Binder provides documentation that details the technical construction of CLASIC/2, including the software components and database.

This binder includes the following sections:

Attachment C: Model Evaluation and Curriculum Vitae of Narges Pourghasemi, M.S.

3.1 Event generation flowchart The flowcharts describe how to calculate the potential loss for each policy and illustrate the generation of catalogs for different events.

3.2 Business Overview This document overviews the CLASIC/2 configuration, application, and components from an industry perspective. It includes a business case using CLASIC/2 to perform a catastrophe loss analysis for a fictional company.

3.3 Architecture overview The documentation in this section describes the CLASIC/2 layers, which is necessary to understand the relationship between the model engine and model framework. The CLASIC/2 logical subsystem provides a flowchart of the CLASIC/2 architecture, including the presentation layer, business layer, database access layer, and common service layer.

3.3.1 Presentation layer This document explains the components of the presentation layer and how the user interacts with the CLASIC/2 software via these components. The CLASVIEW interface component is an automated and interactive tool used to generate reports, including maps and charts, for different events. The information used to populate the reports is extracted from various CLASIC/2 databases.

3.3.2 Business Service layer: This layer is a collection of in-process and out-of-process [D]COM servers and auxiliary DLLs are responsible for: data services, analyzing catastrophe risk for locations and policies using a combination of licensed perils, and saving the results. This document explains the components used to schedule, queue and recover jobs in a multi-server physical machine with multi-thread processing. The data management components, which monitor input/output data and the exchange of company Loss Files and Analysis Loss Files, are documented. The Software Development team’s engineers also explained the various components.

The AIR Geocoder, which is a proprietary utility, determines the latitude/longitude coordinates for each contract location’s street address. The standard census access data (TIGER), which is obtained annually, is used for table look up within the database. Missing address information is validated by the AIR Geocoder and recorded in a log file.

3.3.3 Common service layer This document identifies the framework class libraries, data access component, and configuration files that interact with various CLASIC/2 layers.

Attachment C: Model Evaluation and Curriculum Vitae of Narges Pourghasemi, M.S.

3.3.4 Database Access Layer: This section provides a reference to the CLASIC/2 database objects (tables, views, and stored procedures) and the underlying relationship between the components of the databases.

AIR Worldwide implemented a browser and HTML-based documentation system for the CLASIC/2 engine and model. This system provides technical staff access to class diagrams, software components, software comments, and source code. Software developers use links to view the logical components or class hierarchy, source comments concerning a class or its individual methods, and the actual source code.

Standard C-4 – Implementation

Engineers from the Research and Modeling group and Software Development team explained the interface and coupling of the model and engine. AIR Worldwide is currently documenting how the modeling software and conversion of simulation data is processed and integrated into CLASIC/2.

4.1 – Coding and Documentation Guidelines Regarding the CLASIC/2 Model 21 implementation, AIR Worldwide has modularized the implementation into reasonably sized classes. C++, the language chosen for implementing the project, provides ample support for implementing the CLASIC/2’s object design. Class templates make the code more concise due to code re-use. However, templates also make the code more abstract and thus require more time to understand. The software development team has created sufficient documentation on the internals of CLASIC/2, making it possible for an experienced C++ programmer to understand the source code. The documentation provides flowcharts that describe how the various classes and the functions interact in the Model. These documents are available in the Primary Document Binder.

The hurricane simulation (data) software is written in FORTRAN. Through the use of user-defined data types, the software development team takes advantage of the object- oriented software technology benefits.

AIR Worldwide maintains the following technical documentation and coding standards documents:

Technical Documentation Guidelines C#/.NET Coding Guidelines C++/COM Coding Guidelines FORTRAN Coding Guidelines

These documents provide guidelines for useful software practices to achieve code that is effective, well documented, and maintainable. AIR Worldwide developers follow these guidelines closely.

Attachment C: Model Evaluation and Curriculum Vitae of Narges Pourghasemi, M.S.

4.2 Line Count

AIR Worldwide has developed a tool which is used on the directories corresponding to the CLASIC/2 Engine components, Model 21 component, and the shared model framework that is included as part of the Model 21 component.

4.3 Source Control Management AIR Worldwide software developers use Microsoft Visual Studio.NET, a de-facto industry-standard development platform, to develop, debug, and maintain the model, CLASIC/2 software, and small size data files, as well as for storing requirements and design documentation.

4.4 Traceability AIR Worldwide documentation provides fully traceable component identification using diagrams. In the Primary Document Binder there is a table that contains each of the component names, the number of lines of code, and the number of lines of comments.

4.5 Building CLASIC/2 Final Builder provides the automated process that is used for internal and release builds of CLASIC/2. In some circumstances, it is necessary to manually build CLASIC/2. Both processes are documented.

Standard C-5 – Verification

The CLASIC/2 software development team uses industry-standard methods for ensuring correct operation of the software. Flag Triggered Output Statements is a built-in development flag that enables/disables tracing of the model’s input/output values. The C++ ASSERT() directive has been used extensively in various parts of the code for validation of runtime assumptions related to business logic, intermediate results, component initialization, and execution workflow. Care has been taken to handle exceptions, error returns, bound checking, null pointer testing, and memory de-allocation. A test case methodology applies on the engineering model to validate the result of this model and against the model developed by the Research and Model group.

The CLASIC/2 software team uses AIRTask, an internally developed, web-based issue tracking software, to communicate bug reports among software developers and quality assurance engineers. AIRTasks is also used to report requirement and documentation issues.

The overall testing, including User Interface and integration testing, is performed using Borland SilkTest (version 8) automated testing software and MUTS (Model Unit Testing System). AIR Worldwide uses Query Analyzer, part of the SQL Enterprise manager, to verify and view SQL information and to retrieve data. Beyond Compare is used to compare results and changes in scripts, etc.

The document C-5 binder provides comprehensive testing documentation.

Attachment C: Model Evaluation and Curriculum Vitae of Narges Pourghasemi, M.S.

5.1 Unit Testing For CLASIC/2, the Quality Assurance team has established a set of unit tests for critical functions that play a vital role in calculating hurricane loss distributions.

The Quality Assurance team analyzes these functions for their proper minimum and maximum input values. The team also instruments the code to make these functions accessible to the unit-testing program and cause these functions to log the output data when the unit-testing flag is turned on.

When a quality assurance engineer runs the unit-test program, the engineer may specify the following types of test inputs for each function:

• Custom: the engineer may specify a specific input data set for testing the function.

• Random: the unit-test program generates a random data set for testing the function. As part of the data generation, the program generates a certain amount of out-of- bound data. The purpose of the out-of-bound data is to determine if the function handles erroneous inputs correctly.

• Linear the unit-test program generates a data set by varying each input parameter by a fixed increment within the range of [90% of the minimum, 110% of the maximum]. All possible combinations of the input parameter values are recorded. The test results are intended for trend analysis and boundary checking analysis.

For each test, the input and output data for each function is recorded in a file and the test data is analyzed for proper behavior.

5.2 Integration Testing The mechanism implemented for unit testing can be easily adopted for integration testing. After verifying that functions at certain levels behave correctly, the quality assurance team proceeds to verify the correctness of other functions that invoke only the validated functions. The following is the procedure for validating CLASIC/2.

Before providing the hurricane modeling software to the CLASIC/2 software development team, the hurricane research team validates the hurricane model using theories, visual tools such as MapInfo and ArcGIS, and meteorological science and field observations of actual hurricane damages.

As the CLASIC/2 software development team completes the re-implementation of the hurricane model in C++, the quality assurance team runs specific tests using the research team hurricane model. For each test, the quality assurance team compares the result with the result obtained by running the test using the CLASIC/2 software. If a discrepancy is detected between the two results, the quality assurance team refers the error to the software development team for resolution.

Attachment C: Model Evaluation and Curriculum Vitae of Narges Pourghasemi, M.S.

5.3 Aggregate Testing To ensure broad coverage in CLASIC/2 software testing, the quality assurance team performs certain aggregate tests. Some examples include:

• Average loss for a house of certain construction type and value is computed for all ZIP Codes in the United States.

• Losses for a group of selected locations, building construction types, and four occupancy classes are computed and compared. Effects of deductibles are also computed and compared.

• For certain aggregate tests, the quality assurance team run tests using the previous year’s CLASIC/2 software. For each test, the quality assurance team compares the results with the results obtained by running the current CLASIC/2 software. If a discrepancy is detected between the two results, the quality assurance team refers the case to the research team for further analysis.

5.4 UI Testing AIR Worldwide has defined general guidelines for testing the CLASIC/2 User Interface. The document provides a detailed test plan, using SILKTEST, for each of the major functions supported by the CLASIC/2 User Interface.

Standard C-6 - Model Maintenance and Revision

The CLASIC/2 software development team uses Microsoft Visual SourceSafe (VSS) to track software and documentation change history. The AIR Data Control Workbook is used to log model data that is ready for transfer from the Research and Model group to the Software Development team. The checksum scheme is used to ensure data file and database integrity. All revision enhancements are performed using the VSS locking mechanism and are logged and managed using AIRTasks.

6.1 Source Code, data and Documentation Maintenance Code Both Research and Software Development teams use VSS source control software to track source file changes. VSS maintains the source code revision history and provides the means to undo changes if necessary. The files related to CLASIC/2 documentation are also maintained in VSS.

As described above, the quality assurance team runs certain aggregate tests using current CLASIC/2 software and the previous year’s CLASIC/2 software. The purpose of these tests is to detect any unexpected change in the implementation of the underlying model.

All the requirements, design documents and code, and smaller data files for Model 21 hurricane model are stored in VSS.

Attachment C: Model Evaluation and Curriculum Vitae of Narges Pourghasemi, M.S.

Data The larger sized data files are stored in the file system for which the version history and changes are maintained manually (AIR Data Control Workbook) by tracking date, file name, data type, description, changes relative to previous year, and the name of the person(s) who send and receive the data.

As model data generated by the Research and Modeling group is ready for transfer to the Software Development team, it is logged in the AIR Data Control Workbook document.

AIR Worldwide continually reviews model assumptions in regards to new meteorological and other information and periodically enhances the model. An updated model is identified with a new version number. The AIR Worldwide hurricane model is subject to compliance with the standards established by the Florida Commission on Hurricane Loss Projection Methodology and only this version of the model approved by the Commission will be run for client studies user for Florida rate filings.

Hurricane Data Model File Integrity

Model 21 hurricane model contains multiple data files related to hurricane simulation. Some of these files are quite large, over 100 MB. To prevent overburdening the system, these files are not stored inside VSS. To ensure the integrity and maintain a change history of these important files, AIR Worldwide has instituted the following policy:

Whenever the hurricane modeling group hands over a new set of hurricane modeling data to the software development team, a MD5 and CRC32 digest will be generated for each modeling data file. For each modeling data file, the following information will be stored inside Visual Source Safe: the pathname, the digest files, and the last access date. Each time a developer downloads a copy of the modeling data from the file server, that developer always performs the MD5 and CRC32 checksum verification. If the developer encounters any checksum failures, the pathname and the last access date stored in Visual SourceSafe provide sufficient information to retrieve a correct copy from the backup.

Documentation Software and model-related documentation is generated and maintained by each of the groups: • Communications • Research and Modeling • Software Documentation • Software Research and Development • Software Services • Software QA

Attachment C: Model Evaluation and Curriculum Vitae of Narges Pourghasemi, M.S.

6.2 Tracking The AIRTask software, developed for internal use, is used to track bug reports, general issues, and bug fixes. The issues originated from one person/group are sent by the system to notify the responsible person(s)/group(s) for follow-up.

The web-based tool AirForce, a product of salesforce.com, is used for customer relationship management to log customer communications and interactions on all software issues. The clients can also review issues, news, or hot fixes through the corporate website and logging mechanism.

6.3 Model Revision Policy A major release of CLASIC/2 occurs annually. A major release incorporates new hurricane data from the previous year, improvement in the hurricane model, new user functionality, and possibly new software architecture. Between major releases, AIR Worldwide releases a minor version containing software bug fixes and user functionality enhancements. These software bug fixes and user functionality enhancements do not affect the filed loss cost. The CLASIC/2 software and model versioning methodology is outlined in the Primary Document Binder.

Standard C-7 – Security

AIR Worldwide has instituted multiple security measures to ensure the safety and integrity of its software.

7.1 Access Control AIR Worldwide has implemented access control at both the physical level and at the software level. All employees are required to take a security course.

7.1.1 Physical Access Control AIR Worldwide’s office is housed in a building staffed with security guards 24 hours each day. In order to access AIR Worldwide’s office without an escort, a person has to present a security badge with picture ID and proper electronic encoding. File servers in the office are housed in a separate room. Only employees with special authorization are allowed to enter the room.

7.1.2 Software and Data Access Control Access to AIR Worldwide’s software resources is safeguarded by firewall, internal Network logon (Windows 2000 authentication), external Network logon (remote access server), virtual private network (VPN), and secure SSH File Transfer Protocol (SFTP). The staff enforces this security measure by signing security sheets and changing passwords monthly. Access to the files and folders on AIR Worldwide’s data servers is regulated by

Attachment C: Model Evaluation and Curriculum Vitae of Narges Pourghasemi, M.S. read/write permission. Access to the contents of AIR Worldwide’s database is limited to authorized accounts.

7. 2 Software Back-up Modified files in the network file servers are incrementally backed up daily. A full back up of all the files in the network file servers is performed weekly. At the end of each month, all the incremental and full back ups are stored in an off-site storage facility.

7. 3 Virus Scan AIR Worldwide uses products from Trend Micro to scan incoming emails, email attachments, and any files introduced through external media (e.g. floppy disks or CDs) and covers all servers and workstations. The virus protection is automatically downloaded daily.

AIR Worldwide performs a full virus scan of all the files in the network weekly.

Conclusion

It is my opinion that AIR Worldwide’s implementation of Model 21 within CLASIC/2 is in compliance with the computer standards established by the Florida Commission on Hurricane Loss Projection Methodology. It is also my opinion that the software engineering practices at AIR Worldwide are in accordance with current software industry standards.

Attachment C: Model Evaluation and Curriculum Vitae of Narges Pourghasemi, M.S. Narges Pourghasemi ______P.O. Box 994 Brookline, MA 02446 . 617-489-9552 . [email protected]

PROFILE A versatile and result-oriented Senior Software Engineer with expertise in System Analysis and Product Enhancement. Over 12 years of consistently proven success designing and developing Software, GUIs, E-commerce, web applications, Client/Server applications, Virtual Reality and Distributive systems.

Equally effective working alone or in teams. Excellent Customer-Interface skills. Distinguished academic and teaching experience. Proven ability to quickly grasp new challenges, master new technologies and achieve tangible results. US Citizen.

Technical Proficiencies Languages Java, Java Swing, C/C++, JavaScript, Java Servlets, JSP, ASP, HTML, XML/XSL, Velocity, PERL, X11/Motif, OpenGL, LISP, Assembly, Shell Scripts, awk, sid. Operating UNIX, ksh/bsh/csh, LINUX ( Red Hat 9, SUSE 9.1), AIX , HP-UX, Windows/XP/NT 4.0 Environment Mac OS X, VMS, VOS, IRIX , Apache Tomcat, JSDK, J2EE. Communications HTTP, X.25, MATIP, EDIFACT, TCP/IP and Sockets. Tools/Utilities Java IDE (Eclipse and IntelliJ) ; GUI Builders (Builder Xcessory & UIM/X); SCM (Rational ClearCase & ClearRequest, Visual SourceSafe, AccuRev, CVS, DSEE); IMSL; Debuggers (xde, dbx, gdb,xdb), Census ( issue tracking software), and graphical tools ( Visio, ac3d , gimp). Databases MySQL, JDBC, Oracle 6.0, SQL, Ingres Database. Hardware Sun Solaris, RISC 6000, HP11, Stratus, SGI, Apollo, VAX 11/780, AMIGA2000, Apple.

EXPERIENCE

Independent Software /Web Developer Consultant 2004 – present • CHARLES RIVER ANALYTICS, INC. , Cambridge, MA present Designed and implemented a Graphical User Interface of an editor for the emulator data exploited in a Context-driven Infospace Configuration application for Air Force Research Lab, using Java Swing. Wrote the User Manual for this tool; and reformatted other tools documentations into the User Manuals. • APPLIED INSURANCE RESEARCH (AIR WORLWIDE CO.) , Boston, MA 2006 IT Audit- Review the software and the development process of the Hurricane Insurance Model to ensure that they comply with the Florida Hurricane Commission standards. • KAYAK , Concord, MA 2005 Assisted in development of Internet Travel Product, for a travel meta-search company, using Java, J2EE/Tomcat /Struts, extensive HTTP Protocols, XML/DOM/SOAP/AXIS/WSDL, Velocity, HTML, MySQL, CVS in UNIX and Mac OS environment. • TAXWARE, Salem, MA. 2004 – 2005 Enhanced and corrected a Tax Management application’s front-end interface and software defects, using X/Motif, Builder Xcessory GUI builder, C on variety of Unix/Linux platforms. • NORTHESTERN UNIVERSITY – Teaching Assistant for Java Course. 2004

SITA (formerly EQUANT Application Services), Burlington, MA. 1999 - 2003 Senior Software Engineer- Advance Travel Solutions- Contributed to various projects for a spectrum of clients in the provision of E-Commerce and On-Net application solutions with focus on global airline and travel. The EDI-Application Gateway is constructed around the industry standard seven-layer model for communication architectures. • Conducted the Message Switch System setup for a $20M contract project with CCRA (Canada Customs and Revenue Agency) to provide advanced passenger information on all passengers traveling on in-bound international flights to Canada. On time completion of this work, with high degree of quality, and in the absence of an accountable engineer, helped contribute to a new $4M target achievement for phase one (Message Switch) of the project. • Web application Development: Enhanced the SITA Flight Finder Web Applications for the AirOne (Italian Airline) reservation, for the Canada3000 Airline reservation and in-house utility applications, and for SITA Air Cargo Carriers. Environment: ASP, JSP, HTML, Java, Java Beans, JavaScript and PERL.

Attachment C: Model Evaluation and Curriculum Vitae of Narges Pourghasemi, M.S. • Enhanced communication interface of the SITA Interline Through Check-In (ITCI) system, Air Faring and E-Ticketing Interfaces, to support EDIFACT interchange. • Improved the message exchange interfaces between Amadeus CRS and SITA host servers using the MATIP (Mapping of Airline Traffic over IP) protocol for Type A (Conversational, Host to Host) and Type B traffics. • Wrote utility programs to allow batch files to work under different operating systems after porting the base model and test suite from Stratus/VOS to UNIX; • Evaluated, customized and administered an Issue Tracking system called Census, for in-house and client’s usage. • Assisted for development of a Gateway system to be used to bridge the Canada3000 Tour Operator sales into SITA’s Gabriel CRS (Computer Reservation Systems) for centralized inventory control. • Initiated porting, the complex source base which included the application and the operating environment, from RISC platform to SUN Solaris. The application section consisted of 36 components (70K SLOC); Conformed the updated Software Environment and System Configuration; • Rapidly grasped knowledge of the application systems, core products and CRS interfaces, and the company proprietary software development and runtime environment, in a short period of time and was appointed to provide full time support to a major client (Carlson Wagonlit Travel). o Provided consultant services and developed constant enhancements to the APIs and CRS Gateways based on customer requests and CRS changes. o Interfaced with Client staff to determine change/enhancement requirements. o Coordinated all work and frequent releases. Provided written release notes with each new software release; ensured on time releases. o Maintained API documentation, and handled writing procedures for support of system and status reports. o Handled Y2K compliance issues. • Served as a consultant; assisted in work estimate, enhancements, and tests for American Express. • Worked on several projects concurrently and communicated with cross functional teams located in Atlanta and UK. Environment: Proprietary Software Environment, C based and Scripting Proprietary Languages, C, NT 4.0, X.25, MATIP, TYPE A/B messages, TCP/IP, SUN Solaris , VOS/Stratus, PERL and MS Visual Studio, Source Safe.

U.F.A. INC., Woburn, MA. 1994 - 1999 Senior Software Engineer; Involved in all phases of software development and GUI Design for simulation and modeling of an air traffic management system on a distributed system. • Responsible for Data Preparation component for the major client, DFS (Germany’s Federal Aviation Administration). Involved in design and development/enhancement of graphical editors and support tools. • Created a library from scattered codes for the graphical display of radar geographical backgrounds. The creation of this library, subsequently used in the Situation Display and all Data Preparation applications, made code enhancement much easier. • Designed and developed tools for data extraction and conversion from large external data sources: Jeppesen Sanderson navigation database and ICAO FPL recorded Flight Plans database. • Designed and developed the Graphical User Interfaces (GUI) for Data Preparation utilities. • Evaluated and used the COTS GUI builder (UIM/X) for prototyping and development.

Attachment C: Model Evaluation and Curriculum Vitae of Narges Pourghasemi, M.S. • Designed and developed Radar Message Filter Editor, an X Window/GUI based tool, for the Raytheon Company. • Consulted in pre-sales activities, scheduling, time/cost estimation, and all Software and User specifications issues.

Environment: C, IBM RISC 6000, UNIX, X11/Motif, UIM/X, ClearCase, DSEE, , TCP/IP, Sockets, SGI, SUN, Apollo. ADDITIONAL EXPERIENCE Software Consultant/Independent Contractor/Academic Researcher & Instructor See Addendum.

EDUCATION NORTHEASTERN UNIVERSITY- Boston, M.A. 2004 -present Ph.D. Candidate in Computer System Engineering, Industrial Eng. Dept. Area of Concentration: Software Engineering, User Interface Design/Human factor (User adaptive interfaces).

GEORGE WASHINGTON UNIVERSITY - Washington, D.C. Professional Degree in Engineering; Computer Science/Artificial intelligence; minor in Medical Engineering.

• UNIVERSITY OF NEW HAMPSHIRE - Durham, N.H. M.S. in Computer Science.

• UNIVERSITY OF TEHRAN - B.S. in Computer Science and Mathematics. • Microsoft Windows Embedded Certificate - Developing Embedded Solutions for Windows XP Embedded.

Attachment C: Model Evaluation and Curriculum Vitae of Narges Pourghasemi, M.S.

ADDENDUM

SOFTWARE CONSULTANT/ INDEPENDENT CONTRACTOR

ICT, Irvine, CA. Installed and setup the Oracle RDBMS Server V6 Software and Workstation Tools for MS-DOS on a distributed System. Set up an in-house application database using the Oracle SQL*forms and SQL*menu, thus converting the Telemagic database into the Oracle RDBMS.

INNOVATIONS INC., Washington, D.C. Designed and implemented interactive animation software in C on Amiga2000 for medical instrumentation use.

CODEWORKS CO., Washington, D.C. Designed, implemented, and tested an architectural design report in C. This major product provided architects with comprehensive design reports for their buildings.

INTELLITEK INC., Rockville, MD. AI programmer. Designed, coded, and tested an object generator interface for an object defined language package. Revised and rewrote object defined language (ODL) software from InterLISP to Common LISP.

ACADEMIC EXPERIENCE

NORTHEASTERN UNIVERSITY, Boston, MA. Teaching Assistant- Dept. of Computer System Engineering- Assisted for Java Programming course. Instructor- Dept. of Computer Technology. Taught Numerical Methods course to undergraduate students.

Project: Wrote a program for animation of one user-controlled vehicle and multiple autonomous vehicles in two lane road with curve. Environment: C++, OPENGL, Window XP. Research Assistant - Dept. of Industrial Engineering. Involved in a research project developed to apply AI technology to the analysis and exploration of very large scientific databases. Environment: C, Silicon Graphics IRIS, IRIX.

GEORGE WASHINGTON UNIVERSITY, Washington, D.C. Instructor- Dept. of Electrical Engineering and Computer Science. Taught graduate and undergraduate courses.

Research Assistant, Implemented and tested a data compression algorithm program for medical images. Projects: • Implemented a program to classify patterns from EKG data for Pattern Recognition course. • Developed a program in FranzLisp to solve elementary mathematical world problem using Conceptual Analysis and Dependency methodology for Natural Language Processing Course.

UNIVERSITY OF NEW HAMPSHIRE, Durham, N.H. Research Assistant. Modified and extended the Graphics Kernel System (GKS) software package to standard form. This software aid the user in programming independent of the internal data structure.

Attachment D: U.S. Hurricane Individual Risk Methodology

U. S. Hurricane Individual Risk Methodology

USIRM_0302

Attachment D: U.S. Hurricane Individual Risk Methodology

Copyright © 2003 AIR Worldwide Corporation. All rights reserved. Information in this document is subject to change without notice. No part of this document may be reproduced or transmitted in any form, for any purpose, without the express written permission of AIR Worldwide Corporation (AIR).

If you have any questions regarding this document, contact:

AIR Worldwide Corporation 131 Dartmouth Avenue Boston, Massachusetts 02116 USA

Tel: (617) 267-6645 Fax: (617) 267-8284

Attachment D: U.S. Hurricane Individual Risk Methodology

Executive Summary This document provides an overview of the AIR individual risk analysis and mitigation model for U.S. hurricanes. The mitigation model is included within AIR’s detailed loss estimation software, CLASIC/2. The document has been prepared to assist AIR clients in deriving mitigation factors to be used for property loss costs.

The AIR individual risk analysis and mitigation model for U.S. hurricane is a knowledge- based expert system designed to predict the performance of residential and commercial building construction under wind loads. The model follows a logic-based approach based on engineering principles and knowledge of both building construction and building performance in high winds. Modification functions, which are applied to the base damage functions, have been developed to reflect the performance enhancement or diminution of a wide variety of structural and environmental characteristics. The base damage functions used by the AIR hurricane model are individually developed for a “typical” building (defined as exhibiting “average” performance) within construction classes that are broadly defined, without reference to individual structural characteristics.

AIR’s base damage functions were validated and calibrated using extensive and detailed actual hurricane loss data. The relative abundance of hurricane loss data greatly facilitates the determination of average or “typical” building performance.

For purposes of modeling individual risks, the modification functions referenced above reflect the difference between the performance of a building with known structural and environmental characteristics and that of the typical building. For example, the modification function for a residential wood frame building with hip roof indicates how it would perform differently from a typical residential wood frame building whose roof is mathematically defined to exhibit average performance.

The modification functions are themselves functions of wind speed. That is, the effectiveness of mitigating structural and environmental characteristics vary according to the wind speed. In addition, the combined effect of the modification functions is complex and not necessarily additive or multiplicative.

The AIR individual risk methodology follows a structured approach to quantify the impact of secondary characteristics of a building. The model addresses thirty such characteristics and is flexible enough to handle either partial or complete information. The effects of such features are rigorously combined to develop mitigation curves that are dependent on both wind speed and construction class. These mitigation curves are used directly to modify the base damage functions, which have been validated using insurance claims data. The model can be used efficiently to study mitigation benefits of building features.

AIR has developed loss modification factors for a similar range of building features as in the “public domain” Florida Department of Community Affairs (DCA) study. CLASIC/2 users may use these factors to develop rating credits in a similar manner as presented in the public domain study.

Attachment D: U.S. Hurricane Individual Risk Methodology

Introduction AIR’s hurricane model includes an individual risk analysis module, which was developed using a knowledge-based expert system in a structured approach. Based on structural engineering expertise and building damage observations made in the aftermath of historical hurricanes, more than 30 building features, each of which are considered by the module, have been identified as having a significant impact on the building losses. Options for each feature are identified based on construction practice. Algorithms for modifying the vulnerability functions, for both structural and nonstructural damage, are developed based on engineering principles and building performance observations. The module supports any combination of multiple building features on the building damage and produces a modification function to the vulnerability function. The modification function captures the changes to building vulnerability that result when certain building features are present and when information on such building features is known. The modification function varies with the wind intensity to reflect the relative effectiveness of a building feature when subject to different wind speeds.

This document begins with a general discussion of damage estimation, followed by a section on wind induced loads and resulting damage, a section on building and environment features that affect building performance, and concludes with a section on AIR’s development of loss modification factors for a range of building features similar to those used in the Florida public domain study.

The section on damage estimation within AIR catastrophe modeling technology includes:

• An overview of AIR’s catastrophe modeling technology and AIR’s hurricane model • A discussion of the damage estimation module in particular • Measurement of the effects of wind duration • An overview of construction types

The section on wind induced loads and damage will discuss the effects of wind on engineered and non-engineered building performance.

The section on building and environment features will cover the various options in construction that might affect building performance. A number of roof system elements will be examined individually, followed by a discussion of the individual risk methodology. Examples of mitigation curves will complete the discussion.

The section on wind mitigation factors provides background on the Florida statute requiring consideration of building fixtures and construction techniques in rate filings, along with a table of features and the closest AIR selections that can be used to develop rating credits.

Attachment D: U.S. Hurricane Individual Risk Methodology

Damage Estimation within AIR Catastrophe Modeling Technology This section provides a brief overview of AIR catastrophe modeling technology followed by a detailed discussion of the damage estimation component.

Overview of AIR Catastrophe Modeling Technology Figure 1 below illustrates the component parts of the AIR catastrophe models.

Figure 1: Catastrophe Model Components (shaded)

This first module of the AIR tropical cyclone model is used to create the stochastic storm catalog. One hundred years of historical data on the frequency of hurricanes and their meteorological characteristics were used to fit statistical distributions for each parameter. By stochastically drawing from these distributions, the fundamental characteristics of each simulated storm are generated. These parameters include storm track, landfall location and storm heading at landfall, and the intensity variables of central pressure, radius of maximum winds, and forward speed. The result is a large, representative catalog of potential events.

Once values for each of the important meteorological characteristics have been stochastically assigned, each simulated storm is propagated along its track. Peak wind speeds and wind duration are estimated for each geographical location affected by the storm.

The damage estimation component of the model overlays the local intensity of the simulated event onto a database of exposed properties. The model then calculates the resulting monetary damage by applying mathematical relationships, called damage functions, that capture the effects of the intensity of the event, which varies by location, on the exposed buildings and contents. Separate damage functions are developed for different construction types and occupancy classes. Similarly, separate damage functions for each of building, contents, loss of use, and business interruption are applied to the estimated replacement value of the insured property to calculate losses.

Attachment D: U.S. Hurricane Individual Risk Methodology

In the last component of the catastrophe model, insured losses are calculated by applying the policy conditions to the total damage estimates. Policy conditions may include deductibles by coverage, site-specific or blanket deductibles, coverage limits and sublimits, loss triggers, coinsurance, attachment points and limits for single or multiple location policies, and risk specific reinsurance terms.

After all of the insured loss estimations have been completed, they can be analyzed in ways of interest to risk management professionals. For example, the model produces complete probability distributions of losses, or exceedance probability curves, for gross and net losses and for both annual aggregate and annual occurrence losses. Output may be customized to any desired degree of geographical resolution down to location level, as well as by line of business, and within line of business, by construction class, coverage, etc. The model also provides summary reports of exposures, comparisons of exposures and losses by geographical area, and detailed information on potential large losses caused by the extreme events that make up the tail of the distribution.

Damage Estimation Module AIR scientists and engineers have developed mathematical functions, called damage functions, that describe the interaction between buildings (and their contents) and the local intensity to which they are exposed. Damage functions have also been developed for estimating time element losses. These functions relate the mean damage level, as well as the variability of damage, to the measure of intensity at each location. That is, the AIR model estimates a complete distribution around the mean level of damage for each local intensity and each structural type. Because different structural types will experience different degrees of damage, the damage functions vary according to construction materials and occupancy. Losses are calculated by applying the appropriate damage function to the replacement value of the insured property.

The AIR damage functions incorporate the results of well-documented engineering studies, tests, and structural calculations. They also reflect the relative effectiveness and enforcement of local building codes. AIR engineers refine and validate these functions through the use of post-disaster field survey data and through an exhaustive analysis of detailed loss data from actual events.

The AIR damage functions have been extensively validated using detailed actual loss data provided by a number of clients for various storms. Typically the loss data is available by five-digit ZIP Code, as shown in Figure 2 below.

Attachment D: U.S. Hurricane Individual Risk Methodology

Figure 2: Detailed Loss Data by ZIP Code

Validation was performed by comparing simulated and actual losses for five client companies by state and/or county and by line of business for a number of hurricanes, as shown in Figure 3, below. The five client companies are referred to as A-E.

1,000

Actual 100 Simulated

l l w d e s al w n e r e ie arl in i e e i n E oy rtha ran n g p r Eri rtha nie Earl n dr l Irene F n d - n - rges n n on A- orges A-Bret - B-Er - C D o E-Ea A A-Fran B e A-F A -Be B B-Opa -Bo C-O D-Fran -Bo -Bo A-Bertha A- -G B B -Geor -An D-Be D E A- A B C D-Ge

Figure 3: Actual vs. Simulated Losses

The detailed data available to AIR has enabled the fine-tuning of damage ratios by construction class and by coverage type as well as by wind speed. Examples of the comparison between actual loss data and the AIR vulnerability functions are shown in the graphs below.

Attachment D: U.S. Hurricane Individual Risk Methodology

Actual and Simulated Damage Ratios vs Wind Speed Actual and Simulated Damage Ratios vs Wind Speed Coverage A - Single Company, Single Storm Frame - Single Company, Single Storm Damage Ratio Damage

Damage Ratio Damage Simulated Simulated Actual Actual

Wind Speed (mph) Wind Speed (mph) Figure 4 (a) Coverage A– Single Company, Single Storm; (b) Wood Frame – Single Company, Single Storm

Measuring the Effects of Wind Duration The AIR hurricane model is unique in that it develops a complete time profile of wind speeds for each location affected by the storm, thus capturing the effect of wind duration on structures as well as the effect of peak wind speed.

Duration is particularly important in estimating wind damage, which manifests itself at the weak link in a structural system. As each connector is overwhelmed, loads are transferred to the next point of vulnerability. The longer the duration of high winds, the longer this process will continue and the greater will be the resulting damage.

The AIR hurricane model calculates the cumulative effects of wind. That is, when estimating damage to a property at any point in time, to take into account the extent of the damage that has occurred in the preceding period. Each successive damage ratio is applied, in succession, to the remaining undamaged portion of the exposure from the preceding period.

Construction and Occupancy Types AIR engineers have designed the wind damage functions for the U.S. hurricane model to provide detailed breakdowns of loss estimates by coverage, construction class, and occupancy. The entire model includes six major lines of business (Homeowners, Tenants, Apartments/Condos, Commercial, Mobile Homes, and Autos) which are further broken down into 39 distinct construction classifications as outlined in Appendix 8.1.

Apartments and condominiums frequently receive a degree of engineering attention similar to that given to commercial construction. From a structural viewpoint, therefore, commercial construction and apartments/condominiums are similar. Nevertheless, apartments and condominiums have some building components that make them more susceptible to windstorms than commercial construction. These may include balconies, awnings, and double sliding glass doors. Because these components are less engineered at

Attachment D: U.S. Hurricane Individual Risk Methodology

the design and construction stages and are hence more vulnerable, AIR engineers have developed separate damage functions for apartments and condominiums.

Separate damage functions for each of building, contents, and time element provide not only estimates of the mean, or expected, damage ratio corresponding to each wind speed, but also probability distributions around each mean. In the case of building damageability, the damage ratio is the dollar loss to the building divided by the corresponding replacement value of the building. As can be seen in Figure 5, the model ensures non-zero probabilities of zero and one hundred percent loss (for individual properties).

Figure 5: Representative Damage Function The contents damage ratio is the dollar loss to the contents divided by the replacement value of the contents. Contents vulnerability is a function of both building damage and wind speed. The graph below represents the relationship of mean building damage to mean contents damage used in the AIR models. At low wind speeds, the contents damage function is below the building damage function. This relationship reverses at high wind speeds.

Attachment D: U.S. Hurricane Individual Risk Methodology

Figure 6: Contents Vulnerability Relationship

Contents damageability, in the U.S. regional model, is a function not only of building type, but also of occupancy class. That is, for each occupancy class there exists a contents damage function, which is a function of the building damage ratio. Occupancy provides information on the likely contents present and hence the potential vulnerability. AIR models 45 occupancy classes as outlined in Appendix 8.2.

In the case of time element, the damage ratio represents per diem expenses or business interruption losses associated with the expected number of days that the building is uninhabitable (residential) or unusable (commercial). Time element damageability is a function of the mean building damage, as well as the time it takes to repair or reconstruct the damaged building. The functional relationship between building damage and loss of use is established using detailed published building construction and restoration data and on engineering judgment. The graph below represents the relationship between repair/reconstruction time and the mean building damage.

Figure 7: Time Element Vulnerability Relationship

Wind Induced Loads and Damage This section provides a basic overview of how wind and buildings interact. A discussion of the typical wind forces that apply will help to explain the sorts of wind-induced building damage typically observed.

Wind Induced Loads Close to the earth’s surface, roughness of the surface retards the movement of the wind. As the distance from the surface increases, these friction effects become less and less significant and at a certain height they become negligible. This height at which the surface roughness is negligible is referred to as the gradient height. The layer of air below this height, where the wind is turbulent and its speed increases with height, is known as the atmospheric boundary layer whose height ranges from the ground surface to between 1,000

Attachment D: U.S. Hurricane Individual Risk Methodology

and 2,000 feet. All structures are located in this atmospheric boundary layer and act as bluff bodies to wind.

When wind comes into contact with buildings the airflow streamlines separate at the sharp corners of buildings such as wall corners, eaves, roof ridges and roof corners. This separation induces additional turbulence in airflow causing highly fluctuating pressures on the building surfaces. The direction of wind with respect to the building (angle of attack) is also a significant factor in the magnitude and fluctuation of pressures acting on the surfaces of the building.

Figure 8: Wind Flow around Buildings Can Generate Severe Pressure and Suction Forces

In general, when wind acts on a building, the windward wall experiences a pressure pushing inward (positive pressure) while the sidewall and the leeward wall will experience a suction pressure outward (negative pressure). The roof, depending on its slope, will experience uplift (negative pressure). Wind forces are significantly increased at corners, ridges, and at abrupt changes in the direction of wind flow.

Two important flow mechanisms with respect to buildings are discussed below.

Flow Separation: When wind impinges the front wall, it flows upward past the roofline. The wind is unable to turn abruptly at the roofline and so continues past the roof edge, thus separating from the roof eave. Suction forces are often found on the roof under the separated flow, especially near the separation point. The wind that has separated slowly comes back to its original flow direction there by reattaching on the roof surface. The point of reattachment depends on the dimensions of the buildings, roof geometry, the wind speed, and wind direction relative to the building.

Roof-Corner Vortex: Wind approaching the roof corner at a quartering angle flows up over the roof, and rolls up into two vortices of opposite rotational directions originating at the building corner. These vortices are much like miniature tornadoes, producing high

Attachment D: U.S. Hurricane Individual Risk Methodology

speeds under the vortices. These roof-corner vortices are sometimes called the delta-wing vortices.

Figure 9: Wind Flow Separation around Roof

Wind Damage Traditionally, damage surveys have classified buildings into “engineered” or “non- engineered” structures. Most residential dwellings are classified as non-engineered. A typical example is a wood-frame single-family dwelling, a construction that may not have received much attention from a structural engineer. Most commercial structures, built in accordance with building codes and under the supervision of a structural engineer, are classified as engineered structures. A typical example of an engineered structure is a high- rise reinforced concrete building. In general, an engineered building will have less susceptibility to wind damage than a non-engineered building.

Wind Damage to Non-engineered Buildings Figure 9 below illustrates the dynamic process by which non-engineered buildings are damaged by wind. Wind damage primarily affects non-structural elements, such as different components of the building envelope. In most cases, damaged is fairly localized. Roofs and openings in the façade (e.g., windows and garage doors) are typically the first elements to be damaged by wind. Loss of the first shingle allows wind to penetrate and lift the next shingle. Unsecured slates may peel off; metal roofs may roll up and off. Damage accelerates as wind speeds increase. The loss of the first window, either because of extreme pressure or of wind-induced projectiles, can create a sudden build-up of internal pressure that can blow the roof system off from inside, even if it is properly secured. At high wind speeds, the integrity of the entire structure can be compromised, particularly in cases where the roof provides the lateral stability by supporting the tops of the building’s walls. Structural collapse can happen in extreme wind events. Even if the structure remains intact, once the building envelope is breached contents are vulnerable either due to the wind itself or to accompanying rain.

Attachment D: U.S. Hurricane Individual Risk Methodology

Figure 10: Damage Profile of Non-Engineered Buildings

Wind Damage to Engineered Buildings For engineered buildings, damage typically occurs to the following building (nonstructural) components:

• Mechanical equipment • Roofing • Wall cladding • Breaching of doors and windows

Complete structural collapse is a rare event for well-engineered buildings. Even so, damage to non-structural components of a building can amount to significant financial loss.

Typical damage patterns can be seen in the figures below.

Figure 11: Damage to Mechanical Figure 12: Roof Damage Equipment

Attachment D: U.S. Hurricane Individual Risk Methodology

Figure 13: Damage to Cladding Figure 14: Damage to Windows

Attachment D: U.S. Hurricane Individual Risk Methodology

Building and Environmental Features AIR’s wind individual risk analysis model was developed using a structured, knowledge- based expert system using structural engineering expertise and building damage observations made in the aftermath of actual hurricanes. In the AIR model, options for each feature are identified based on construction practice. Algorithms for modifying the vulnerability functions, for both structural and nonstructural damage, are developed based on engineering principles and building performance observations. The module supports the effects of any combination of building features on the building damage and produces a modification function to the vulnerability function. The modification function captures the changes to building vulnerability that result when certain building features are present and when information on such building features is known. The modification function varies with the wind intensity to reflect the relative effectiveness of a building feature when subjected to different wind speeds.

Figure 15 illustrates the application of modification factors to the basic damage functions

Basic Damage Function (Wood Frame) Damage Ratio Damage

Wind Speed

Figure 15 (a): Basic Damage Function

Attachment D: U.S. Hurricane Individual Risk Methodology

Non-Engineered Shutters Engineered Shutters Reduction (%)

Wind Speed

Figure 15 (b): Reduction in Damage for Engineered vs. Non-Engineered Shutters

Basic Damage Function Engineered Shutters Damage Ratio

Wind Speed

Figure 15 (c): Basic Damage Function and Modified Function for Engineered Shutters

Basic Damage Function Nonengineered Shutters Engineered Shutters

None Damage Ratio

Wind Speed

Figure 15 (d): Envelope of Damage Functions, All Protection Options

Attachment D: U.S. Hurricane Individual Risk Methodology

The first step in the development of the individual risk model is to identify building and environmental characteristics that impact the performance of a building in high winds. These features are selected based on research papers and knowledge of building performance in high wind as obtained in the course of hurricane damage surveys. The AIR model includes four groupings of building and environmental features that influence damageability. These are:

1. Nonstructural features, (e.g., cladding, etc.) 2. Structural features (e.g., roof and wall systems) 3. General building features (e.g., building condition, occupancy) 4. Environmental features (e.g., tree exposure, surrounding terrain)

A set of the building characteristics is shown in Table 1.

Attachment D: U.S. Hurricane Individual Risk Methodology

Table 1: Building Features

Residential Commercial–Non-Engineered Commercial–Engineered *Construction Type Construction Type Construction Type Roof Geometry Occupancy Type Occupancy Type Roof pitch Roof Geometry Number of Floors Roof Anchorage Roof Pitch Glass Type Roof Covering Roof Anchorage Glass Percent Roof Covering Attachment Roof Covering Window Protection Roof Deck Roof Covering Attachment Wall Siding Type Roof Deck Attachment Roof Deck Wall Type Roof Age Roof Deck Attachment Small Debris Window Protection Roof Age Large Missiles Glass Type Window Protection Roof Deck Glass Percent Glass Type Roof Covering Wall Type Glass Percent Floor of Concern Wall Siding Type Wall Type Wall Attached Structures Tree Exposure Wall Siding Type Building Condition Small debris Tree Exposure Age Large missiles Small debris Internal Partition Walls Wall Attached Structures Large missiles Exterior Doors Exterior Doors Wall Attached Structures Appurtenant Structures Number of Floors Exterior Doors Roof Geometry Terrain roughness Number of Floors Roof Age Building Condition Terrain roughness Roof Attached Structures Occupancy Building Condition Age Age Appurtenant Structures Appurtenant Structures Roof Attached Structures Roof Attached Structures Internal Partition Walls Internal Partition Walls Building Foundation Connection Building Foundation Connection

Attachment D: U.S. Hurricane Individual Risk Methodology

Possible Options for Building Features The AIR model includes various options for each building and building environment feature. For example, glass is, in general, vulnerable to wind. It is therefore important that the model includes and evaluates the impact of commonly used glass types on building vulnerability. Options for each building feature are shown in Table 2.

Table 2: Options and Description of Building Features

Characteristic Description Valid Options and indices Comments Roof geometry Addresses the shape of the 0 = unknown/default Geometry of the roof affects the roof. 1 = Flat intensity of wind pressures and the 2 = Gable end without bracing resulting uplift resistance. 3 = Hip 4 = Complex 5 = Stepped 6 = Shed 7 = Mansard 8 = Gable end with bracing 9 = Pyramid 10 = Gambrel Roof pitch Addresses the roof slope. 0 = unknown/default Low pitch roofs have greater uplift 1 = Low (<10°) forces acting on the roof as compared to 2 = Medium (10° to 30°) high pitch roofs. 3 = High (> 30°)

Roof covering Addresses the nature of 0 = unknown/default Damage to the roof covering can result material used to cover the 1 = Asphalt Shingles in significant water damage to the roof. 2 = Wooden Shingles interior of the building and contents. 3 = Clay/Concrete Tiles 4 = Light Metal Panels 5 = Slate 6 = Built-up Roof with Gravel 7 = Single Ply Membrane 8 = Standing Seam Metal Roofs 9 = Built-up Roof without Gravel 10 = Single Ply Membrane Ballasted 11= FBC Equivalent

Roof deck Material and construction 0 = unknown/default Roof decks transfer the roof loads to the Information Submitted in Compliance with the 2006 Standards of the 293 Florida Commission on Hurricane Loss Projection Methodology © 2007 AIR Worldwide Corporation

Attachment D: U.S. Hurricane Individual Risk Methodology

Characteristic Description Valid Options and indices Comments type of the roof deck of 1 = Plywood underlying joists and purlins. building. 2 = Wood Planks 3 = Particle Board/OSB Damage to the roof deck results in the 4 = Metal Deck with Insulation Board breach of the building envelops 5 = Metal Deck with Concrete resulting in significant building and 6 = Precast Concrete Slabs interior damage. 7 = Reinforced Concrete Slabs 8 = Light metal Roof covering Nature of the connections 0 = unknown/default Damage to the roof can result due to the attachment used to secure the roof 1 = Screws improper attachment of the roof covering to the roof deck. 2 = Nails/Staples covering to the underlying roof deck. 3 = Adhesive/Epoxy 4 = Mortar Roof deck attachment Nature of the connections 0 = unknown/default Damage to the roof can result due to the used to secure the roof 1 = Screws/Bolts improper attachment of the roof deck to deck to the underlying roof 2 = Nails the underlying roof support system. support system. 3 = Adhesive/Epoxy 4 = Structurally Connected 5 = 6d nails @ 6”/12” 6 = 8d nails @ 6”/12” 7 = 8d nails @ 6”/6” Roof anchorage Nature of the connections 0 = unknown/default Roof anchorage provides the means to used to secure the roof 1 = Hurricane Ties establish a load path to transfer wind support system to the 2 = Nails loads from the roof to the walls. walls. 3 = Anchor Bolts Hurricane straps provide for such 4 = Gravity/Friction anchorage. 5 = Adhesive/Epoxy 6 = Structurally Connected 7 = Clips Roof age Number of Years the roof Numeric value indicating number of the years the roof Roofs loose their strength to resist wind in place. in place loads with age and hence is more 0 = unknown/default vulnerable. 2 = 0-10 years 1 = 10 -20 years 3 = 20 and up

Attachment D: U.S. Hurricane Individual Risk Methodology

Characteristic Description Valid Options and indices Comments

Wall type Materials used for external 0 = unknown/default Different types of wall materials offer walls of building. 1 = Brick/Un-reinforced Masonry varying degrees of resistance to wind 2 = Reinforced Masonry induced lateral loads. Breach in the wall 3 = Plywood can result in internal pressure buildup. 4 = Wood Planks 5 = Particle Board/OSB 6 = Metal Panels 7 = Precast Concrete Elements 8 = Cast-in-Place Concrete 9 = Gypsum board Wall sidings Materials used for 0 = unknown/default Wall sidings offer protection of walls weathering protection of 1 = Veneer Brick/Masonry from wind and rain. walls 2 = Wood shingles Different types of wall siding materials 3 = Clapboards have varying degrees of wind load 4 = Aluminum/Vinyl Siding resistance. 5 = Stone Panels 6 = Exterior Insulation Finishing System (EIFS) Breach in wall sidings can expose wall 7 = Stucco to wind and rain, resulting in water intrusion and internal pressure buildup. Glass type Type of glass used in 0 = unknown/default Different glass types have varying building. 1 = Annealed degrees of resistance to wind loads and 2 = Tempered debris impact. 3 = Heat Strengthened 4 = Laminated 5 = Insulating Glass Units Glass percent The percent area of the 0 = unknown/default Different glass types have varying walls covered by glass. 1 = Less than 5% degrees of resistance to wind loads and 2 = Between 5% and 20% debris impact. 3 = Between 20% and 60% 4 = Greater than 60% Window Protection Describes the nature of 0 = unknown/default Protecting the windows can reduce the wind protection systems 1 = No protection potential damage to a building. used. 2 = Non-engineered storm shutters 3 = Engineered storm shutters

Attachment D: U.S. Hurricane Individual Risk Methodology

Characteristic Description Valid Options and indices Comments

Exterior Doors Describes the nature of the 0 = unknown/default Exterior doors are weak in resisting exterior doors in the 1 = Single Width Doors wind loads and so is the frame that holds building. 2 = Double Width Doors the doors. They deflect considerably 3 = Reinforced Single Width Doors under high wind loads and thus fail. 4 = Reinforced Double Width Doors 5 = Sliding Doors 6 = Reinforced Sliding Doors Building-foundation Connection type between 0 = unknown/default Transfers the vertical and lateral loads connection the structure and 1 = hurricane Ties on the building to the foundation. foundation. 2 = Nails 3 = Anchor Bolts 4 = Gravity/Friction 5 = Adhesive/Epoxy 6 = Structurally Connected Mechanical Systems Description of the 0 = unknown/default mechanical and other 1 = Chimneys equipment on top of roofs. 2 = A/C Units 3 = Skylights 4 = Parapet walls 5 = Overhangs/rake (8 to 36 in.) 6 = Dormers 7 = Other 8 = No attached structures 9 = overhangs/rake (less than 8 in.) 10= overhangs/rake (greater than 36 in.) 11= waterproof membrane 12= Secondary Water Resistance (SWR) Wall Attached Components of a property 0 = unknown/default Attached structures are often more structures that are not an integral part 1 = Carports/canopies/porches vulnerable than the main building of the main building but 2 = Single Door Garages especially if there is inadequate are physically attached to 3 = Double Door Garages anchorage. it. 4 = Reinforced Single Door Garages 5 = Reinforced Double Door Garages 6 = Screened Porches/glass patio doors 7 = Balcony 296 Information Submitted in Compliance with the 2006 Standards of the Florida Commission on Hurricane Loss Projection Methodology © 2007 AIR Worldwide Corporation

Attachment D: U.S. Hurricane Individual Risk Methodology

Characteristic Description Valid Options and indices Comments 8 = no attachment

Appurtenant structures Components of a property 0 = Unknown/default Appurtenant structures may require a that are not an integral part 1 = Detached Garage different treatment in analysis from the of the main building and 2 = Pool Enclosures main building. are not physically 3 = Shed connected to it. 4 = Masonry boundary wall 5 = Other fence 6 = No Appurtenant Structures Internal partition walls Describes the nature of the 0 = Unknown/default Internal partitions, when effective, can interior partition walls. 1 = Wood protect the interior of the building when 2 = Gypsum Boards the building envelope is breached. 3 = Plastered Masonry 4 = Brick 5 = Other Tree Exposure Describes the tree hazard 0 = Unknown/default Trees snap in strong winds and can around the building. 1 = No damage adjacent buildings. 2 = Yes Small debris source Describes the potential of 0 = Unknown/default Small debris, such as, roof gravel, trash small debris in a radius of 1 = No bins, or tree branches, may be picked up 200 ft 2 = Yes by high wind and becomes missiles to breach window glass.

Small debris can be carried into all elevations of building facades at high wind speed. Large missile source Describes the potential of 0 = Unknown/default Garden furniture, or wood planks or large missiles in a radius 1 = No studs dislodged from nearby buildings of 100 ft 2 = Yes may become missiles at high wind speed to breach building envelope.

Attachment D: U.S. Hurricane Individual Risk Methodology

Characteristic Description Valid Options and indices Comments

Terrain roughness Describes the terrain 0 = unknown/default Terrain type is empirically a good conditions 1 = Terrain Type A indicator of missile sources. 2 = Terrain Type B 3 = Terrain Type C For example, buildings located in 4 = Terrain Type D Terrain A (e.g., large city center) are more likely to have gravel impact than Terrain type B (suburban residential area) due to the common roofing construction practice to the median and high-rise buildings while buildings in Terrain B are more likely have large missiles impact due to the residential roof and/or wall framing member failure from nearby buildings. Average building Describes average height 0 = unknown/default height of buildings adjacent to the Numeric value indicating the number of stories building of interest

Attachment D: U.S. Hurricane Individual Risk Methodology

Example of Individual Risk Characteristics: Roof System The section below provides an example of how individual elements of a structural system (in this case, the roof system) combine to influence the damageability and loss due to hurricanes.

The main function of a roof is to enclose the building space and to protect it from the damaging effects of weather elements such as rain, wind, heat, and snow. Consideration is also given to factors such as strength and stability under anticipated loads, heat insulation, lighting, ventilation, sound insulation, aesthetics etc.

In the AIR individual risk analysis model, a roof system is comprised of the following features:

• Roof covering • Roof covering attachment • Roof deck • Roof deck attachment • Roof geometry • Roof age • Roof pitch

A brief description of some of the features that influence damageability and loss due to hurricanes is provided in the following sections. I. Roof Covering The roof covering is the material covering the framework of the roof structure to safeguard the roof against the weather.

The roof covering is fixed to the underlying structure by means of a range of fittings and fixtures. The climatic conditions have a marked influence on the performance and durability of roof coverings. Strong winds may blow off roof coverings such as slates, tiles, and asphalt shingles when they are not properly fixed in position. Extreme temperature changes may cause the material to crack and joints to leak, if not properly protected. Atmospheric effects of fog, salt, air, smoke and other gases may result in corrosion of metal roofing if not protected by painting. (Clay tiles, slates, shingles, and built up roof coverings are unaffected by atmospheric action.) The various effects described above can result in poor performance, reduced life, or both.

It is the roof-deck to which roof coverings are fastened. The decks are supported on structural members such as girders, trusses, or rigid frames. In the case of shell roofs, the decks serve as principal supporting member. In some cases, the roof covering and the deck are combined into one unit, such as corrugated roofing. Because of these relationships, it is desired that both the type of roof covering and the type of roof-deck should be selected concurrently.

Attachment D: U.S. Hurricane Individual Risk Methodology

The weight of roof covering affects the design, weight and the cost of both roof-deck and supporting structure or framework. A heavier roof covering requires a stronger supporting structure, which adds to the cost. For example, sheet metal coverings are very lightweight, shingles can be classified as light to medium in weight, whereas clay tiles and slates are considered to be heavy roof-coverings. Supporting structure and roof deck are designed appropriate to the weight of the chosen roof covering.

The figure below shows typical damage to roof coverings.

Figure 16: Damage to Roof Covering.

II. Roof Deck The roof deck transfers the roof loads to the underlying trusses or rafters. Damage to the roof deck constitutes a breach of the building envelope and can result in significant building and interior damage. Some of the commonly used roof decks are plywood, precast concrete slabs, reinforced concrete slabs, and light metal.

Figure 17: Damage to Roof Deck

Attachment D: U.S. Hurricane Individual Risk Methodology

III. Roof Geometry Roof geometry largely determines the magnitude of aerodynamic loads experienced by a particular roof. The geometry affects the intensity of wind pressures and the resulting uplift resistance. Common roof shapes are gable and hip, although a variety of roof shapes are possible. A brief description of some of the roof shapes is provided below.

Gable Roof. This roof slopes in two directions so that the end formed by the intersection of slopes is a vertical triangle.

Figure 18: Illustration and Example of Gable Roof

Hip Roof. This roof slopes in four directions such that the end formed by the intersection of slopes is a sloped triangle.

Figure 19: Illustration and Example of Hip Roof

Mansard Roof. Like the hip roof, this roof also slopes in four directions, but there is a break in each slope.

Figure 20: Illustration of Mansard Roof

Attachment D: U.S. Hurricane Individual Risk Methodology

Individual Risk Methodology As explained in Section 4, building aerodynamics is a complex phenomenon. The ultimate performance of a building depends on the interaction of several building components. Moreover, the damage due to wind is progressive, which means that a failure that begins at a localized level can eventually grow to a catastrophic level. Thus it is important to recognize the way damage progresses and the role and importance of building components at each stage of failure.

The individual risk methodology follows a structured, logical approach that groups building characteristics according to their function. In this way the methodology reflects the contribution of each characteristic to the overall building performance. This knowledge-based methodology relies on expert experience in wind engineering, damage observation from post-disaster surveys, and data from wind tunnel experiments. The ultimate goal is to develop a modification function that is applied to the base damage function. The result is a modified damage function that reflects the impact of one or more selected building characteristics.

Weightings are used to combine the effects of features whose interaction is complex and not necessarily additive. These are introduced to evaluate features that modify the performance of the system. If we consider roof system as an example, roof age, roof pitch, and roof geometry modify the performance of the roof as a whole and therefore the weight should be used as a multiplier. The weights are dependent on wind speed and construction class, and are appropriately selected to reflect the importance of a feature at certain levels of a building’s damage state.

Evaluating Building Performance There are two primary metrics for evaluating the impact of a building or environmental feature on overall building performance. The first is a weighted value assigned to the various options for building or environmental features. The value for any given option of any given feature reflects the relative prevalence of use among the options and is independent of other features. That is, the value is designed such that the most commonly used option is assigned a value close to 1.0. The implication is that a building with this option is expected to perform very similarly to the average, or “typical” building represented by the base damage functions.

The value assigned for an option that is considered to be more vulnerable (less wind resistive) than the most commonly chosen one is greater than 1.0. That is, a building with this option will be more vulnerable than the average building. Similarly, the value assigned for an option that is considered to be less vulnerable (more wind resistive) than the most prevalent one will be less than 1.0. Such a building will be less vulnerable than the average building. The value for a given option is a constant. If no information is available on the option, the default value is 1.00, which means that the base damage function is used without modification.

Attachment D: U.S. Hurricane Individual Risk Methodology

The second metric is a value of one of two types. One type is used to develop simple weighted averages. These are used to evaluate the loss contribution of several features that together constitute a system, such as roof. They are wind speed dependent, that is, the contribution of each feature varies with wind speed. For example, a roof may consist of three features: roof covering, roof deck and roof attachment. The loss contribution to the roof system from these three features is expected to be different at different wind speeds. At low wind speeds, the roof covering drives the damage since it is at relatively low wind speeds that damage to roof covering occurs. As wind speeds increase, the roof deck becomes vulnerable. In this case, roof deck failure will result in loss of roof covering regardless of the type (or option) of roof covering present. Therefore, as wind speed increases, the weight for roof deck increases. In contrast, at higher wind speeds, the weight for roof covering decreases because it is already lost. The sum of the weights for a system should add up to 1.0.

The second type of weight is used to combine the effects of features whose interaction is complex and not necessarily additive. These are introduced to evaluate features that modify the performance of the system. If we consider roof system as an example, roof age, roof pitch, and roof geometry modify the performance of the roof as a whole and therefore the weight should be used as a multiplier. The weights are dependent on wind speed and construction class, and are appropriately selected to reflect the importance of a feature at certain levels of a building’s damage state.

Example of Wind Speed Dependency As noted above, the performance of building features is wind speed dependent. This can be further explained by considering in detail a single feature, such as roof-wall anchorage. Roof-wall anchorage provides the means to establish a load path to transfer wind loads from the roof to the walls.

This anchorage can be provided through:

1. Hurricane straps 2. Structural connections 3. Nails 4. Epoxy/adhesive 5. Anchor bolts 6. Gravity/friction

By selecting hurricane straps as one of the options for “roof wall anchorage” we can observe varying mitigation benefits in three different wind speed regimes.

• For lower wind speeds, the mitigation benefits (in terms of reduced damage) of hurricane straps are similar to any other roof-wall anchorage mechanism. Hence mitigation benefits are not high in this wind speed regime. • At higher wind speeds, hurricane straps are very effective in reducing the damage. It is in this wind regime that the presence of straps is most important. Hence mitigation benefits are high.

Attachment D: U.S. Hurricane Individual Risk Methodology

• At very high wind speeds, the effectiveness of hurricane straps in reducing the damage decreases. Hence, mitigation benefits are not high.

Because of its weakness, there are no mitigation benefits for “epoxy/adhesive” as an option for “roof wall anchorage”. In fact, “penalties” are applied and the size of the penalty is wind speed dependent.

• At low wind speeds, the effectiveness of “epoxy/adhesive” is similar to other roof-wall anchorage mechanisms. Hence we do not see any large penalties being exacted in this wind speed domain. • At higher wind speeds, however, the choice of a weak roof-wall anchorage has a significant impact on damage. Hence we see high penalty being exacted. • At very high wind speeds, most roof-wall anchorage mechanisms would be ineffective. Thus the penalty for choosing a weak one is small.

The above two examples illustrate how different wind regimes affect the accrual or non- accrual of mitigation benefits for a particular feature, as does Figure 21 below.

Figure 21: Roof-Wall Anchorage

Sample Mitigation Curves Failure of windows during a hurricane is often due to the impact of flying debris or to exceeding the pressure capacity of the window. Failure of windows is a breach of the building envelope and subsequently allows wind and water to enter the building. Even if there is no structural damage, this breach of building envelope and subsequent wind and water intrusion would cause damage to contents and building interior finishes, which can lead to substantial loss. Protecting the windows is critical in reducing the potential damage to a building. One mitigation option is the installation of engineered storm shutters.

As can be seen in Figure 22, the percentage reduction in damage achieved by the installation of engineered storm shutters is wind speed dependent At lower and high wind

Attachment D: U.S. Hurricane Individual Risk Methodology speeds the percentage reduction in damage is comparable to that of any other equivalent window protection mechanism. Hence we do not see any higher order percentage reduction of damage at lower and higher wind speeds. The effectiveness of engineered storm shutters is greatest in the middle wind regime where shutters have the greatest marginal impact on damage reduction.

Figure 22: Modification Function for Engineered Storm Shutters

Damage to the roof covering can also result in significant water damage to the interior of the building and contents. A sample modification curve for roof covering comprised of slates is provided in Figure 23.

Figure 23: Modification Function for Slate Roofing

Mitigation Credits for Construction to Florida Building Code (FBC 2001) For buildings built to the minimum requirements of Florida Building Code (FBC 2001), AIR identified six unique building categories in the entire State of Florida, as listed in Table 3, by taking into consideration the design wind speed, the terrain exposure category, and the requirements of Wind-borne Debris Region (WBDR) and High-Velocity Hurricane Zone (HVHZ). Figure 24 shows the geographical locations of these building categories in

Attachment D: U.S. Hurricane Individual Risk Methodology

Florida. The vulnerability functions for these six unique building categories were derived by selecting the relevant building features and options from the AIR individual risk module that meet the minimum requirements of the Florida Building Code 2001. For example, AIR building category 6 is located in the High Velocity Hurricane Zone (HVHZ) and is designed for a wind speed of 146 mph and Exposure C (Open country), as specified in FBC 2001. The code stipulates all openings to be protected in High Velocity Hurricane Zone. Table 4 illustrates the roof covering, the roof deck, the roof-deck attachment, roof anchorage and opening protection options selected for residential wood frame category 6 building to meet the minimum requirements of the Florida Building Code.

Table 3: Building Categories as per Florida Building Code (FBC 2001) Building ID Wind Speed Exposure WBDR 1 <120 B No 2 ≥ 110 C Yes 3 ≥ 120 B No 4 ≥ 120 B Yes 5 ≥ 120 C Yes 6 HVHZ* C Yes

* Broward and Miami-Dade counties.

Attachment D: U.S. Hurricane Individual Risk Methodology

Figure 24: Buildings built to the minimum requirements of Florida Building Code (FBC 2001)

Table 4: Model Parameters for Building Category 6 as per Minimum Requirements of FBC 2001

Parameter Building 6 Wind Speed 146 mph Exposure C WBDR Yes HVHZ Yes Roof Covering FBC Equivalent Roof Deck Plywood Roof-Deck Attachment 8d@6"/6" Roof Anchorage Hurricane Straps Opening Protection Engineered Shutters Doors Impact Resistant - Reinforced

Attachment D: U.S. Hurricane Individual Risk Methodology

Evaluating Wind Loss Mitigation Credits with the CLASIC/2 Location Detail Record

Background

On June 6, 2002, the Florida Department of Insurance issued an Informational Memorandum (0470M) which outlined provisions of Florida Statute Section 627.0629(1). The new statute requires that rate filings received by the Florida Department of Insurance on or after June 1, 2002 include credits for “fixtures or construction techniques demonstrated to reduce the amount of loss in a windstorm”. The statute further requires all insurers to make a rate filing which includes actuarially reasonable differentials by February 28, 2003.

This document provides guidance to our clients who wish to use CLASIC/2 for evaluating these wind mitigation credits. According to the statute, the following 6 areas must be considered.

• Roof strength • Roof covering performance • Roof-to-wall strength • Wall-to-floor-to-foundation strength • Opening protection • Window, door, and skylight strength

The Information Memorandum states that other construction techniques have also been demonstrated to influence loss and thus should be considered as well:

• Roof shape • Wall Construction • Opening Protection for non glazed openings (e.g., doors) • Gable End Bracing for roof shapes other than hip

The CLASIC/2 Location Detail Record The location detail record (record 64 in the UPX file and record 74 in the UFX file) contains several building and environmental features that influence loss from windstorms. The following features are of particular relevance to the Florida Statute:

Attachment D: U.S. Hurricane Individual Risk Methodology

Mitigation Category CLASIC/2 Location Detail Field

Enhanced Roof Streng h Roof Deck Roof Deck Attachment Roof Covering Roof Covering Attachment Roof Geometry Roof Pitch Year Roof Built

Roof Covering Performance Roof Covering Roof Covering Attachment

Roof to Wall Strength Roof Anchorage Wall Type Wall Siding

Wall to Floor to Founda ion Strength Foundation Connec ion

Opening Protection Window Protection

Window,Door,Skylight Strength Glass Type Exterior Doors Wall Attached Structures

Roof Shape Roof Geometry

Wall Construction Wall Type Wall Siding

Opening Protection for Non Glazed Openings Exterior Doors Wall Attached Structures

Note that some features can be used to address more than one area of interest. For example, while roof geometry is a separate category it also influences roof strength.

Developing Mitigation Credits In 2001 the Florida Department of Community Affairs (DCA) commissioned a study to estimate the loss reduction potential of different wind resistive building features. Loss cost relativities were presented for a set of primary rating factors, with various options for each factor. The Florida Department of Insurance recognizes the “public domain” DCA study as one basis for deriving credits, but notes that insurers may rely upon other studies.

AIR has developed loss modification factors for a similar range of building features. CLASIC/2 users may use these factors to develop rating credits in a similar manner as presented in the public domain study. It is important to note that the AIR and DCA studies were developed independently, and as a result there are differences in the methods, features and conclusions. While there is no direct mapping of the rating tables used in the DCA study to the AIR features, AIR has developed a list of AIR features that provide a similar range of relativities. The DCA features and closest AIR selections are shown below:

Attachment D: U.S. Hurricane Individual Risk Methodology

DCA Feature Similar AIR Category and Selections

Roof Cover Roof Covering + Roof Covering Attachment Non FBC Equivalent Asphalt Shingles + Nails FBC Equivalent FBC Equivalent+Nails

Roof Deck Attachment Roof Deck + Roof Deck Attachment A Plywood +6d nails @ 6”/12” B Plywood + 8d nails @ 6”/12” C Plywood + 8d nails @ 6”/6”

Roof Wall Connection Roof Anchorage Toe Nails Nails/Screws Clips Clips Single Wraps Hurricane Ties Double Wraps Hurricane Ties

Opening Protection Window Protection None None Basic Non-Engineered Shutters Hurricane Engineered Shutters

Roof Shape Roof Geometry Hip Hip Other Gable End without Bracing

No Secondary Water Resistance No Secondary Water Resistance Secondary Water Resistance Secondary Water Resistance

The AIR study made use of the same thirty-one points used to define locations in the public domain study. The analysis showed that loss relativities did not vary significantly by location. The resulting AIR wind loss mitigation relativities are therefore presented as the mean of the relativities across all locations, and take into account locations in each of the wind speed and terrain exposure combinations that result from the Florida Building Code.

Attachment D: U.S. Hurricane Individual Risk Methodology

About AIR Worldwide Corporation AIR pioneered the development and application of catastrophe loss estimation technology– a technology that provided companies, for the first time, valuable tools to assess and manage their catastrophe risk. Today, AIR provides clients with a full suite of integrated products for underwriting, pricing, portfolio management, risk transfer and financing.

AIR has developed models to estimate potential catastrophe losses from all major natural hazards, including hurricanes, earthquakes, extra tropical cyclones, tornadoes, hailstorms and flood, for more than 35 countries throughout North and South America, the Caribbean, Europe and the Pacific Rim. By incorporating the most recent scientific data and research results, AIR is able to provide clients with the most reliable loss estimates available on the market today.

AIR is also the premier provider of weather risk management services. In response to the increasing demand for accurate weather and climate information for evaluating new weather risk management opportunities, AIR provides weather data and climate forecasts for the weather risk management market around the world.

AIR has created a broad range of software solutions to serve the diverse needs of our clients, among them: CATMAP®/2, CATRADER®, CLASIC/1™ and, most recently, CLASIC/2™. Web-based applications include ALERT™ (AIR Loss Estimates in Real Time), ClimateCast™ and AIRProfiler®.

Attachment D: U.S. Hurricane Individual Risk Methodology

Appendices

Construction Classes

Construction Class Description Wood Frame Wood frame structures tend to be mostly low rise (one to three stories, occasionally four stories). Stud walls are typically constructed of 2 inch by 4 or 6-inch wood members vertically set 16 or 24 inches apart. These walls are braced by plywood or by diagonals made of wood or steel. Many detached single and low-rise multiple family residences in the United States are of stud wall wood frame construction. Light Wood Frame Light wood frame structures are typically not built in the United States but would be found in other countries such as Japan. In Hawaii, this classification would include single wall (studless) construction framed with light timber trusses. Masonry Veneer "Masonry veneer" are wood frame structures faced with a single width of brick or stone. Corrugated metal ties support the masonry and prevent it from falling away, but allow for differential movement between the masonry and the frame. Heavy Timber "Heavy Timber" structures typically have masonry walls with heavy wood column supports and the floor and roof decks are 2-3 inch tongue-and-groove planks. Masonry Masonry buildings when the other detailed technical characteristics of the building are unknown. Adobe Adobe construction uses adobe (clay) blocks with cement or cement-clay mixture as mortar and roof consists of timber frame with clay tiles or in some cases metal roofing. Rubble Stone Masonry "Rubble stone masonry" consists of low rise perimeter load bearing walls composed of irregular stones laid as coursed or uncoursed rubble in a cement mortar bed with floor and roof joists constructed with wood framing. Unreinforced Masonry - Unreinforced masonry buildings consist of structures in which there is no steel Bearing Wall reinforcing within a load bearing masonry wall, floors, roofs, and internal partitions in these bearing wall buildings are usually of wood. Reinforced Masonry Reinforced masonry construction consists of load bearing walls of reinforced brick or concrete-block masonry. Floor and roof joists constructed with wood framing are common. Reinforced Masonry Reinforced masonry construction consists of load bearing walls of reinforced brick or Shear Wall (with MRF) concrete-block masonry. Reinforced masonry buildings with "Moment Resisting Frames" carry lateral loads by bending. "Shear Walls" are continuous reinforced brick, or reinforced hollow concrete block walls extending from the foundation to the roof and can be exterior walls or interior walls. Reinforced Masonry Reinforced masonry construction consists of load bearing walls of reinforced brick or Shear Wall (without concrete-block masonry. "Shear Walls" are continuous reinforced brick, or reinforced MRF) hollow concrete block walls extending from the foundation to the roof and can be exterior walls or interior walls. Joisted Masonry Masonry exterior walls with roof of combustible materials on non-combustible supports.

Reinforced Concrete Concrete buildings when the other detailed technical characteristics of the building are unknown. Reinforced Concrete Building constructed with reinforced concrete columns and beams, as well as reinforced Shear Wall (With MRF) concrete floor and roof. “Moment Resisting Frames” carry lateral loads by bending. “Shear Walls” are continuous reinforced concrete extending from the foundation to the roof and can be exterior walls or interior walls.

Attachment D: U.S. Hurricane Individual Risk Methodology

Construction Class Description Reinforced Concrete Building constructed with reinforced concrete columns and beams, as well as reinforced Shear Wall (Without concrete floor and roof. Reinforced concrete “Shear Walls” are continuous reinforced MRF) concrete, extending from the foundation to the roof and can be exterior walls or interior walls. This category typically consists of buildings, with a concrete box structural system with shear walls. The entire structure, along with the usual concrete diaphragm, is typically cast in place. Reinforced Concrete Buildings constructed with reinforced concrete columns, beams, and slabs. Reinforced MRF – Ductile concrete . “Moment Resisting Frames” carry lateral loads due to earthquakes by bending. This kind of structural system can sustain large deformations, and absorb energy without brittle failure. Reinforced Concrete Buildings constructed with reinforced concrete columns, beams, and slabs. “Moment MRF – Non-Ductile Resisting Frames” carry lateral loads due to earthquakes by bending. This kind of structural system can sustain large deformations, and absorb energy without brittle failure. These structures have insufficient reinforcing steel embedded in the concrete and thus display low ductility. Tilt-Up Tilt-up buildings are constructed with reinforced concrete wall panels that are cast on the ground and then tilted upward into their final positions. These wall units are then anchored to the foundation and attached to each other. The roof and floor decks are typically wood. More recently, the wall panels are fabricated off-site and trucked in. These buildings tend to be one or two stories in height. Precast Concrete The precast frame is essentially a post and beam system in concrete where columns, beams and slabs are prefabricated and assembled on site. Steel Steel frame buildings consists of steel columns and beams and the other detailed technical characteristics of the building are unknown. Light Metal “Light Metal” buildings have steel frames spanning the short dimension of the building and have diagonal tie downs in the long dimension of the building. These buildings are usually clad with lightweight metal or asbestos reinforced concrete siding, often corrugated. They typically are one-story structures without interior columns. They are often used for agricultural structures, industrial factories and warehouses. Braced Steel Frame Buildings constructed with steel columns and beams that are braced with diagonal steel members to resist lateral forces.

Steel MRF – Perimeter Buildings constructed with steel columns and beams that use only the frame members on the periphery of the structure to carry lateral loads. The internal beams and columns only carry the gravity load to the foundation. Steel MRF – Distributed Buildings constructed with steel columns and beams to carry lateral loads distributed throughout the building. The diaphragms are usually concrete, sometimes over steel decking. This structural type is seldom used for low-rise buildings. Long Span Building constructed with steel frame and metal siding and roof of wood or other combustible material. Typically gymnasiums or auditoriums. Semi-Wind Resistive A licensed engineer does not design the structure but an attempt is made to build in accordance with an accepted wind building code; code compliance is not assured. Some engineering input may have occurred. Most of the details in a wind resistive structure are found in a semi-wind resistive structure, but not all components are wind resistive. Wind Resistive The structure must be designed by a licensed engineer to comply with the wind code. During construction, particular attention should be paid to details. Characterized by the presence of properly sized wind resistant connectors, adequate bracing and a continuous load path from the roof to the foundation, i.e., the roof is tied to the walls, the floors are attached to each other (for two or more stories) and the walls are tied to the foundation

Attachment D: U.S. Hurricane Individual Risk Methodology

Construction Class Description Mobile Homes Represents a weighted average of tie-down types, including no tie-downs. This code would be used for mobile homes (manufactured homes) when the tie down information is unknown. Mobile home with no A manufactured or mobile home is a home built entirely in a factory under a federal tie-downs building code administered by HUD. The Federal Manufactured Home Construction and Safety Standards (put out by HUD) went into effect June 15, 1976. Manufactured homes may be single- or multi-section and are transported to the site and installed. The federal standards regulate manufactured housing design and construction, strength and durability, transportability, fire resistance, energy efficiency and quality. A mobile home with no tie-downs is an older home manufactured to the voluntary standards of the industry. In particular, these structures do not have any securing device to connect the home to the ground. Mobile home with Mobile homes (manufactured homes) that has at least one of the following types of tie- partial tie-downs downs: Baling straps , Diagonal tie, ground anchor, tie strap, or vertical tie Mobile home with full A manufactured home that has all the necessary anchoring and support systems such as tie-downs piers, footers, ties, anchoring equipment, ground anchors and other equipment that supports the manufactured home and secures it to the ground in a manner according to the required manufactured home standard.

Attachment D: U.S. Hurricane Individual Risk Methodology

Occupancy Classes

Clasic/2 Code Category Occupancy Class 300 Unknown Unknown

301 Residential General Residential 302 Residential Permanent Dwelling - Single Family 303 Residential Permanent Dwelling - Multi Family 304 Residential Temporary Lodging 305 Residential Group Institutional Housing 306 Residential Apartments/Condos/Flats 307 Residential End Terrace 308 Residential Mid Terrace

311 Commercial General Commercial 312 Commercial Retail Trade 313 Commercial Wholesale Trade 314 Commercial Personal and Repair Services 315 Commercial Professional, and Technical 316 Commercial Health Care Services 317 Commercial Entertainment and Recreation 318 Commercial Parking

321 Industrial General Industrial 322 Industrial Heavy Fabrication and Assembly 323 Industrial Light Fabrication and Assembly 324 Industrial Food and Drug Processing 325 Industrial Chemical Processing 326 Industrial Metal and Minerals Processing 327 Industrial High Technology 328 Industrial Construction 329 Industrial Petroleum 330 Industrial Mining

341 Religion and Nonprofit Religion and Nonprofit 342 Religion and Nonprofit Church 343 Government General Services 344 Government Emergency Services 345 Education Education

Clasic/2 Code Category Occupancy Class

351 Transportation Highway 352 Transportation Railroad 353 Transportation Air 354 Transportation Sea and Inland Waterways

Attachment D: U.S. Hurricane Individual Risk Methodology

361 Utilities Electrical 362 Utilities Water 363 Utilities Sewer 364 Utilities Natural Gas 365 Utilities Telephone and Telegraph

371 Miscellaneous Communication 372 Miscellaneous Flood Control 373 Miscellaneous Agriculture 374 Miscellaneous Greenhouse

Attachment D: U.S. Hurricane Individual Risk Methodology

References Bhinderwala S., “Insurance Loss Analysis of Single Family Dwellings Damaged in Hurricane Andrew,” Master’s Thesis, Clemson University, 1995. “Building Performance: Hurricane Andrew in Florida, Observations, Recommendations and Technical Guidance.” FEMA and Federal Insurance Administration, December 21, 1992. “Building Performance Assessment: Hurricane Fran in North Carolina, Observations, Recommendations and Technical Guidance,” FEMA, Mitigation Directorate, March 19, 1997. Cermark. J.E., “Wind Tunnel Testing of Structures,” Journal of Engineering Mechanics Division, ASCE, Vol. 106, No. EM6, December 1980. Crandell J.H., “Statistical Assessment of construction characteristics and performance of homes in Andrew and Opal,” Journal of Wind Engineering and Industrial Aerodynamics, 77 and 78, 1998, pages 695-701. Cross, Mark, L., and Thorton, John, H., “The Probable Effect of an Extreme Hurricane on Texas Catastrophe Plan Insurers,” The Journal of Risk and Insurance, 50(3): 417- 444, 1983. “Exterior Insulation and Finish Systems,” Natural Hazard Mitigation Insights Publication No. 7, ISSN 1089-6058, February 1997. Friedman, Don G., “Natural Hazard Risk Assessment for an Insurance Program,” The Geneva Papers on Risk and Insurance, International Association for the Study of Insurance Economics (Geneva Association), Geneva, Switzerland, Volume I. Number 30, pages 57-128, January 1984. Freidman, D.G., “Assessing Hurricane Impact on Human Settlements.” Ministry of Human Settlements and Public Works, Government of Mexico, Symposium on Hurricane Floods: Their Effect on Human Settlements, La Pax, Baja California Sur, Mexico, November 24-27, 1981. Friedman, Don G., “Computer Simulation in Natural Hazard Assessment,” University of Colorado, Program on Technology, Environment and Man. Monograph # NSF-R-A-E 15-002, 1975. Harris P.L., Beason W.L. and Minor J.E., “The Effects of Thickness and Temper on the Resistance of Glass to Small Missile Impact,” Institute of Disaster Research, Texas tech University, May 1978. “Hurricanes of 1992”, Proceedings of Hurricanes of 1992, Andrew and Iniki One Year Later, ASCE, Miami, Fl, 1993. “Hurricane Camille - August 1969: A Survey of Structural Damage along the Mississippi Gulf Coast,” National Bureau of Standards report, December 4, 1970. Jianming Yin. “Probabilistic Study of Wind Pressures on Buildings,” Ph.D. Thesis, Texas Tech University, 1996. Kareem, Ahsan, “Hurricane Alicia: One Year Later,” ASCE Specialty Conference, Galveston, TX, August 16-17, 1984. Liu, Henry, “Wind Engineering- A Handbook for Structural Engineers,” Prentice Hall, 1991. Marshall, R.D., “Fastest Wind Speeds in Hurricane Alicia,” National Bureau of Standards Technical Note 1197, Washington DC, US Department of Commerce, NIST, 1984.

Attachment D: U.S. Hurricane Individual Risk Methodology

Mehta, K.C., Minor, J.E., and Reinhold, T.A., “Wind Speed-Damage Correlation in Hurricane Frederick,” Journal of the Structural Division, ASCE, Vol. 109, Number 1, Paper No. 17626, January 1983. Mehta, K.C, Cheshire, R.H., and McDonald, J.R., “Wind Resistance Categorization of Buildings for Insurance,” Journal of Wind Engineering and Industrial Aerodynamics, 41-44, pages 2617-2628, 1992. Mehta, K.C., Smith, D.A., and Cheshire, R.H., “Knowledge-based System for wind damage to buildings.” Proceedings of Hurricanes of 1992, Andrew and Iniki One Year Later, ASCE, Miami, Fl, 1993. Minor J. “A Simple Window Glass Design Chart,” Proceedings of the 8th US National Conference on Wind Engineering, Baltimore, MD, 1997. Minor J., “New Building Code Provisions for Window Glass,” Proceedings of the NSF/WERC Symposium on High Winds and Building Codes, Kansas City, MO, Nov 2-4, 1987. Minor, J.E. and Beason, W.L. “Window Glass Failure in Windstorms,” Journal of the Structural Division, ASCE, Volume 101, No. ST1, Proceedings Paper 11834, January 1976. Minor, J.E. and Mehta, K.C. and McDonald, J.R., “Wind Damage Observations and Implications,” Journal of the Structural Division, ASCE, Vol. 105, No ST11, November 1979. Minor, J.E., Mehta, K.C., and McDonald, J.R., “Failures of Structures Due to Extreme Winds,” Journal of the Structural Division, ASCE, Vol. 98, No. ST11, 1972. “Performance of Metal Buildings in High Winds,” Natural Hazard Mitigation Insights Publication No. 9, Jan1999, ISSN 1089-6058. Rosen H. J, Construction Materials for Architecture, Chapter 7: “Thermal and Moisture Protection,” Chapter 8: “Glass and Curtain Walls,” John Wiley and Sons, 1985. Sandri P. “An Artificial Neural Network for Wind-Induced Damage Potential to Non-Engineered Buildings,” Ph.D. Thesis, Texas Tech University, 1996. Savage, Ruldoph P., Baker, J., Golden, H., Kareem, A., and Manning, B.R., “Hurricane Alicia, Galveston Houston, Texas August 17-18, 1983,” U.S. Department of Commerce Sciaudone, J.C., Feuerborn, D., Rao, G.N., and Daneshvaran, S., “Development of Objective Wind Damage Functions to Predict Wind Damage to Low-Rise Structures,” Proceedings of the 7th U.S. National Conference on Wind Engineering, Baltimore, MD, 1997. Simiu, E., and Scanlan, R.H., Wind Effects on Structures, Wiley-Interscience, 1986. Smith T.L. and McDonald J.R., “Preliminary Design Guidelines for Wind-Resistant Roofs on Essential Facilities,” Proceedings of the 7th U.S. National Conference on Wind Engineering, Vol II, Baltimore, MD, 1997. Sparks, P.R., Schuff, S.D., and Reinhold, T.A., “Wind Damage to Envelopes of Houses and Consequent Insurance Losses,” Proceedings of Wind, Rain and Building Envelope, Invitational Seminar, University of Western Ontario, May, 1994. Stubbs, N., and Perry D.C., “Damage Simulation Model for Building and Contents in Hurricane Environment,” Proceedings ASCE Structures Congress XIV, 989-996, 1996.

Attachment D: U.S. Hurricane Individual Risk Methodology

Sutt E., Muralidhar K., and Reinhold T., “Roof Sheathing Uplift Resistance for Hurricanes,” Clemson University Report, 1998. Tryggvason, B.V., Surry, D., and Davenport, A.G., “Predicting Wind Induced Response in Hurricane Zones,” Journal of the Structural Division, ASCE, Volume 102, No ST12, December 1976. Unanwa C. O., “A Model for Probable Maximum Loss in Hurricanes,” Ph. D. Thesis, Texas Tech University, 1997. Unanwa C.O., McDonald J.R., Mehta K.C. and Smith D.A, “Development of Wind Damage bands for Buildings,” Journal of Wind Engineering and Industrial Aerodynamics, Vol. 84, January 2000. Wallace G. F., “Mitigating Damages in Hawaii's Hurricanes, A Perspective on Retrofit Options,” Insurance Institute for Property Loss Reduction, Occasional Paper Series OP-3, August 1993. Watford S. W., “A Statistical Analysis of Wind Damages to Single family Dwellings Due to Hurricane Hugo,” Master’s Thesis, Clemson University, 1991. Wiggins, J.H., “Natural Hazards Tornado, Hurricane, Severe Wind Loss Models,” U.S. Department of Commerce. PB-294 594, December 1976.

Attachment E: Remapped ZIP Codes

FCHLPM Remapped FCHLPM Remapped FCHLPM Remapped FCHLPM Remapped ZIP ZIP ZIP ZIP ZIP ZIP ZIP ZIP 32004 32082 32231 32207 32461 32413 32644 32626 32006 32073 32232 32209 32463 32428 32654 32640 32007 32177 32235 32206 32509 32526 32655 32643 32026 32083 32236 32205 32511 32507 32658 32615 32030 32068 32237 32257 32512 32507 32662 32640 32035 32034 32238 32244 32513 32503 32663 32686 32041 32097 32239 32277 32516 32506 32664 32667 32042 32044 32240 32250 32520 32501 32681 32667 32050 32068 32241 32223 32521 32507 32683 34449 32056 32055 32245 32216 32522 32502 32692 32680 32067 32073 32247 32207 32523 32501 32697 32054 32072 32087 32255 32207 32524 32501 32704 32712 32079 32043 32260 32259 32530 32583 32706 32744 32085 32084 32267 32233 32537 32536 32710 32818 32099 32220 32276 32209 32538 32567 32715 32714 32105 32180 32290 32202 32540 32541 32716 32714 32111 34472 32302 32301 32549 32548 32718 32708 32115 32114 32313 32304 32559 32526 32719 32708 32116 32118 32314 32301 32560 32533 32721 32720 32120 32114 32315 32303 32562 32561 32722 32720 32121 32119 32316 32304 32572 32570 32723 32724 32122 32114 32318 32305 32573 32501 32727 32726 32123 32119 32323 32322 32574 32501 32728 32725 32125 32117 32326 32327 32575 32501 32733 32792 32126 32118 32329 32320 32576 32501 32739 32738 32133 32179 32330 32351 32581 32501 32747 32771 32135 32136 32335 32334 32582 32501 32752 32779 32138 32148 32337 32317 32588 32578 32753 32713 32142 32136 32341 32340 32589 32501 32756 32757 32147 32148 32345 32344 32590 32501 32762 32765 32149 32148 32353 32351 32591 32501 32768 32712 32151 32110 32357 32331 32592 32501 32772 32771 32157 32181 32360 32334 32593 32501 32774 32763 32158 32159 32361 32336 32594 32502 32775 32754 32160 32656 32362 32305 32595 32502 32777 32757 32170 32169 32395 32301 32596 32502 32781 32780 32173 32174 32399 32301 32597 32501 32782 32796 32175 32174 32402 32401 32598 32502 32783 32780 32178 32177 32406 32405 32602 32601 32790 32789 32182 32134 32410 32401 32604 32603 32791 32779 32183 32179 32411 32408 32610 32611 32793 32792 32185 32148 32412 32401 32612 32611 32794 32751 32192 32617 32417 32407 32613 32611 32795 32746 32193 32112 32422 32433 32614 32608 32799 32746 32198 32114 32432 32442 32616 32615 32802 32801 32201 32202 32434 32433 32627 32601 32816 32826 32203 32209 32447 32446 32633 32640 32834 32828 32228 32227 32452 32425 32634 32686 32853 32803 32229 32218 32454 32459 32635 32605 32854 32804 32230 32254 32457 32456 32639 34449 32855 32805 32856 32806 33022 33020 33255 33155 33466 33461 32857 32807 33041 33040 33256 33156 33468 33458 32858 32808 33044 33042 33257 33157 33474 33436

Attachment E: Remapped ZIP Codes

FCHLPM Remapped FCHLPM Remapped FCHLPM Remapped FCHLPM Remapped ZIP ZIP ZIP ZIP ZIP ZIP ZIP ZIP 32859 32809 33045 33040 33261 33161 33475 33455 32860 32810 33051 33050 33265 33175 33481 33431 32861 32811 33052 33050 33266 33166 33482 33445 32862 32827 33061 33060 33269 33169 33488 33433 32867 32817 33072 33069 33280 33162 33497 33428 32868 32818 33074 33064 33283 33173 33499 33487 32869 32809 33075 33067 33296 33156 33503 33598 32872 32822 33077 33071 33299 33122 33508 33511 32877 32837 33081 33021 33302 33304 33509 33511 32878 32828 33082 33028 33303 33301 33521 34785 32885 32827 33083 33023 33307 33334 33524 33540 32886 32801 33084 33024 33310 33311 33526 33525 32887 32821 33090 33030 33318 33322 33530 33567 32890 32809 33092 33032 33320 33321 33537 33523 32891 32803 33093 33063 33329 33324 33539 33542 32893 32814 33097 33073 33335 33334 33550 33584 32896 32801 33101 33128 33336 33311 33564 33566 32897 32801 33102 33122 33337 33324 33568 33569 32898 32801 33107 33122 33338 33324 33571 33573 32899 32815 33110 33150 33339 33306 33574 33576 32902 32901 33111 33131 33340 33311 33575 33570 32906 32905 33112 33122 33345 33351 33583 33584 32910 32907 33114 33134 33346 33316 33586 33570 32911 32905 33116 33176 33348 33308 33587 33527 32912 32904 33119 33139 33349 33304 33593 33523 32919 32901 33121 33131 33355 33325 33595 33594 32923 32922 33124 33181 33359 33319 33601 33602 32924 32922 33148 33131 33388 33324 33608 33621 32932 32931 33151 33127 33394 33301 33622 33607 32936 32935 33152 33122 33402 33480 33623 33607 32941 32901 33153 33138 33416 33406 33630 33607 32954 32953 33159 33139 33419 33404 33631 33607 32956 32955 33163 33180 33420 33410 33633 33607 32957 32958 33164 33162 33421 33411 33650 33607 32959 32927 33188 33187 33422 33417 33651 33602 32961 32960 33192 33193 33424 33426 33655 33602 32964 32960 33195 33125 33425 33426 33660 33619 32965 32962 33197 33157 33427 33486 33661 33619 32969 32966 33231 33131 33429 33432 33662 33619 32970 32967 33233 33133 33439 33438 33663 33607 32971 32967 33234 33134 33443 33441 33664 33607 32978 32958 33238 33150 33447 33444 33672 33602 33001 33050 33239 33139 33448 33446 33673 33603 33002 33014 33242 33142 33454 33467 33674 33604 33008 33009 33243 33143 33459 33440 33675 33605 33011 33010 33245 33145 33464 33460 33677 33607 33017 33015 33247 33122 33465 33462 33679 33629 33680 33610 33831 33830 34107 34102 34478 34474 33681 33611 33835 33860 34133 34135 34483 34472 33682 33612 33836 33837 34136 34135 34487 34448 33684 33634 33840 33803 34137 34114 34489 34488 33685 33615 33845 33844 34138 34141 34492 34491 33686 33616 33846 33813 34139 34141 34603 34601 33687 33617 33847 33830 34140 34114 34605 34601 33688 33618 33848 34758 34143 34142 34611 34606

Attachment E: Remapped ZIP Codes

FCHLPM Remapped FCHLPM Remapped FCHLPM Remapped FCHLPM Remapped ZIP ZIP ZIP ZIP ZIP ZIP ZIP ZIP 33689 33619 33851 33844 34146 34145 34636 34601 33690 33609 33854 33898 34204 34203 34656 34653 33694 33624 33855 33898 34206 34205 34660 34683 33697 33612 33856 33898 34216 34217 34661 34601 33728 33716 33858 33837 34218 34217 34673 34668 33729 33716 33862 33852 34220 34221 34674 34667 33730 33713 33863 33860 34230 34236 34679 34667 33731 33701 33867 33853 34250 34221 34680 34652 33732 33702 33871 33870 34260 34243 34681 34683 33733 33713 33877 33859 34264 34203 34682 34683 33734 33704 33882 33880 34265 34266 34697 34698 33736 33706 33883 33880 34267 34266 34712 34711 33737 33707 33885 33881 34268 34266 34713 34714 33738 33708 33888 33884 34270 34243 34729 34715 33740 33706 33902 33901 34272 34275 34740 34787 33741 33706 33906 33907 34274 34275 34742 34741 33742 33702 33910 33990 34276 34236 34745 34741 33743 33710 33911 33901 34277 34231 34749 34748 33744 33708 33915 33990 34278 34234 34755 34715 33747 33711 33918 33903 34280 34209 34760 34787 33757 33755 33930 33935 34281 34207 34770 34769 33758 33765 33932 33931 34282 34207 34777 34787 33766 33763 33938 33948 34284 34285 34778 34787 33769 33765 33944 33471 34295 34223 34789 34788 33775 33772 33945 33922 34421 34420 34948 34950 33779 33771 33949 33950 34423 34429 34954 34983 33780 33781 33951 33950 34430 34432 34958 34957 33784 33713 33965 33912 34445 34442 34973 34972 33802 33803 33970 33936 34447 34448 34979 34950 33804 33805 33975 33935 34451 34453 34985 34952 33806 33803 33994 33905 34460 34461 34991 34990 33807 33813 34101 34102 34464 34465 34992 34997 33820 33830 34106 34102 34477 34474 34995 34994 33826 33825