Probabilistic Earthquake Disaster Scenario for San José, Costa Rica

PROBABILISTIC EARTHQUAKE DISASTER SCENARIO FOR SAN JOSÉ, COSTA RICA

San José, Costa Rica - December 2016 Acknowledgement

This report forms part of the USAID/OFDA funded Preparing Rescue and Emergency Personnel to Ameliorate the Response to Earthquakes (USAID/OFDA PREPARE program). It was originally published under the title: Phase 1: Assessment of Earthquake Risks.

The USAID/OFDA PREPARE program has been made possible thanks to the support and generosity of the American people through the United States Agency for International Development (USAID) and its Office of Foreign Disaster Assistance (OFDA). Miyamoto International, Inc. administers and executes the resources of USAID and manages the implementation of the program in conjunction with the governments of Pasto, Colombia; Costa Rica; El Salvador and Mexico.

The goal of the program is to strengthen risk management policy and practice of national and municipal Disaster Risk Reduction/Management (DRR/DRM) institutions for a reduction in internally displaced people (IDPs), lives lost, less people injured and less economic disruption in the selected PREPARE cities: Pasto, Colombia; San José, Costa Rica; San Salvador, El Salvador and Guadalajara, Mexico.

This report is the result of a multi-stakeholder engagement and consultation process which involved authorities from the national and municipal level governments, academic institutions, and private sector partners.

We hereby acknowledge the contributions of, and thank the agencies, institutions and partners, for their valuable time, guidance and support.

December 2016

Submitted by

Miyamoto International, Inc.

www.miyamotointernational.com

DISCLAIMER: This publication is made possible by the support of the American People through the United States Agency for International Development (USAID). The opinions, findings and conclusions stated herein are those of the authors and do not necessarily reflect the views of USAID or the United States Government. Executive Summary Probabilistic seismic risk assessment for the canton of San José in Costa Rica was undertaken as part of Phase I of the Preparing Rescue and Emergency Personnel to Ameliorate the Response to Earthquakes (PREPARE) program. The objectives of the project were to assess the expected values of the building damage, fatalities (for both daytime and nighttime scenarios), and debris volume that would result from a potential earthquake. The findings can then be used to prepare policies or plans of action to lessen the human and socioeconomic impact that would be caused by future earthquakes in San José.

The current research data and available maps were used to develop the design-level (475-year return period) seismic hazard and the site conditions that were used as the probabilistic seismic input for risk analysis. As part of the risk assessment program, satellite imagery was examined and field surveys were conducted to collect data for a pool of representative buildings.

The exposure data for the area of study is presented in Table 1. The collected data was used to divide buildings into various groups of similar construction. This approach formed the basis for the exposure model. For each building typology, fragility functions were then developed. The consequence (structural damage, fatality, and debris) functions that correspond to each damage state (DS) for a given building type formed the last piece of input.

The input files were then used to run Monte Carlo simulations with the OpenQuake risk engine. The earthquake risk analysis was conducted for all building assets of the exposure model, and the risk results of each building asset were accumulated with respect to 196 barrios individually and for the canton of San José as a whole. The distribution for individual barrios is beneficial for government and city officials in allocating resources for risk preparation and mitigation.

Table 1. Exposure data No. of Built area, Occupants Occupants buildings km2 (daytime) (nighttime) 85,800 26.9 472,000 352,000

Table 2 summarizes the estimated earthquake risks of the entire canton of San José. Analysis showed that:

• The number of buildings that are expected to be yellow-tagged (moderately damaged) or red-tagged (severely damaged or collapsed) is estimated at about 51,000 structures, or approximately 60% of the building stock. • Depending on the time of the event, approximately 3,000 fatalities (for an estimated rate of 0.7%) is anticipated. • The generated debris volume of 4,940,000 m3 is significant and must be accounted for.

Table 2. Expected values of earthquake risk for the canton of San José Structural Fatalities Fatalities Yellow-tagged Red-tagged Debris damage daytime nighttime buildings buildings volume % Area, km2 % No. % No. % No. % No. (106) m3 42% 11.35 0.64% 3,000 0.76% 2,700 33% 28,000 26% 22,500 4.94

The spatial distribution of the 22,500 red-tagged buildings is presented in Figure 1. As shown in the figure, a large concentration of red-tagged buildings is in a few barrios.

The high physical damage and fatality rates from an earthquake that are computed in this report are not unex- pected, and they highlight the need for development of a risk mitigation program. As part of such a program, it is recommended that the following strategies be implemented:

• Provide a seismic strengthening program for key structures that are identified as exhibiting the most risk because of their inherent structural vulnerability, density of occupants, and importance. • Establish a post-earthquake damage assessment program. It is critical to train and certify engineers and to establish logistics. Such a program will improve response and recovery efforts after major earthquakes. • Establish communication and public outreach programs. It is critical to communicate results and the abovementioned recommendations. Communities should be informed about earthquake risk and risk reduction methods. • Optimize allocation of emergency response and recovery resources by identifying the most vulnerable regions.

The results, findings, and conclusions that are presented in this report are based on a seismic risk analysis derived from internationally recognized references and state-of-the-art analytical techniques. However, as with any engineering project, the underlying methods and analysis are based on certain assumptions and engineering judgment. Addi- tionally, the findings are based on a given design-level scenario earthquake intensity and correspond to the expected values or means. That is, the results present the expected outcome for an average event from a sample of a large pool of events with similar intensities. As such, the findings in this report include a certain level of uncertainty (inherent in risk assessment) and should not be extrapolated directly to a future seismic event. Accordingly, these assumptions and variations should be accounted for when interpreting the findings and applying the results for future planning.

SJ Phase 1 Report_2016-11-15 4 © 2016 Miyamoto International, Inc. Contents Executive Summary...... 3 1. Introduction...... 8 1.1 Project background...... 8 1.2 Phase I description...... 8 2. Earthquake Hazard for San José...... 12 2.1 Overview...... 12 2.2 Bedrock acceleration...... 12 2.3 Site class...... 12 2.4 Seismic design parameters...... 14 3. Exposure Model...... 15 3.1 Over view...... 15 3.2 Canton of San José building asset and occupancy distribution...... 15 3.3 Zones for surveyed buildings...... 18 3.4 Building typology...... 18 3.4.1 Overview...... 18 3.4.2 Building exposure inventory...... 19 4. Fragility and Damage Functions...... 22 4.1 Building fragility functions...... 22 4.1.1 FEMA Hazus default values...... 22 4.1.2 Fragility function modifications...... 23 4.1.3 Fragility parameters for San José buildings...... 24 4.2 Consequence functions...... 32 4.2.1 Structural damage...... 32 4.2.2 Fatalities...... 34 4.2.3 Debris volume...... 36 5. Risk Assessment Methodology...... 38 5.1 Over view...... 38 5.2 Risk analysis algorithm...... 38 5.3 Monte Carlo simulation (MCS)...... 39 5.3.1 Over view...... 39 5.3.2 Methodology...... 39 6. Risk Assessment Results...... 42 6.1 Over view...... 42 6.2 Findings...... 46 6.2.1 Over view...... 46 6.2.2 Expected human loss (fatalities)...... 46 6.2.3 Expected building damage (by colored tag categories)...... 47 6.2.4 Expected building damage (damage ratio and debris volume)...... 47 6.3 Risk distribution maps...... 51 6.4 Aggregated results...... 55 6.4.1 Overview...... 55 6.4.2 Physical damage and building tags...... 55 6.4.3 Fatalities...... 55 6.4.4 Discussion...... 55 7. Conclusions...... 56 8. References...... 58 Appendixes...... 59

SJ Phase 1 Report_2016-11-15 5 © 2016 Miyamoto International, Inc. List of Tables Table 1. Exposure data...... 3 Table 2. Expected values of earthquake risk for the canton of San José...... 3 Table 3. Measured shear velocities...... 13 Table 4. Key statistics for the canton of San José...... 15 Table 5. Building typology that was used in analysis...... 19 Table 6. Building fragility classifications...... 22 Table 7. FEMA Hazus default fragility function parameters...... 23 Table 8. Adjustment factors for standard deviations...... 24 Table 9. Modified Hazus fragility function parameters for San José...... 24 Table 10. Hazus default structural damage ratio...... 32 Table 11. Modified structural damage ratios that were used for San José...... 34 Table 12. FEMA Hazus default fatality ratios...... 34 Table 13. Modified fatality rates that were used for San José...... 36 Table 14. Debris volume ratios that were used for San José...... 37 Table 15. Key exposure data for barrios...... 42 Table 16. Exposure data for the canton of San José...... 55 Table 17. Expected values of structural loss...... 55 Table 18. Expected values of fatalities...... 55 Table 19. Building area and population exposure for the canton of San José...... 56 Table 20. Expected values of vulnerability for earthquake risk for the canton of San José...... 56

SJ Phase 1 Report_2016-11-15 6 © 2016 Miyamoto International, Inc. List of Figures Figure 1. Spatial distribution of red-tagged buildings...... 4 Figure 2. Boundaries of the canton of San José...... 9 Figure 3. Examples of building typologies in the canton of San José, Costa Rica...... 10 Figure 4. RESIS II PGA distribution of Costa Rica (Climent et al 2008)...... 12

Figure 5. Vs30 estimates (Schmidt et al. 2005)...... 13 Figure 6. RESIS II PGA around the canton of San José with soil amplification...... 14 Figure 7. Exposure model of the canton of San José (see Table 15 for barrio ID)...... 15 Figure 8. Spatial distribution of building assets...... 16 Figure 9. Spatial distribution of occupants...... 17 Figure 10. Zoning according to development type...... 18 Figure 11. Building composition based on building type (normalized)...... 20 Figure 12. Building composition based on development pattern (normalized)...... 20 Figure 13. Building composition based on number of stories (normalized)...... 21 Figure 14. Distribution of the 576 surveyed buildings...... 21 Figure 15. Fragility functions for nonengineered light structures...... 25 Figure 16. Fragility functions for unreinforced masonry...... 26 Figure 17. Fragility functions for confined/reinforced masonry...... 27 Figure 18. Fragility functions for reinforced concrete moment frame...... 28 Figure 19. Fragility functions for reinforced concrete shear wall...... 29 Figure 20. Fragility functions for steel moment frame...... 30 Figure 21. Fragility functions for steel braced frame...... 31 Figure 22. Fragility functions for unreinforced masonry informal area...... 32 Figure 23. Economic loss rates in the United States and in Costa Rica, by USGS PAGER .....33 Figure 24. Fatality rates in the United States and in Costa Rica, by USGS (USGS et al. 2009).35 Figure 25. Flowchart for analysis...... 38 Figure 26. Process flow using the OpenQuake risk engine...... 39 Figure 27. Investigation of the sufficient number of MCSs for this project...... 40 Figure 28. Distribution of MCS outcomes for a sample building...... 40 Figure 29. Flowchart of seismic damage estimation for this project...... 41 Figure 30. Distribution of fatality ratios for barrios...... 48 Figure 31. Sorted distribution of fatalities in barrios...... 48 Figure 32. Distribution of building damage categories for barrios...... 49 Figure 33. Sorted distribution of red-tagged and red-tagged + yellow–tagged buildings...... 49 Figure 34. Distribution of building damage ratios for barrios...... 50 Figure 35. Sorted distribution of debris volume...... 50 Figure 36. Spatial distribution of structural damage...... 52 Figure 37. Spatial distribution of fatalities...... 53 Figure 38. Spatial distribution of red-tagged buildings and debris volume...... 54

1.1 Project background

The PREPARE program intends to develop a new disaster risk reduction (DRR) and disaster risk management (DRM) program in the canton of San José, Costa Rica. The multiyear program, with financial support from the United States Agency for International Development/Office of U.S. Foreign Disaster Assistance (USAID/OFDA), includes cooperation and support of local Costa Rican and Colombian partner organizations. The targeted ben- eficiaries are the citizens of the canton of San José who live in zones that are at high risk for future earthquakes.

The PREPARE program aims to provide national and municipal DRR institutions with a clearer picture of the probable impact of an earthquake. The program also wants to help these institutions meet their goals of reducing casualties and lessening the socioeconomic impact of future earthquakes.

The two overarching PREPARE objectives are:

• To strengthen earthquake-response planning and preparedness of national and municipal DRR institutions in San José (Costa Rica) and Pasto (Columbia). This objective fits within the OFDA Geological Hazards subsector. • To strengthen the risk management policy and practice of national and municipal DRR institutions for a reduction in fatalities, injuries, financial costs, and economic disruptions. This objective fits within the OFDA Policy and Planning subsector and the Capacity Building and Training subsector. Three main PREPARE components are to be implemented during three phases:

• Phase I: Assessment of Earthquake Risks. Assess seismic hazards and seismic risk to determine the probabilistic damage to building structures and probable fatalities among the residents in each municipality. • Phase II: Analysis of Earthquake Scenarios and Planning for Response. Analyze earthquake scenarios based on the findings from risk assessments; review plans, policies, and practices for the response, including rapid damage assessments and debris management. • Phase III: Technical Training. Implement DRR training activities based upon a review of the results of the earlier phases. The aim is that after completing the PREPARE program, the partner organizations will have gained knowledge for conducting seismic risk assessments and analyzing earthquake scenarios, and will continue to improve their DRR and DRM capacity in the future. This report focuses on Phase I of the project.

1.2 Phase I description The canton of San José, Costa Rica, is in a high seismic zone and is at high risk for severe future earthquakes. Many of the newer buildings in the canton of San José have been constructed by using modern seismic codes, are well constructed, and meet high seismic standards (and thus are anticipated to perform satisfactorily during earthquakes). The canton also houses numerous older structures that are vulnerable to earthquake damage. Many older structures are not well built, especially in poorer neighborhoods. The most recent earthquake in Costa Rica was in 2012. It had a moment magnitude (Mw) of 7.6 and was centered in Nicoya. The 1991 Mw 7.8 Limón Earthquake resulted in nearly 50 fatalities and caused collapse of many buildings.

SJ Phase 1 Report_2016-11-15 8 © 2016 Miyamoto International, Inc. The map in Figure 2 indicates the boundaries of the canton of San José. This canton, including its districts, was the focus of this study.

Figure 2. Boundaries of the canton of San José

Figure 3 presents examples of some of the construction types in the canton of San José. The primary building types are as follows:

• Non-engineered (informal) construction is typically one- or two-story residential with a light frame and corrugated metal roofing. • Unreinforced masonry is typically one to two stories, was constructed before the adoption of new codes, and includes both residential and historic buildings. • Confined/reinforced masonry is newer construction, low-rise (three or fewer stories), and used as residential or commercial buildings, and includes wood or metal rafters and sheathings. • Reinforced concrete moment frame is low- to high-rise and mixed commercial and residential use, and includes masonry infill. • Reinforced concrete shear wall is low- to mid-rise and is used for both commercial and residential buildings (not shown in the figure) • Steel moment and braced frames are mid- to high-rise, are for commercial occupancy, and are in downtown and business districts. • Light metal is typically single story for industrial use and includes a corrugated metal roof and siding. • Unreinforced masonry (informal) is typically one- to two-story residential with wood or metal flooring/roofing and has minimal capacity and ductility.

Confined masonry Reinforced masonry

Concrete moment frame Steel frame

Light metal Unreinforced masonry informal area

Figure 3. Examples of building typologies in the canton of San José, Costa Rica

SJ Phase 1 Report_2016-11-15 10 © 2016 Miyamoto International, Inc. The PREPARE program is based upon OpenQuake, a suite of open-source software that allows the use of data and applications that are already developed or that are in process. The suite comprises the Platform, the Engine, and a variety of tools for modeling, accessing Global Earthquake Model (GEM) products, and sharing data and findings. GEM provides a set of tools and models for hazard and risk analysis, including the GEM Inventory Data Capture Tools (IDCT).

This set of software allows users to collect and modify building exposure information, which can then be added to the fast-growing Global Exposure Database (GED). The data in GED is more relevant to the building types and construction methods that are used in developing countries such as Costa Rica. The IDCT data collection tools relied on two sources: (1) satellite remote sensing analysis, and (2) direct observation on the ground by using mobile data collection tools. This data was then used to specify fragility functions that will be used for different building types and damage states (DSs).

2.1 Overview The seismic hazard that was evaluated by the RESIS II project (Climent et al. 2008) was used to develop the seismic hazard input (peak ground acceleration, or PGA) for the design-level (475-year) earthquake. The site response was prepared by using the bedrock acceleration and microzonation (site class) data for San José.

2.2 Bedrock acceleration Figure 4 shows PGA distribution of Costa Rica as estimated by RESIS II. As the map shows, the bedrock acceleration is in the range of 0.45g to 0.5g1 for San José.

Figure 4. RESIS II PGA distribution of Costa Rica (Climent et al 2008

2.3 Site class By using the Spectral-Analysis-of-Surface-Waves (SASW) method, the Seismic Engineering Laboratory of the En- gineering Research Institute of the University of Costa Rica measured the 30-m shear velocity (Vs30) for San José (Schmidt et al. 2005). Those authors estimated a bedrock PGA of 0.53g (which is consistent with the values shown in Figure 4). They also conducted SASW testing at several locations and classified the soil condition as stiff; see Figure 5.

*PGA values in the figure are shown in the gal (cm/sce2) units which is 1/981 g.

Figure 5. Vs30 estimates (Schmidt et al. 2005)

Their findings are reproduced in Table 3. Given the measured shear velocities, the soils can be classified as site class S3 per building code Colegio ( Federado 2010). For this soil condition, the code (CR 2010) prescribes an amplification factor of 1.2 to be applied to the bedrock PGA to obtain the site PGA.

Table 3. Measured shear velocities Location Vs30, m/sec Rohrmoser 232 Firestone 238 Napoleón Que- 250 sada Hospital 278 Hatillo 290 Desamparados 314 Parque Nacional 276

Figure 6 presents an enlarged map of PGA values around the canton of San José with a soil amplification factor. For the study area of the canton of San José, the site PGA was approximately 0.63g.

Figure 6. RESIS II PGA around the canton of San José with soil amplification

3.1 Overview The exposure model was developed through a statistical methodology by using the field survey of buildings (GEM 2014). Appendix B presents the technical report that ImageCat prepared, which details the development of the exposure model to be used as input for OpenQuake.

3.2 Canton of San José building asset and occupancy distribution The key data for the target area is presented in Table 4. The citywide map division is shown in Figure 7.

Table 4. Key statistics for the canton of San José Political divisions Buildings Occupants Barrios Districts No. Area, m2 Day Night 196 11 85,800 26,900,000 472,000 352,000

The spatial distribution of building assets and occupants for the target area is presented in Figure 8 and Figure 9, respectively.

Figure 7. Exposure model of the canton of San José (see Table 15 for barrio ID)

Number of buildings

Area of buildings

Figure 8. Spatial distribution of building assets

Daytime

Nighttime

Figure 9. Spatial distribution of occupants

SJ Phase 1 Report_2016-11-15 17 © 2016 Miyamoto International, Inc. 3.3 Zones for surveyed buildings The canton of San José was divided into numerous zones based on the following homogenous development pattern:

• Cemetery/open space • Informal (low-rise) • Industrial (industry and commercial, low-rise) • Single family (residential, low-rise) • Urban (residential and commercial, low- and mid-rise) • High urban ([densely populated] residential and commercial; low-, mid-, and high-rise) • Commercial (commercial, low- and mid-rise)

As seen in Figure 10, the target area was divided into 654 polygons (zones) according to the seven development types.

Figure 10. Zoning according to development type

Next, the zones of Figure 10 were coordinated with the barrios of Figure 7. Data was processed for each barrio and then was summed for the target area.

3.4 Building typology

3.4.1 Overview Buildings in the canton of San José were categorized into eight construction types based on the lateral-force-resisting system (LFRS) and the construction material. Seven of the construction types were further subdivided into two groups based on the number of stories. The resulting 15 building types are listed in

Table 5. Building typology that was used in analysis Type LFRS and material Stories 01 1–3 Nonengineered light structure 02 4+ 03 1–3 Unreinforced masonry 04 4+ 05 1–3 Confined/reinforced masonry 06 4+ 07 1–3 Reinforced concrete moment frame 08 4+ 09 1–3 Reinforced concrete shear wall 10 4+ 11 1–3 Steel moment frame 12 4+ 13 1–3 Steel braced frame 14 4+

15 Unreinforced masonry informal area 1–3

3.4.2 Building exposure inventory

A total of 2,575 building assets were included in the exposure pool that was used in the studies; see Table A.1. Data from these buildings was obtained by using satellite imagery and development patterns and supplemented by field surveys. This section provides information about the makeup of this pool of structures.

Figure 11 through Figure 13 present the distribution of buildings in the exposure model by using different criteria. The data is shown for both the number of structures and the total area of structures. For both cases, the data is normalized to either the total number or the area of buildings. In addition:

• Figure 11 presents the percentage of each type of building from Table 5 in the pool of exposure buildings. Note that the low-rise masonry and concrete buildings constitute the bulk of the buildings in the model. • Figure 12 presents the percentage of each type of development in the pool of exposure buildings. Note that the single-family and the urban are the prominent development types of the buildings in the model. • Figure 13 presents the percentage of building stories in the pool of exposure buildings. Note that the low- rise (one- to three-story) buildings contribute approximately 99% of the number of buildings and 89% of the area of buildings in the exposure model.

Figure 12. Building composition based on development pattern (normalized)

The building exposure model also included data from the 576 buildings that were surveyed in the field. The spatial distribution of these buildings is shown in Figure 14. Table A.2 presents detailed information for these surveyed buildings.

Figure 14. Distribution of the 576 surveyed buildings

4.1 Building fragility functions

4.1.1 FEMA Hazus default values The FEMA Hazus (FEMA 2001a) methodology was used to classify the building types into various fragility bins; see Table 6.

Table 6. Building fragility classifications FEMA Hazus notation Type LFRS and material Height1 Classification Code2 01 L W1/L/S3 Low-code Non-engineered light structure 02 M and H W1/M/S3 Low-code 03 L URML Low-code Unreinforced masonry 04 M and H URMM Low-code Moder- 05 L RM1L ate-code Confined/reinforced masonry Moder- 06 M and H RM1M ate-code Moder- 07 L C1L Reinforced concrete moment ate-code frame Moder- 08 M and H C1M ate-code Moder- 09 L C2L ate-code Reinforced concrete shear wall Moder- 10 M and H C2M ate-code Moder- 11 L S1L ate-code Steel moment frame Moder- 12 M and H S1M ate-code Moder- 13 L S2L ate-code Steel braced frame Moder- 14 M and H S2M ate-code Unreinforced masonry informal 15 L URML Precode area

The default (U.S.) values for the PGA fragility functions for the stated Hazus building types for various damage states (DSs) are presented in Table 7. FEMA Hazus provides a descriptive narrative for each DS and for each building type. For example, for the reinforced concrete moment frame building (C1), the document states:

Slight DS: Flexural or shear type hairline cracks in some beams and columns near joints or within joints.

L, M, and H denote low, moderate, and high for 1–3, 4–7, and 8+ stories. The code designation represents the expected relative compliance in design, detailing, and construction with modern seismic codes. SJ Phase 1 Report_2016-11-15 22 © 2016 Miyamoto International, Inc. Moderate DS: Most beams and columns exhibit hairline cracks. In ductile frames some of the frame elements have reached yield capacity indicated by larger flexural cracks and some concrete spalling. Nonductile frames may exhibit larger shear cracks and spalling. Extensive DS: Some of the frame elements have reached their ultimate capacity indicated in ductile frames by large flexural cracks, spalled concrete and buckled main reinforcement; nonductile frame elements may have suffered shear failures or bond failures at reinforcement splices, or broken ties or buckled main reinforcement in columns which may result in partial collapse. Complete DS: Structure is collapsed or in imminent danger of collapse due to brittle failure of nonductile frame elements or loss of frame stability. Approximately 13% (low-rise), 10% (mid- rise) or 5% (high-rise) of the total area of C1 buildings with Complete damage is expected to be collapsed.

Table 7. FEMA Hazus default fragility function parameters DS median, g3 Type Ln (Std. dev.) DS1 DS2 DS3 DS4 01 0.14 0.22 0.37 0.49 0.64 02 0.10 0.15 0.29 0.39 0.64 03 0.14 0.20 0.32 0.46 0.64 04 0.10 0.16 0.27 0.46 0.64 05 0.22 0.30 0.50 0.85 0.64 06 0.18 0.26 0.51 1.03 0.64 07 0.16 0.23 0.41 0.77 0.64 08 0.13 0.21 0.49 0.89 0.64 09 0.18 0.30 0.49 0.87 0.64 10 0.15 0.26 0.55 1.02 0.64 11 0.15 0.22 0.42 0.80 0.64 12 0.13 0.21 0.44 0.82 0.64 13 0.20 0.26 0.46 0.84 0.64 14 0.14 0.22 0.53 0.97 0.64 15 0.13 0.17 0.26 0.37 0.64

4.1.2 Fragility function modifications To account for the construction in Costa Rica and the seismic source, the FEMA Hazus default fragility median and standard deviations were modified as explained in this section. The Hazus default median values correspond to the following: Western United States (WUS) spectral shape, M7 earthquake, site class D, and more than 80 km from the rupture. By contrast, for San José, the design earthquake has the following characteristics: WUS spectral shape, average magnitude of 6.5 (Climent et al. 2008), site class D (S3 in CR 2010) with a soil amplification (Fv) of 1.5, and less than 10-km epicentral distance. Thus, the correction factor for the median values from the ground motion variation is computed from:

Eq. 1. =Factor = (Spectral shape at ~10 km)*(1.5/Fv)=1.8*(1.5/1.5)=1.8

Furthermore, FEMA Hazus values were developed for U.S. construction and for U.S. code provisions. To account for the variation between U.S. and San José construction, a median reduction value of 0.85 was applied to the median values.

* DS1=Slight, DS2=Moderate, DS3=Extensive, DS4=Complete

SJ Phase 1 Report_2016-11-15 23 © 2016 Miyamoto International, Inc. Assuming that the uncertainties are independently distributed, the total uncertainty can then be computed from the square root of the sum of the squares of hazard and quality uncertainties, or:

Table 8 presents the modification forEq. the 2. standard β_TOT=√(β_SH^2+β_DC^2 deviation. )

Table 8. Adjustment factors for standard deviations Parameter Hazus default San José 0.5 0.5 bSH Seismic hazard Design/construction β 0.4 0.4/0.85 DC quality 0.64 0.687 bTOT Total

4.1.3 Fragility parameters for San José buildings

The modified fragility function parameters that are suitable for San José and that are used in analysis are presented in Table 9.

Table 9. Modified Hazus fragility function parameters for San José DS median, g Std. dev. Type DS1 DS2 DS3 DS4 (ln) 01 0.209 0.332 0.571 0.750 0.687 02 0.146 0.224 0.449 0.592 0.687 03 0.214 0.306 0.490 0.704 0.687 04 0.153 0.245 0.413 0.704 0.687 05 0.337 0.459 0.765 1.301 0.687 06 0.275 0.398 0.780 1.576 0.687 07 0.245 0.352 0.627 1.178 0.687 08 0.199 0.321 0.750 1.362 0.687 09 0.275 0.459 0.750 1.331 0.687 10 0.230 0.398 0.842 1.561 0.687 11 0.230 0.337 0.643 1.224 0.687 12 0.199 0.321 0.673 1.255 0.687 13 0.306 0.398 0.704 1.285 0.687 14 0.214 0.337 0.811 1.484 0.687 15 0.199 0.260 0.398 0.566 0.687

SJ Phase 1 Report_2016-11-15 24 © 2016 Miyamoto International, Inc. Figure 15 through Figure 22 present the sets of plots for the various DS fragility functions for the 15 types of buildings that are studied in this report. The figures were generated by using the San José-modified parameters of Table 9.

Low-rise (type 01)

Mid- or high-rise (type 2)

Figure 15. Fragility functions for nonengineered light structures

Low-rise (type 03)

Mid- or high-rise (type 04)

Figure 16. Fragility functions for unreinforced masonry

Mid- or high-rise (type 06)

Figure 17. Fragility functions for confined/reinforced masonry

Mid- or high-rise (type 08)

Figure 18. Fragility functions for reinforced concrete moment frame

Mid- or high-rise (type 10)

Figure 19. Fragility functions for reinforced concrete shear wall

Mid- or high-rise (type 12)

Figure 20. Fragility functions for steel moment frame

1.0

0.9

0.8

0.7

0.6

0.5

0.4 DS1 0.3 DS2

Probability of Exceeding a DS 0.2 DS3 0.1 DS4 0.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 PGA (g)

Mid- or high-rise (type 14)

Figure 21. Fragility functions for steel braced frame

Figure 22. Fragility functions for unreinforced masonry informal area

4.2 Consequence functions 4.2.1 Structural damage The consequence functions that relate structural damage to DS were based on the values provided by FEMA Hazus and modified for San José. Table 10 presents the FEMA Hazus default physical (structural) damage ratios for the building types that are under consideration.

Table 10. Hazus default structural damage ratio DS Type DS1 DS2 DS3 DS4 01 0.02 0.10 0.40 1.00 02 0.02 0.10 0.40 1.00 03 0.02 0.10 0.40 1.00 04 0.02 0.10 0.40 1.00 05 0.02 0.10 0.40 1.00 06 0.02 0.10 0.40 1.00 07 0.02 0.10 0.40 1.00 08 0.02 0.10 0.40 1.00 09 0.02 0.10 0.40 1.00 10 0.02 0.10 0.40 1.00 11 0.02 0.10 0.40 1.00 12 0.02 0.10 0.40 1.00 13 0.02 0.10 0.40 1.00 14 0.02 0.10 0.40 1.00 15 0.02 0.10 0.40 1.00

For application to San José, the structural damage ratios must be modified. The structural damage ratio can be considered proportional to the economic loss and thus can be computed by using the research data for the economic loss rate.

The economic loss rate, r, is defined as:

Eq. 3. r=(Economic loss)/(Economic exposure)

The building loss rate that corresponds to a given building damage state was obtained by applying the regional modification factor to the FEMA Hazus model. The modification factors were estimated based on work by U.S. Geological Survey (USGS) researchers (USGS et al. 2011). Specifically, because the structural loss rates that were developed in Hazus were based on U.S. statistical data, the regional modification factors to convert the Hazus model (i.e., the U.S.) to Costa Rica had to be developed. Figure 23 shows the loss rates with respect to the Modified Mercalli Intensity (MMI) that were developed by USGS and that were used for this modification.

Eq. 4. rCR = rUS * (Regional modification factor)

Figure 23. Economic loss rates in the United States and in Costa Rica, by USGS PAGER (USGS et al. 2011)

For an MMI 8.5 to 9.0 (PGA 0.51g 0.62g) earthquake in Costa Rica, the figure estimates a damage ratio of 25% to 45%. Also note that the damage ratio is higher in Costa Rica than in the United States. However, this value must be calibrated based on previous earthquake surveys. For example, during the 2009 M6.2 San José and 2016 M7.8 Ecuador earthquakes, the damage ratios were approximately 6.7% and 15%, respectively. Thus, the estimated 25% to 45% damage ratio is reasonable.

The default Hazus damage ratios were therefore modified as follows: (i) Light frame and steel buildings: no change and (ii) concrete and masonry buildings: apply a regional modification factor of 1.125 to DS1 through DS3.

The revised damage ratios that were used in this study are presented in Table 11.

SJ Phase 1 Report_2016-11-15 33 © 2016 Miyamoto International, Inc. Table 11. Modified structural damage ratios that were used for San José DS Type DS1 DS2 DS3 DS4 01 0.02 0.10 0.40 1.00 02 0.02 0.10 0.40 1.00 03 0.0225 0.1125 0.45 1.00 04 0.0225 0.1125 0.45 1.00 05 0.0225 0.1125 0.45 1.00 06 0.0225 0.1125 0.45 1.00 07 0.0225 0.1125 0.45 1.00 08 0.0225 0.1125 0.45 1.00 09 0.0225 0.1125 0.45 1.00 10 0.0225 0.1125 0.45 1.00 11 0.02 0.10 0.40 1.00 12 0.02 0.10 0.40 1.00 13 0.02 0.10 0.40 1.00 14 0.02 0.10 0.40 1.00 15 0.0225 0.1125 0.45 1.00

4.2.2 Fatalities The consequence functions that relate fatality rate to DS were based on the values provided by FEMA Hazus and modified for San José. Table 12 presents the FEMA Hazus default fatality ratios for the building types that are under consideration.

Table 12. FEMA Hazus default fatality ratios DS Type DS1 DS2 DS3 DS4 01 0 0.0000007 0.000022 0.00316 02 0 0.0000007 0.000022 0.00316 03 0 0.0000260 0.000052 0.03284 04 0 0.0000260 0.000052 0.03284 05 0 0 0.000024 0.02167 06 0 0 0.000026 0.01918 07 0 0 0.000022 0.02167 08 0 0 0.000024 0.01918 09 0 0 0.000022 0.02167 10 0 0 0.000024 0.01918 11 0 0 0.000022 0.01418 12 0 0 0.000024 0.01169 13 0 0 0.000022 0.01418 14 0 0 0.000024 0.01169 15 0 0.0000260 0.000052 0.03284

The fatality rate, v, is defined as:

Eq. 5 v=Fatalities/(Population exposure)

The fatality rate that corresponds to a given building damage state was obtained by applying the regional modification factor to the FEMA Hazus model. The modification factors were estimated based on work by USGS researchers (USGS et al. 2009). Specifically, because the fatality rates that were developed in Hazus were based on U.S. statistical data, the regional modification factors to convert the Hazus model (i.e., the U.S.) to Costa Rica had to be developed. Figure 24 shows the fatality rates with respect to the MMI that were developed by USGS and that were used for this modification.

Eq. 6 vCR = vUS * (Regional modification factor)

Figure 24. Fatality rates in the United States and in Costa Rica, by USGS (USGS et al. 2009)

For an MMI 8.5 to 9.0 (PGA 0.51g to 0.62g) earthquake in Costa Rica, the figure estimates a fatality rate of 0.1% to 0.4%. Also, note that the fatality rate is higher in Costa Rica than in the United States. However, this value must be calibrated based on previous earthquake surveys. For example, during the 1910 M6.7 San José, 2009 M6.2 San José, and 2016 M7.8 Ecuador earthquakes, the fatality rates were approximately 5.4%, 0.28%, and 0.02% to 2.2%, respectively. Thus, the estimated 0.1% to 0.4% fatality rate is reasonable.

The default Hazus fatality ratios were therefore modified by multiplying fatality rates by a factor of 1.292 (the average value computed from the previously mentioned earthquakes and extrapolated to M8.5 to M9.0).

Table 13. Modified fatality rates that were used for San José DS Type DS1 DS2 DS3 DS4 01 0 0.0000009 0.0000284 0.0040817 02 0 0.0000009 0.0000284 0.0040817 03 0 0.0000336 0.0000672 0.0424183 04 0 0.0000336 0.0000672 0.0424183 05 0 0 0.0000310 0.0279904 06 0 0 0.0000336 0.0247742 07 0 0 0.0000284 0.0279904 08 0 0 0.0000310 0.0247742 09 0 0 0.0000284 0.0279904 10 0 0 0.0000310 0.0247742 11 0 0 0.0000284 0.0183158 12 0 0 0.0000310 0.0150996 13 0 0 0.0000284 0.0183158 14 0 0 0.0000310 0.0150996 15 0 0.0000336 0.0000672 0.0424183

4.2.3 Debris Volume The debris volume was evaluated based on the methodology that was developed for the 2010 Haiti Earthquake and was calibrated with the measured volumes during that event. The methodology was modified to account for San José construction types.

The calculation was based on the following:

1 = + + Eq. 7 w *c f N(c s At c w Lht w c m ) A Where:

• w = the unit debris volume in m3 per m2 of building area • A = building footprint in m2 • N = number of stories • At = floor slab volume per stories • Lhtw = wall volume per stories • cf, cs, cw, cm = modification factors

Table 14. Debris volume ratios that were used for San José Debris volume, Type m3/(m2 of floor area)/ story 01 0.398 02 0.398 03 0.830 04 0.830 05 0.830 06 0.830 07 0.830 08 0.830 09 0.830 10 0.830 11 0.453 12 0.453 13 0.453 14 0.453 15 0.830

5.1 Over view Figure 25 presents the flowchart that was used in the seismic risk assessment for San José. As the flowchart shows, the seismic hazard, building fragility data, and exposure model are input to the processing engine. For this project, the OpenQuake engine that was developed by the Global Earthquake Model Foundation (GEM 2016) was used as the processing module.

Figure 25. Flowchart for analysis

5.2 Risk analysis algorithm The probabilistic risk assessment used Monte Carlo simulations (MCSs). To obtain convergence in results, 10,000 MCSs were performed. The risk analysis procedure was as follows (see Figure 26):

• Select a scenario earthquake:

o Use fragility and exposure data. o Run OpenQuake engine and determine the DS distributions. o By using the consequence functions and the obtained DSs, compute structural damage, fatality, and debris volume. • Repeat the process 10,000 times. • Compute the expected value (mean) for quantities of interest. • Post-process and aggregate for the barrios and the canton of San José.

Figure 26. Process flow using the OpenQuake risk engine

5.3 Monte Carlo simulation (MCS

5.3.1 Over view In analysis, the seismic intensity (PGA) was assumed to have lognormal distribution, and the random variables of PGA that were based on the distribution were generated for all MCS cases (10,000 variables at each location). The building damage probability was then estimated for each case by each PGA variable, and the building fragility function was represented by lognormal distribution. This analysis process was repeated for the specified number of MCSs. Upon completion, the mean (expected) value of building damage of all the simulation cases was obtained for each building by applying the consequence function, and these damage values were aggregated according to their corresponding barrio.

5.3.2 Methodology The seismic damage due to a target earthquake intensity is probabilistically estimated through building damage by applying building fragility and PGA distribution. Because seismic damage estimation contains several uncertainties, a probabilistic estimation must be performed to obtain the expected damage by using either a theoretical approximation method or a numerical simulation method.

For this project, MCS, a numerical simulation method, was used to analyze the seismic damage. The Monte Carlo technique is one of the computational simulation approaches that relies on random sampling to obtain numerical results. The main concept of MCS is to estimate the mean value and the variability of the response of a complex system by using a reasonable subset of the solution space. The subspace is determined by sampling the original space, which means that numerous simulations are required to obtain a reliable result.

For this project, the appropriate number of simulations to achieve a reasonable result, one that converges on the mean value of seismic damage, had to be found. The results for a range of simulation runs are presented in Figure 27. Inspection of the figure reveals that the results vary substantially when only small numbers of simulations (e.g., 5, 10, and 50 realizations) are run. Likewise, the solution is still non-convergent with 100 and 500 simulations. The simulation results converge at approximately 10,000 realizations (i.e., the result for 10,000 simulations and beyond is consistent).

SJ Phase 1 Report_2016-11-15 39 © 2016 Miyamoto International, Inc. From this convergence investigation, 10,000 simulations was found to yield reasonably stable results and was selected for the MCS for this project. Here, the building damage state was assumed randomly and was probabilistically distributed in the simulation for each realization, and the mean value of seismic damage was then calculated by applying consequence functions.

Figure 27. Investigation of the sufficient number of MCSs for this project

Figure 28 presents MCS outcomes for a sample building2. For this particular building, out of 10,000 simulations, approximately 1,400, 800, 2,350, 3,600, and 1,850 outcomes fall into the No Damage, Slight, Moderate, Exten- sive, and Complete damage states, respectively. The distributions vary from building to building depending on the site seismicity and the building fragility

Figure 28. Distribution of MCS outcomes for a sample building

The flowchart shown in Figure 29 illustrates the damage estimation procedure using MCS for this project. As discussed earlier in this report, the mean (i.e., expected) value of structural damage, fatalities, and debris volume will be the key risk assessment parameters that are used in the evaluation.

**DSO implies no damage

6. Risk Assessment Results 6.1 Over view For each building, the estimated loss (structural damage, fatalities, and debris volume) was computed based on the expected value (mean) from 10,000 Monte Carlo simulations. The analysis results were expressed in relative values (damage probability). This data was then converted to damaged area, fatalities, and debris volume by using the building footprint, number of stories, and number of occupants. Then these numbers were aggregated for each barrio and were summed over barrios to obtain values for the canton of San José. In addition, the different damage states were grouped into three categories corresponding to the expected level of post-earthquake damage, such as green-, yellow-, and red-tagged (FEMA 2001b).

Key exposure data (building number, total area, and occupants) for the 196 barrios and the canton of San José is provided in Table 15. Refer to Table A.3 for a more detailed exposure database for the barrios.

Table 15. Key exposure data for barrios Barrio No. of Building area, Occupants No. Barrio name ID buildings m2 Day Night 1 QUINCE DE SETIEMBRE 1012 1,111 197,095 1,662 5,825 2 VEINTICINCO DE JULIO 1016 697 190,587 2,862 4,745 3 AEROPUERTO 99091 55 103,459 2,090 1,891 4 AHOGADOS (parte) 601 63 28,193 464 285 5 ALFA 901 54 22,916 603 148 6 ALMENDARES 301 154 34,429 938 848 7 AMERICAS 801 464 348,535 9,129 1,438 8 AMON 101 279 149,860 3,845 608 9 ANGELES 302 503 143,909 3,835 1,841 10 ANONOS 9908 154 29,289 461 726 11 ARANJUEZ 102 358 208,072 5,373 968 12 ARBOLES 703 450 385,560 9,112 1,527 13 ASTURIAS 902 106 13,696 174 1,635 14 ASUNCION 903 663 236,125 1,854 1,415 15 BAJO CAÃ‘ADA 99111 55 13,602 297 272 16 BAJO TORRES 704 58 21,197 441 625 17 BAJOS DE LA UNION 201 156 85,519 2,073 745 18 BALCON VERDE 803 166 75,130 1,120 558 19 BELGRANO 1002 255 53,900 388 524 20 BELLA VISTA 401 31 21,818 569 34 21 BENGALA 1101 495 95,023 769 2,693 22 BILBAO 1102 356 71,693 983 1,047 23 BOLIVAR 303 219 92,509 1,803 1,103 24 BRIBRI 904 1,240 186,926 2,590 6,130 25 CALDERON MUÃ‘OZ 502 292 96,520 1,664 1,626 26 CALIFORNIA (parte) 103 130 75,213 1,969 254 27 CAMELIAS 99112 88 26,544 542 344 28 CAÃ‘ADA SUR 1103 329 66,374 782 1,430

SJ Phase 1 Report_2016-11-15 42 © 2016 Miyamoto International, Inc. Table 15. Key exposure data for barrios Barrio No. of Building area, Occupants No. Barrio name ID buildings m2 Day Night 29 CARIT 304 109 27,538 466 588 30 CARLOS MARIA JIMENEZ 403 87 21,659 156 405 31 CARMEN 1104 822 264,138 3,062 3,371 32 CARMEN 104 330 355,612 9,276 572 33 CARRANZA 705 181 48,423 877 610 34 CASCAJAL 1105 341 96,220 1,180 1,348 35 CENTRO CATEDRAL 9904 273 206,797 5,396 373 36 CENTRO MERCED 9902 324 277,566 7,265 933 37 CERRITO 503 221 36,305 484 904 38 CERRO AZUL 1106 94 20,404 146 174 39 CLARET 202 369 89,702 1,216 2,064 40 COCACOLA 203 80 54,422 1,212 286 41 COLOMBARI 1107 293 53,189 401 879 42 COLON (parte) 804 245 107,942 1,911 999 43 KENNEDY 1112 760 199,001 2,241 2,695 44 SANTANDER 718 77 15,861 419 135 45 CORAZON DE JESUS 306 195 27,033 209 547 46 CORAZON DE JESUS 706 161 26,888 712 716 47 CORDOBA 504 568 170,580 1,548 1,888 48 CRISTAL 707 608 182,303 2,443 2,064 49 CRISTO REY 307 1,206 301,351 6,506 6,608 50 CUBA 308 613 234,463 4,232 2,596 51 DEL PINO 805 57 14,883 123 451 52 DOLOROSA (parte) 404 286 128,031 3,361 620 53 DOMINGO SABIO 1108 151 27,166 243 1,210 54 DOS PINOS 405 12 22,321 596 307 55 BOSQUE 602 922 197,493 1,480 3,102 56 ELECTRIONA 99071 35 22,339 267 354 57 EMPALME 105 138 64,691 1,700 352 58 ESCALANTE 106 512 264,216 6,937 1,321 59 FARO 606 240 53,541 643 748 60 FATIMA 607 360 118,790 1,696 494 61 FAVORITA NORTE 905 513 161,737 1,125 1,465 62 FAVORITA SUR 906 561 248,778 2,998 1,561 63 LA CAJA 710 334 165,754 4,257 233 64 FLORENTINO CASTRO 99072 75 64,976 1,640 783 65 FRANCISCO PERALTA (parte) 406 278 149,547 3,918 577 66 GALICIA 907 57 8,603 140 1,635 67 GEROMA 908 1,537 503,389 4,456 4,639 68 GONZALEZ LAHMANN 407 568 259,892 6,811 1,039 69 GONZALEZ VIQUEZ 408 294 150,088 3,942 772 70 GUACAYAMA 1109 49 37,584 894 296 71 GUELL 409 328 73,899 1,331 1,410 72 HATILLO 1 1004 646 143,250 1,089 2,778 73 HATILLO 2 1005 1,177 291,760 2,954 4,807 74 HATILLO 3 1006 861 190,428 1,465 3,735

SJ Phase 1 Report_2016-11-15 43 © 2016 Miyamoto International, Inc. Table 15. Key exposure data for barrios Barrio No. of Building area, Occupants No. Barrio name ID buildings m2 Day Night 75 HATILLO 4 1007 1,043 205,580 1,658 3,986 76 HATILLO 5 1008 1,263 148,288 1,191 4,699 77 HATILLO 6 1009 1,332 158,310 1,388 6,795 78 HATILLO 7 1010 626 130,758 1,112 2,728 79 HATILLO 8 1011 2,208 205,019 2,130 8,948 80 HATILLO CENTRO 1003 569 270,373 4,917 2,432 81 HISPANIA 909 49 7,666 102 1,001 82 HISPANO 608 393 111,705 1,494 1,621 83 HOGAR PROPIO 1111 93 20,851 148 163 84 HOLANDA 806 98 53,107 920 362 85 IGLESIAS FLORES 204 333 58,272 467 1,550 86 JARDIN 506 386 113,616 1,908 1,848 87 JARDINES DE AUTOPISTA 709 428 127,493 2,138 1,360 88 JAZMIN 1110 205 45,078 648 1,222 89 LA ARBOLEDA 9905 90 18,440 134 213 90 CABAÃ‘AS 603 198 46,818 354 803 91 LA CARPIO 711 3,531 993,889 21,555 21,555 92 LA CRUZ 410 433 106,826 1,231 1,707 93 GLORIA 505 571 121,159 1,761 2,277 94 PEREGRINA 715 955 172,590 1,343 3,234 95 LA SALLE 808 413 212,749 3,702 1,781 96 ANIMAS 702 114 215,782 4,020 828 97 LUISAS 507 394 75,547 574 1,373 98 LIBERTAD 910 616 88,745 750 3,079 99 LINCOLN 610 502 153,062 1,110 1,661 100 LLANOS DEL SOL 912 585 197,207 3,602 1,496 101 LOMALINDA 809 90 33,125 227 118 102 LOMAS DE OCLORO 411 155 25,019 342 763 103 LOMAS DE SAN FRANCISCO 611 181 27,928 208 671 104 LOMAS DEL RIO 911 1,493 210,131 1,712 6,264 105 LOPEZ MATEO 1113 1,083 237,454 2,460 5,763 106 MANGOS 508 70 38,177 984 137 107 LUJAN 412 958 263,768 5,854 3,335 108 LUNA PARK 1114 330 93,177 1,167 2,309 109 MAALOT 612 246 91,864 1,339 1,096 110 MAGNOLIAS 99113 267 55,343 421 997 111 MANTICA 205 264 171,134 4,491 820 112 MARIA REINA 913 155 30,206 246 870 113 MARIMIL 713 26 24,811 650 104 114 MENDEZ 613 114 30,805 239 639 115 MERCED 310 547 501,282 13,090 1,083 116 METROPOLIS 914 4,349 397,835 5,045 12,120 117 MEXICO 206 1,718 530,681 10,345 6,966 118 MILFLOR 413 89 22,480 596 230 119 MOJADOS 1116 348 50,462 402 1,371

SJ Phase 1 Report_2016-11-15 44 © 2016 Miyamoto International, Inc. Table 15. Key exposure data for barrios Barrio No. of Building area, Occupants No. Barrio name ID buildings m2 Day Night 120 MONGITO 1117 228 49,340 430 1,138 121 MONSERRAT 714 36 21,838 587 739 122 MONTE AZUL 1118 465 106,864 963 1,224 123 MONTEALEGRE 509 428 159,455 2,009 1,352 124 MORENO CAÃ‘AS 510 886 201,661 2,162 5,394 125 MORENOS 810 571 232,229 5,377 1,922 126 MUSMANI 1119 213 50,905 492 766 127 NACIONES UNIDAS 414 282 108,511 2,107 857 128 NAVARRA 915 124 17,482 200 1,635 129 NIZA 811 139 51,985 1,003 323 130 OTOYA 107 110 42,317 1,114 261 131 PACIFICA 614 1,121 325,780 2,333 3,778 132 PACIFICO (parte) 311 249 171,001 4,045 675 133 PARQUE LA PAZ 99114 13 5,079 55 166 134 PASO ANCHO 1120 931 216,992 2,774 4,823 135 PASO DE LA VACA 207 145 78,181 2,066 659 136 PAVAS (centro) 916 2,106 655,141 8,599 5,585 137 PINOS 312 462 70,500 1,283 2,223 138 PITAHAYA 208 704 281,787 6,650 2,086 139 PRESIDENTES 1121 281 89,436 1,576 1,188 140 PUEBLO NUEVO 917 158 24,910 215 959 141 QUESADA DURAN 511 361 103,309 1,549 1,278 142 RANCHO LUNA 812 110 72,753 1,906 493 143 RESIDENCIAL DEL OESTE 918 638 88,103 708 2,570 144 RINCON GRANDE 919 292 35,793 1,576 1,635 145 ROBLEDAL 716 446 161,257 1,576 853 146 ROHRMOSER (parte) 920 478 303,745 7,921 1,330 147 ROHRMOSER (parte) 813 358 236,331 6,173 547 148 ROMA 814 92 33,851 413 267 149 ROSITER CARBALLO 717 283 34,510 242 1,117 150 SABANA 815 41 61,295 877 486 151 SAGRADA FAMILIA 1013 1,116 224,877 2,289 5,961 152 SALUBRIDAD 313 456 66,331 581 2,701 153 SAN BOSCO 314 650 293,956 7,734 1,784 154 SAN CAYETANO (parte) 416 623 147,285 1,288 2,337 155 SAN DIMAS 512 58 25,689 675 132 156 SAN FRANCISCO DE DOS RIOS (centro) 615 669 294,592 6,371 3,703 157 SAN GERARDO (parte) 513 100 28,211 360 171 158 SAN MARTIN 1123 218 62,694 730 1,624 159 SAN PEDRO 922 1,297 165,034 3,180 7,986 160 SAN SEBASTIAN (centro) 1124 876 314,663 6,304 3,826 161 SANTA BARBARA 923 251 565,050 12,776 2,242 162 SANTA CATALINA 924 1,073 292,164 2,482 3,721 163 SANTA FE 925 219 49,704 340 510 164 SANTA LUCIA 316 234 99,195 2,603 462

SJ Phase 1 Report_2016-11-15 45 © 2016 Miyamoto International, Inc. Table 15. Key exposure data for barrios Barrio No. of Building area, Occupants No. Barrio name ID buildings m2 Day Night 165 SANTA ROSA 1125 333 79,712 1,098 1,758 166 SATURNO 719 24 25,138 668 280 167 SAUCES 616 810 270,876 2,891 2,969 168 SAUCITOS 617 127 21,417 164 396 169 SEMINARIO 1126 9 36,288 901 187 170 SILOS 317 63 70,782 1,728 387 171 SOLEDAD 417 381 150,584 3,938 441 172 TABACALERA 418 66 29,163 762 67 173 TIRIBI 1014 37 9,550 64 366 174 TOVAR 816 67 16,981 465 469 175 TREBOL 514 171 52,237 558 570 176 TRIA NGULO 926 242 77,475 553 655 177 UJARRAS 515 269 74,742 930 1,142 178 URUCA (centro) 720 529 472,088 11,044 1,769 179 VASCONIA 419 388 83,824 1,122 870 180 VILLA ESPERANZA 927 1,196 211,374 2,121 9,822 181 VUELTA DEL VIRILLA 721 96 56,948 1,077 549 182 I GRIEGA 609 284 94,267 2,518 1,325 183 YOSES SUR 517 410 155,889 3,179 1,066 184 ZAPOTE (centro) 518 462 199,387 4,313 813 185 ZONA INDUSTRAL BARZUNA 9903 110 176,441 4,508 766 186 ZONA INDUSTRIAL 99073 468 779,189 18,402 2,536 187 ZONA INDUSTRIAL PAVAS OESTE 99092 239 264,801 5,159 2,059 188 SOROBARU 1129 361 65,951 502 1,230 189 ZURQUI 618 91 20,300 314 312 190 SAN FRANCISCO 315 215 96,851 2,532 261 191 LA LUISA 807 62 11,672 246 246 192 CAMELIAS 9906 57 22,343 595 271 193 CALIFORNIA (parte) 402 92 41,751 1,094 165 194 DOLOROSA (parte) 309 197 73,908 1,927 111 195 PACIFICO (parte) 415 274 151,410 3,980 855 196 COLON (parte) 305 210 90,631 1,968 692 San José Canton 85,800 26,900,000 472,000 352,000

6.2.1 Over view The following sections present aggregated data for the barrios and the canton of San José. Both percentages and absolute values are presented. The expected values of physical damage and human loss for building assets and for barrios are presented in Table A.4 and Table A.5, respectively.

6.2.2 Expected human loss (fatalities) Because it is not known when an earthquake will occur, the fatality estimates in this section are based on the average values of daytime and nighttime fatalities for each barrio. Figure 30 presents the expected distribution of fatality ratios for each barrio. Note that the distribution of fatality ratios varies from 0.3% to 2.4%, with a mean value of close to 0.6%. Figure 31 shows the distribution of fatalities (in people) for the barrios, sorted by number of fatalities. The dashed horizontal line in the figure depicts the average value for all barrios. Note that for the most vulnerable zones, the fatalities exceed the average value by a large margin, whereas for nearly 60 barrios, the expected fatalities are five or fewer. This data can be used to allocate emergency and medical resources to the most vulnerable areas.

6.2.3 Expected building damage (by colored tag categories) After an earthquake, a damage survey of the affected area would be performed. The distribution of assigned tag colors (green, yellow, and red, for safe, limited, and prohibited occupancy, respectively) is evaluated here based on the expected building damage from a design-level earthquake. Figure 32 presents the expected distribution of assessment tag colors for each barrio. Note that the distribution of green-, yellow-, and red-tagged building ratios are uniform for most of barrios. However, for a few barrios, damage that is more extensive is anticipated.

As shown in Figure 33, for approximately 180 barrios, the anticipated red-tagged buildings would be close to 25% of the building population, and red- and yellow-tagged buildings would be close to 60%. However, for the remaining areas, much worse performance is expected. In the worst handful of barrios, the percentage of red-tagged buildings would be close to 70%, and the percentage of combined yellow- and red-tagged buildings would be approximately 80%. Therefore, extensive damage and large numbers of fatalities would be anticipated for these locations. This data can be used to dispatch the assessment team to the most adversely affected areas to determine whether buildings are safe to reoccupy and to reduce the need for temporary housing.

6.2.4 Expected building damage (damage ratio and debris volume) After a design-level earthquake, a damage survey of the affected area would be performed and estimates of the building damage extent and type would be developed. Additionally, provisions would be necessary for debris removal to allow the city to recover and return to normal operation.

Figure 34 presents the expected distribution of building damage for each barrio. Note that the distribution is uniform at approximately 40% for most of the barrios. However, for a few barrios, more severe structural damage is anticipated. Figure 35 shows the sorted distribution of debris volume (in m3) for the barrios,. The dashed horizontal line in the figure depicts the average value for all barrios. Note that for the most affected zones, the amount of debris is significantly larger than the average. These are likely the zones with a combination of more buildings and higher damage ratio. This data can be used to allocate construction equipment and personnel to the most vulnerable areas to assist in recovery.

SJ Phase 1 Report_2016-11-15 47 © 2016 Miyamoto International, Inc. Figure 31 shows the distribution of fatalities (in people) for the barrios, sorted by number of fatalities. The dashed horizontal line in the figure depicts the average value for all barrios. Note that for the most vulnerable zones, the fatalities exceed the average value by a large margin, whereas for nearly 60 barrios, the expected fatalities are five or fewer. This data can be used to allocate emergency and medical resources to the most vulnerable areas.

Figure 30. Distribution of fatality ratios for barrios

Figure 31. Sorted distribution of fatalities in barrios

90%

80%

70%

60%

50%

40% Tag distribution 30%

20%

10%

0% 301 102 704 103 706 805 905 907 409 909 506 711 702 809 508 205 914 714 614 916 511 920 615 923 719 417 926 721 807 305 1012 1101 1104 9902 1107 1008 1119 1013 99071 99073 Barrio No.

Figure 32. Distribution of building damage categories for barrios

Figure 33. Sorted distribution of red-tagged and red-tagged + yellow–tagged buildings

Figure 35. Sorted distribution of debris volume

SJ Phase 1 Report_2016-11-15 50 © 2016 Miyamoto International, Inc. 6.3 Risk distribution maps The graphical distribution of findings from probabilistic risk analyses are presented on the following pages. Figure 36 through Figure 38 present the spatial distribution of structural damage, fatalities, and red-tagged buildings and debris volume, respectively. As discussed previously, the values presented correspond to the mean (expected) values.

In the figures, the color distribution indicates the expected intensity of each consequence. The data from these maps can be used to identify the barrios that are most susceptible to earthquake losses, which can then be priori- tized for allocation of resources for seismic retrofit and earthquake preparedness. In particular:

• The distribution of fatalities in barrios differs significantly for daytime and nighttime earthquake scenarios. This difference is attributed to citizens’ commuting to work from their households during the day. As such, it is imperative that both scenarios be considered for risk planning. • Depending on the consequence parameter chosen, various barrios show increased vulnerability. Howev- er, certain barrios appear to be vulnerable for multiple risks. These barrios may need extra attention when planning risk mitigation and preparedness programs.

Ratio of highly damaged buildings

Building area Figure 36. Spatial distribution of structural damage

Daytime

Nighttime Figure 37. Spatial distribution of fatalities

Red-tagged buildings

Debris volume, m3 Figure 38. Spatial distribution of red-tagged buildings and debris volume

6.4.1 Overview Table 16 presents the exposure data for the studied area. The canton of San José is home to approximately 352,000 (nighttime) to 472,000 (daytime) occupants and has nearly 85,800 buildings. It is important to keep these numbers in mind when reviewing the aggregated data.

Table 16. Exposure data for the canton of San José Buildings Population No. Area, m2 Daytime Nighttime 85,800 26,900,000 472,000 352,000

6.4.2 Physical damage and building tags The anticipated physical damage to the built area that is subject to the design-level earthquake is listed in Table 17. Note that approximately 60% of the buildings would be yellow- or red-tagged. The damage area is nearly 42% of the total building area, and the earthquake could result in over 4,940,000 m3 of debris.

Table 17. Expected values of structural loss Damage Yellow-tagged Red-tagged Volume, m3 % Area, m2 % No. % No. 42% 11,350,000 33% 28,000 26% 22,500 4,940,000

6.4.3 Fatalities The anticipated fatalities from a design-level earthquake are listed in Table 18. The area could experience close to 3,000 fatalities, which is nearly 0.7% of the population of the canton.

Table 18. Expected values of fatalities No. of fatal- Fatality % Time of day ities Daytime 0.64% 3,000 Nighttime 0.76% 2,700

6.4.4 Discussion

The results of earthquake risk estimation for the canton of San José showed that significant structural damage and a moderately high fatality rate are to be expected in a design earthquake. The assessment also showed that some barrios are particularly vulnerable to adverse consequences from such an event. These findings point to the need to develop an earthquake preparedness program, including allocation of resources, retrofitting of vulnerable buildings, earthquake planning and preparedness, and development of a post-earthquake assessment and recovery program. Note that the results that are discussed in this report are based on probabilistic analysis that used both engineering assumptions and engineering judgment. Furthermore, results were obtained from a given design-level scenario earthquake. Finally, the presented results are the mean (average) values.

Because several uncertainties are considered in the stochastic risk analysis, the results essentially contain a certain level of variation that comes from probabilistic distribution. Therefore, such variations should be accounted for when interpreting the findings and applying the results for future planning.

SJ Phase 1 Report_2016-11-15 55 © 2016 Miyamoto International, Inc. 7. Conclusions Experience from past and recent earthquakes in Central and South America has shown that extensive damage affects the entire built environment, resulting in loss of life and causing physical damage that can be a significant portion of the country’s GDP. Within Central America, San José—the capital and the major economic center of Costa Rica, with a population of approximately 450,000—is the subject of this report.

The risk assessment algorithm used the following parameters as input: (1) design-level seismic hazard; (2) citywide exposure data, including structural properties and number of occupants; (3) building fragility for the common building types; and (4) consequence functions, relating the number of fatalities, structural damage, and debris volume to the building damage state.

By using available data, site class and bedrock acceleration data were combined to develop surface accelerations for the design earthquake. The seismic design parameters for peak ground accelerations (PGAs) were computed and were estimated at approximately 0.6g. Field surveys of 576 buildings (0.7% of the building stock), satellite image data, and development information were used to estimate the exposure data. The exposure data was used to divide buildings in various groups of similar construction type. Building fragility parameters that are suitable for San José construction were determined by using well-known worldwide resources. The parameters for consequence functions were based on global standards and were modified for the construction types that are found in San José. The risk exposure data is presented in Table 19.

Table 19. Building area and population exposure for the canton of San José No. of Built area, Occupants Occupants buildings m2 (daytime) (nighttime) 85,800 26,900,000 472,000 352,000

Probabilistic seismic risk analysis was then performed by using the OpenQuake engine to compute the associated risks. The data for individual buildings was then aggregated to obtain the expected value responses for the individual 196 barrios and was summed for the canton of San José. The key aggregated results from the seismic risk analysis are presented in Table 20.

The analysis results that Table 20 presents show that:

• The number of buildings that are expected to be yellow-tagged (moderately damaged) or red-tagged (severely damaged or collapsed) is estimated at about 51,000 structures, or approximately 60% of the building stock. • Depending on the time of event, approximately 3,000 fatalities (for an estimated rate of 0.7%) is anticipated.

• The generated debris volume of 4,940,000 m3 is significant and must be accounted for.

Table 20. Expected values of vulnerability for earthquake risk for the canton of San José Structural Fatalities Fatalities Yellow-tagged Red-tagged Debris vol- damage daytime nighttime buildings buildings ume % Area, m2 % No. % No. % No. % No. m3 42% 11,350,000 0.64% 3,000 0.76% 2,700 33% 28,000 26% 22,500 4,940,000

SJ Phase 1 Report_2016-11-15 56 © 2016 Miyamoto International, Inc. The high physical damage and fatality rates from an earthquake that are computed in this report are not unex- pected, and they point to the need for development of a risk mitigation program. As part of such a program, it is recommended that the following strategies be implemented:

• Provide a seismic strengthening program for key structures that are identified as having the most risk because of their inherent structural vulnerability, density of occupants, and importance. • Establish a damage assessment program for earthquake hazard. It is critical to train and certify engineers and to establish logistics. Such a program will improve response and recovery efforts after major earthquakes. • Optimize allocation of emergency response and recovery resources by identifying the most vulnerable regions. • Establish communication and public outreach programs. It is critical to communicate results and the abovementioned recommendations. Communities should be informed about earthquake risk and risk reduction methods.

The results, findings, and conclusions that are presented in this report are based on a seismic risk analysis derived from internationally recognized references and state-of-the-art analytical techniques. However, as with any engineering project, the underlying methods and analysis are based on certain assumptions and engineering judgment. Additionally, the findings are based on a given design-level scenario earthquake intensity and correspond to the expected values or means. That is, the results present the expected outcome for an average event from a sample of a large pool of events with similar intensities. As such, the findings in this report include a certain level of uncertainty (inherent in risk assessment) and should not be extrapolated directly to a future seismic event. Ac- cordingly, these assumptions and variations should be accounted for when interpreting the findings and applying the results for future planning.

Climent, Á., Rojas, W., Alvarado, G.E., and Benito, B. (2008). Proyecto Resis II Evaluación de la amenaza sísmica en Costa Rica. Colegio Federado de Ingenieros y de Arquitectos de Costa Rica (2010). Código Sísmico de Costa Rica. Instituto Tecnológico de Costa Rica, Costa Rica. CR (2010) Codigo Sismico de Costa Rica, Instituto Tecnológico de Costa Rica, Cartago, Costa Rica. Federal Emergency Management Agency (FEMA) (2001a). Hazus-MH 2.1, Multi-hazard Loss Estimation Methodology, Earthquake Model. Federal Emergency Management Agency, Washington, DC, USA. Federal Emergency Management Agency (FEMA) (2001b). Hazus-MH MR5, Advanced Engineering Building Module (AEBM), Technical and User’s Manual. Federal Emergency Management Agen- cy, Washington, DC, USA. Global Earthquake Model Foundation (GEM) (2016). The OpenQuake-engine User Manual, Global Earthquake Model (GEM) Technical Report 2016-03. Global Earthquake Model Foundation (GEM) (2014). User guide: Tool for spatial inventory data development, Global Earthquake Model (GEM) Technical Report 2014-05. Schmidt, V., Moya, A., Climent, Á., Rojas, W., and Boschini, I. (2005). Microzonificación sísmica de San José, Costa Rica. Universidad de Costa Rica. U.S. Geological Survey (USGS), Jaiswal, K., Wald, D.J., and Hearne, M. (2009). Estimating Casualties for Large Earthquakes Worldwide Using an Empirical Approach, Open-File Report 2009-1136. U.S. Geological Survey (USGS), Jaiswal, K., and Wald, D.J. (2011). Rapid Estimation of the Economic Consequences of Global Earthquakes, Open-File Report 2011-1116.

Download links here:

English miyamotointernational.com/prepare-phase-1-report-appendix-a-english

Spanish miyamotointernational.com/prepare-phase-1-report-appendix-a-espanol

APPENDIX B: Building exposure model

The report for the building exposure model, which was developed by ImageCat, is presented in the following section on the next page.

14th June 2016 Contents

Contents 1. Introduction ...... 1

2. Development Patterns ...... 3 3. Simplified Mapping Schemes ...... 9 4. Development Pattern GIS layer Metadata ...... 10 5. Barrio-level building counts by structure type ...... 12

Contact Information Charles Huyck ImageCa, Inc.. 400 Oceangate, Suite 1050 Long Beach, CA 90802 Telephone: (562) 628-1675 Email: [email protected]

C 1. Introduction This exposure database was developed through a statistical process based on “mapping schemes” and “Development Patterns.” Mapping schemes refer to allocating the ration of buildings by occupancy type, structure type, number of stories, era, or any number of “branches” defined by the user. A mapping scheme is then associated with development patterns within a GIS database through a user-defined class key to apply these the building ratios to the aggregate building counts by development pattern to yield an exposure estimate of the number and size of buildings for each zone. Given the detail of the digitized development patterns and the requirement to provide data at the barrio level, these counts have been aggregated to the barrio unit.

For San Jose, the city was divided into approximately 400 zones indicating 6 development patterns. These patterns were established by the land use interpretation expert and the project engineer after comparing the existing patterns against those used on previous projects. Typically one or two classes are added, but in the case of San Jose, delineation was straight forward. In establishing the development patterns, the project engineer weighs the benefits of increasing the number of classes of the development patterns (differentiation) against the need for a statistically robust sampling data set given the resources available.

The term “Development Pattern” is used to define areas which are sufficiently homogenous as to be characterized by a single mapping scheme- which is up to the discretion of the project engineer. Even in highly heterogeneous areas where assigning a single scheme is difficult, these areas will ultimately represent a class for building allocation. For example, a historic central business district that also contains new development may be identified by the user as a single zone with a special type, even though it is not particularly homogenous. Homogeneity is an ideal, but often not feasible. It was chosen over the more common term “land use zones” in order to emphasize the ability to delineate areas based on any criteria that may be important for the characterization of risk- including, but not exclusive to land use. It is ultimately a statistical tool.

Next, referring to the GEM guidelines outlined in the IDCT protocol: In-field Sampling Protocol a sampling strategy to collect building data. This was done by Georgiana Esquivias working closely with Diana Ubico. A significant and robust sampling dataset was collected by city staff using the IDCT tools. Data is collected from all the staff members and aggregated into a single file and checked for accuracy without any issues. In addition, GIS staff provided shapefiles of building footprints and the preferred unit of aggregation, barrio boundaries. The building footprints were used to estimate both the number and square footage of buildings.

This process resulted in the 3 key data required for developing an exposure data set: 1) the number and square meter area of buildings (provided by the footprint data); 2) the zones delineating development patterns; and 3) the building samples consistent with the GEM

1 taxonomy. These data are then loaded into SIDD. For each type of user zone defined in the development pattern dataset, SIDD creates a preliminary mapping scheme. The mapping scheme is simply a statistical summary of the percentage of sampled buildings in each category defined by the GEM Building Taxonomy (S. Brzev, A.W. Charleson, K. Jaiswal, C. Scawthorn, 2012). The SIDD tool is then used in an iterative fashion to examine the mapping schemes, adjust the mapping schemes, review and adjust average building size to characterize the development patterns based on knowledge of the city. For San Jose, there was a significant simplification of mapping schemes. When the project team settled on the mapping schemes, they are applied to the exposure. The resulting file exposure dataset allocates the footprints aggregated for each individual development pattern polygons into structural classes in the mapping scheme.

The process described above is typical for the development of an exposure database for loss estimation. SIDD is designed specifically to simplify this process and allow experts to share and expand the assumptions inherent in developing exposure data. It is important to recognize that the process of creating these files requires a high level user with experience in the complications of creating exposure data through a sampling strategy. For example, there is a trade-off between the size of the development pattern zones used and the number of “remainders,” or allocations of less than one, which result from the application of the mapping scheme. A smaller zone leads to a more accurate loss assessment from a spatial perspective, but also results in more remaindered structures. Another trade-off is between the complexity of building classification and allocation of structures. With a more detailed building classification comes the ability to characterize the vulnerability with finer precision (given damage functions). However, with more classification categories there are more remaindered structures and a potentially less robust assessment in aggregate. The process is not designed to alleviate or override these critical decisions inherent in the process of developing building exposure, but to streamline the process of making them.

C 2. Development Patterns

The following descriptions provide the basis for assigning development patterns to specific zones within San Jose. These zones are used to develop a sampling strategy and apply a weighted distribution of building stock to the Barrio level.

Development Pattern 0

This development pattern includes regions typically occupied by open spaces or cemeteries. The occasional support, maintenance or restroom buildings may be present

Development Pattern 1

This development pattern includes informal developments. Building footprints areas are very small, with little (to no) space between neighboring “buildings”. The number of stories rarely exceeds one. Construction is typically non-engineered and built of local materials.

Development Pattern 2

C This development pattern is found throughout the San Jose region and is dominated typically by large, 1-story warehouses with pitched roofs and regular in shape. This development pattern is predominately found along the outskirts of the city with the predominate structural systems being comprised of reinforced masonry, reinforced concrete and/or steel framed.

Development Pattern 3

This development pattern is found throughout San Jose, however is focused primarily along the southeast and western regions. The area is dominated by low-rise structures (1 to 2- story), with smaller footprints and typically tightly spaced. Some larger, mid-rise (4 story+) can be found within the development pattern. The predominant structural systems within the region are reinforced and confined masonry and reinforced concrete frames.

C Development Pattern 4

This development pattern is characterized by urban areas predominately occupied by low (with the occasional mid-rise) structures. Buildings within the development pattern are tightly spaced and will typically be found within or around the city center. Building footprints are typically larger than the other development patterns, with the most common structural systems being masonry (reinforced or confined) and reinforced concrete frames.

C Development Pattern 5

This development pattern is typical of a central business district of any major city. Building footprints are larger and story heights typically are within the 1 to 3 story range, however a number of mid and high-rise structures are present. Large offices, apartment complexes and government buildings will be found within this particular development pattern. The most common structural systems within the region are reinforced concrete frames and reinforced masonry structures.

C Development Pattern 6

This development pattern is typically comprised of large, low-rise warehouses, fairly regular in shape. Buildings are tightly spaced, however parking lots for the commercial stores are commonly found throughout. The long and narrow warehouses have pitched roofs and are typically either steel framed, reinforced masonry or reinforced concrete frames.

C 3. Simplified Mapping Schemes

A total of 119 lateral force resisting system/building height combinations were observed by the field survey team. Each combination has a unique material (e.g. concrete, masonry, steel), material technology (e.g. CMU, cast-in-place, precast, etc.), material property (e.g. cement mortar, bolted/welded connections, etc.), lateral force resisting system (e.g. moment frame, infilled frame, shear wall, etc.) and number of stories. To reduce the number of total structural attribute combinations, the Miyamoto team provided seven common building material and lateral force resisting systems found within the region, and two height bins to categorize each GEM taxonomy string could directly “map” to.

Included in the building material and LFRS are:

1. Wood/Informal/Non-engineered construction 2. Unreinforced masonry 3. Confined/Reinforced masonry 4. Reinforced Concrete Moment Frame 5. Reinforced Concrete Shear Wall 6. Steel Moment Frame 7. Steel Braced Frame

For the number of stories, each building was either identified as: 1. 1 to 3 Floors 2. 4 Floors or above

Using the GEM taxonomy string provided by the field surveys, the ImageCat team manually interpreted the 119 combinations and assigned each one the seven appropriate LFRS and appropriate story height bin.

Mapping from GEM Taxonomy to Simplified Category

GEM Taxonomy Simplified Category C99+PC/HBET:4,20 C99 Reinforced Concrete Moment Frame, 4F+ +PC/HBET:1,3 Reinforced Concrete Moment Frame, 1-3F

MR+CB99+MOC/LFLSINF/HBET:1,3 MR+CB99+MOC/LWAL/HBET:1,3 Confined/Reinforced Masonry, 1-3F S+SR/HBET:1,3 Steel Moment Frame, 1-3F S+SO/LFM/HBET:1,3 Steel Moment Frame, 1-3F CR+CIP/HBET:1,3 Reinforced Concrete Moment Frame, 1-3F CR+CIP/LFINF/HBET:1,3 Reinforced Concrete Moment Frame, 1-3F W+WO/LN/HBET:1,3 Wood/Informal/Non-engineered construction, 1-3F W+WLI/LFM/HBET:1,3 Wood/Informal/Non-engineered construction, 1-3F

C 4. Development Pattern GIS layer Metadata Using remotely sensed imagery the analyst at ImageCat, Inc. have created development pattern zones for the canton of San Jose, Costa. Per request of Miyamoto International the development patterns were segmented by the barrios data set provided by Municipality of San Jose. ImageCat utilized the parcel data set to assign development patterns and infilled the remaining area with the originally delineated development pattern zone clipped to the barrio extent.

The data is provided in a custom Transverse Mercator projection to match the data provided by the Municipality of San Jose. It is not “UTM”

Filename:

SJ_Development_Patterns_20160614.* Attribute fields:

ID: Unique identifier for each feature Barrio: Barrio name; obtained from San Jose Municipality Barrios layer CodeBarrio: Barrio numerical code; obtained from San Jose Municipality Barrios layer District: District numerical code; obtained from San Jose Municipality Barrios layer Latitude: Internal latitudinal location of the feature Longitude: Internal longitudinal location of the feature

C DevPatCode: Development pattern code

SPATIAL REFERENCE INFORMATION

Coordinate System: CRTM05 WKID: 5367 Authority: EPSG

Projection: Transverse_Mercator False_Easting: 500000.0 False_Northing: 0.0 Central_Meridian: -84.0 Scale_Factor: 0.9999 Latitude_Of_Origin: 0.0 Linear Unit: Meter (1.0) Geographic Coordinate System: GCS_CR05 Angular Unit: Degree (0.0174532925199433) Prime Meridian: Greenwich (0.0) Datum: D_Costa_Rica_2005 Spheroid: WGS_1984 Semimajor Axis: 6378137.0 Semiminor Axis: 6356752.314245179 Inverse Flattening: 298.257223563

C 5. Barrio-level building counts by structure type As described in the introduction, the surveyed data was cross referenced with the digitized development patterns to estimate the number of building by type. This data was aggregated to the barrio level.

Filename:

Barrio-level_building_count_20160614.csv

ASSETID: unique identifier for the table POLYGONID: Link to geographic ID in the provided barrio shaefile DISTRICTID: ID of district in canton DISTRICT: name of district in canton BARRIOCODE: code of barrio BARRIONAME: name of barrio CentroidX: Longitude of the weighted centroid of the barrio (internal to polygon) CentroidY: Latitude of the weighted centroid of the barrio (internal to polygon) BLDGTYPE: Building type of the Miyamoto taxonomy NUMBLDGS: Estimated number of buildings. Total for a given Barrio will match the number of building footprints provided by the city. TOTALAREA: Estimated area for the buildings in square meters. Total for a given Barrio will match the area of building footprints provided by the city- with an adjustment for number of stories as sampled for a given structure type.

The maps below present the results.

C 1

To: Kit Miyamoto CC: Ron Eguchi, Charlie Huyck From: Z. Hu Date: September 7, 2016

Re: Confidence Limits on Building Count Data – Costa Rica Study

Reviewing our process, we are highly confident in our estimate of number of buildings given that these estimates are constrained by building footprint data that was obtained by the ImageCat team during the course of the project. Thus, the main uncertainty arises in the quantification of the distribution of buildings by structural type which was generated through a limited sampling of ground survey data. In order to quantify the level of uncertainty surrounding structural type distributions, we have assumed a binomial distribution that technically is designed to represent the probability of observing a building of type t in N buildings, i.e., a binomial distribution evaluated at m successes in N trials.

To assess the confidence of our estimates, the three following methods are used.

1) Normal approximation method This is a conventional method of calculating the confidence interval given sample survey data; it is also referred to as a Wald interval. Table 1 contains confidence ranges for each building type t. It should be read as follows, there is a 95% confidence level that the actual ratio of building type t within the total population of buildings is within the specified range.

Table 1 : Confidence Interval (CI) by Building Type (Ratio) using Normal Approximation Method

1 95% confidence CI lower CI upper Confined/Reinforced Masonry, 1-3F 60.3% 68.4% Confined/Reinforced Masonry, 4F+ 0.0% 1.2% Reinforced Concrete Moment Frame, 1-3F 10.0% 15.7% Reinforced Concrete Moment Frame, 4F+ 4.5% 8.7% Steel Braced Frame, 1-3F 0.0% 0.9% Steel Moment Frame, 1-3F 2.4% 5.8% Unreinforced Masonry, 1-3F 1.2% 4.1% Wood/Informal/Non-engineered construction, 1-3F 6.2% 11.0%

2) Agresti-Coull interval This is an alternative confidence interval based on a similar principle. We used this method to verify that the result generated by the conventional implementation is valid. Table 2 contains the resulting ranges which are very similar to results given in Table 1.

Table 2: Confidence Interval by Building Type (Ratio) using Agresti-Coull Method

95% confidence CI lower CI upper Confined/Reinforced Masonry, 1-3F 60.3% 68.2% Confined/Reinforced Masonry, 4F+ 0.1% 1.6% Reinforced Concrete Moment Frame, 1-3F 10.3% 15.9% Reinforced Concrete Moment Frame, 4F+ 4.8% 9.0% Steel Braced Frame, 1-3F 0.0% 1.4% Steel Moment Frame, 1-3F 2.7% 6.1% Unreinforced Masonry, 1-3F 1.6% 4.4% Wood/Informal/Non-engineered construction, 1-3F 6.5% 11.2%

3) Standard error for each stratified development pattern Our exposure generation methodology combines basic stratified sampling methods with interpreted zones of consistent construction. Survey samples are aggregated by building type for each development pattern. The overall exposure numbers are generated by applying these distributions to the total number of buildings estimated in each development pattern- in the case of Costa Rica this was received directly from the client. Method 3 computes confidence intervals for each development pattern separately so that we can understand and explain the uncertainty in each pattern. Table 3 contains the confidence range for each building type ratio while Table 4 provides confidence ranges by total building counts.

Please note that for development pattern 0 - mostly open space - the building distribution from pattern 3 was used --- therefore, these two development patterns share the same confidence range. And for development pattern 1, we have assumed that all buildings are of the informal type, therefore no confidence limit is calculated.

2 Table 3: Confidence Interval by Building Count Ratio Development Pattern 2 3 4 5 6 CI CI CI CI CI CI CI CI CI CI 95% confidence lower upper lower upper lower upper lower upper lower upper Confined/Reinforced Masonry, 1-3F 49.2% 84.1% 74.8% 87.2% 64.5% 78.8% 30.2% 47.8% 46.0% 64.6% Confined/Reinforced Masonry, 4F+ 0.0% 2.1% 0.0% 4.5% Reinforced Concrete Moment Frame, 1-3F 0.0% 10.7% 2.1% 9.5% 5.6% 15.3% 13.7% 28.3% 12.7% 27.7% Reinforced Concrete Moment Frame, 4F+ 0.0% 17.1% 0.0% 3.4% 22.7% 39.3% 0.0% 4.8% Steel Braced Frame, 1-3F 0.0% 2.1% 0.0% 2.8% Steel Moment Frame, 1-3F 0.0% 2.1% 0.3% 5.7% 0.0% 4.5% 8.2% 21.5% Unreinforced Masonry, 1-3F 3.2% 11.4% 0.0% 3.4% Wood/Informal/Non- engineered construction, 1-3F 6.8% 37.6% 1.6% 8.6% 5.6% 15.3% 0.5% 7.5% 2.5% 12.4%

Development Pattern Definitions: 0: This development pattern includes regions typically occupied by open spaces or cemeteries. 1: This development pattern includes informal developments. Building footprints areas are very small, with little (to no) space between neighboring “buildings” 2: This development pattern is found throughout the San Jose region and is dominated typically by large, 1-story warehouses with pitched roofs and regular in shape. 3: This development pattern is found throughout San Jose, however is focused primarily along the southeast and western regions. 4: This development pattern is characterized by urban areas predominately occupied by low (with the occasional mid-rise) structures. 5: This development pattern is typical of a central business district of any major city. 6: This development pattern is typically comprised of large, low-rise warehouses, fairly regular in shape.

3 Table 4: Confidence Intervals by Building Count 0 1 2 3 4 5 6 Total No. of Buildings 837 5784 1537 56963 18100 1098 1452 CI CI CI CI CI CI CI CI CI CI CI CI Lower upper Lower upper Lower upper Lower upper Lower upper Lower upper Confined/Reinforced Masonry, 1-3F 626 730 756 1,293 42,625 49,680 11,666 14,268 332 525 668 938 Confined/Reinforced Masonry, 4F+ 1 384 1 50 Reinforced Concrete Moment Frame, 1-3F 18 80 1 164 1,217 5,436 1,008 2,774 150 311 185 402 Reinforced Concrete Moment Frame, 4F+ 1 263 1 620 249 432 1 70

Steel Braced Frame, 1-3F 1 384 1 31

Steel Moment Frame, 1-3F 1 17 1 1,182 49 1,032 1 50 120 313 Unreinforced Masonry, 1- 3F 27 95 1,818 6,498 1 620 Wood/Informal/Non- engineered construction, 1- 3F 14 72 5784 105 578 929 4,892 1,008 2,774 5 83 37 179

For additional information, see https://en.wikipedia.org/wiki/Binomial_proportion_confidence_interval and “Development Patterns” section in “Exposure Data Development” document.

4 Summary

For purposes of representing the uncertainty in building structural class assignments, we recommend that the results of Table 1 be used. To represent the range or variability of risk or loss results, one might also use the upper and lower bounds at the 95% confidence level as a means of representing the variability in overall loss estimates.

Tables 2 through 4 are provided as a means of validating the results of Table 1 and as a more detailed view of how the results vary across different development patterns.

-oOo-