Analyzing fire suppression crew production, efficiency, and composition on large wildland fires Kate Marcille Forest Engineering, Resources & Management, OSU Western Forest Economists Meeting May 19, 2014 Outline
Background ◦ Wildfire activity ◦ Fire suppression and management ◦ Handcrews Data ◦ Painting the management picture Approach ◦ Objective 1: Crew productivity ◦ Objective 2: Daily fireline factors ◦ Objective 3: Do we dig it? Current Status ◦ Specification search Moving Forward ◦ Substitutes? ◦ Improving modeling approaches Wildfire activity
. Increase in fire size and duration Acres burned by U.S. wildfires: 1961-2012 . Higher severity and intensity across the landscape . More interaction with Wildland Urban Interface (WUI) 10-year moving . Heightened societal awareness of average values at risk
Source: National Interagency Fire Center Fire Suppression: Costs and losses
Increasing cost of fire suppression Forest Service Budget Allocation Wildland Fire ◦ Over one billion in emergency suppression $10 Management expenditures $9
$8
◦ Huge portion of USDA Forest Service budget
$7
Greater loss $6 ◦ Risk to ecological and other resource values $5 ◦ WUI expansion and conflicts $4
Billions (2012 dollars) (2012 Billions $3
The financial impact of wildfire $2 management challenges the ability of the $1 $0 Forest Service to meet societal demands 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 and achieve forest health objectives. Source: USDA Forest Service Federal fire management decisions GENERAL WILDLAND FIRE DECISION-MAKING CHALLENGES DECISION-MAKING CHALLENGES . Uncertainty . Lack of incentives . Risk attitudes and preference . Open checkbook . Safety and exposure . Liability . Multiple objectives . Socio-political pressures
All these factors complicate the process of making cost- effective management decisions.
On-the-ground operations Improving decision-making space ◦ Better information ◦ Detailed and specific data ◦ Reassess current approaches and understanding
Improving fire management ◦ Where are resources being deployed? ◦ When are they being used? ◦ What types of resources are utilized each day?
Handcrews: Boots on the ground
Career and temporary wildland firefighters Agency and privately contracted Multi-level types ◦ Type I Interagency Hotshot Crews (IHCs) ◦ Type II ◦ Type II Initial Attack (IA) Assigned an array of tasks and responsibilities by division ◦ Direct, indirect, mop-up, rehab, point protection What are crews accomplishing on the ground?
http://www.fs.usda.gov/detail/alabama/news-events/?cid=STELPRDB5433456 Estimating resource productivity
San Dimas Production Rates, Broyles (2011) ◦ Estimation of productive capacity “…collection of more precise operational Holmes and Calkin (2013) data could help reduce uncertainty Modeled wildfire suppression inputs, xit, as they relate to specified output, DFLit regardingβ the relative importance of DFLit = γxit
factorsDFL thatit: daily contribute fireline constructed to productivity on fire i and day t shortfalls.”xit: resources used (handcrews, dozers, engines, helicopters) γ: constant β: vector of Holmesparameters and Calkin (2013)
Found that handcrew productivity was about 20% of the standard rates, on average
Saddle Complex Scott Mountain
Fontenelle
Schultz
Tecolote
Cost Fire Year State Acres (million $) Tecolote 2010 New Mexico 812 5.5 Schultz 2010 Arizona 15,075 9.5 Scott Mountain 2010 Oregon 3,464 4.6 Saddle Complex 2011 Montana 15,866 4.5 Fontenelle 2012 Wyoming 65,220 12.65 Data crosswalk
Rocky Mountain Research Station operational data Resource Ordering Status System (ROSS) ◦ daily fireline constructed ◦ mobilization and demobilization dates ◦ line by resource type ◦ quantity of each resource assigned ◦ assignment categorization ◦ crew level (Type I, II, IA) ◦ Direct, indirect, mop-up, rehab, point protection ◦ agency or contract crew (link to IAPs)
Daily Incident Action Plans (IAPs) Incident Command System daily reports ◦ crew name (ICS-209) ◦ division crew composition ◦ Incident Management Team type ◦ daily division assignment ◦ fire size in acres to date ◦ daily weather data ◦ estimated costs to date ◦ daily fire behavior information ◦ daily percentage of fire containment (DPC)
Objective 1: Productivity Mimic Holmes and Calkin (2013) production model ◦ handcrews only ◦ observed daily fireline
DFLit = f(handcrews, cumulative ERC, max windspeed, IMT dummy)
HDiFLit= f(handcrews, cumulative ERC, max windspeed, IMT dummy)
www.interfire.org Objective 2: Daily fire line factors
Crew composition ◦ contract vs. agency Daily fire assignments ◦ predominant daily assignment: direct, indirect, mop-up, rehab
DFLit = f(# of agency crews, total crews, mission type, ERC, maxwind, maxtemp, minhumid, IMT)
HDiFLit = f(# of agency crews, total crews, mission type, ERC, maxwind, maxtemp, minhumid, IMT) a) How productive are different resource mixes at constructing line? b) How does the mix of mission types impact the daily fireline constructed?
www.pacforestsupply.com Scott Mtn Mission Types Mission Type for All Fires 1% Objective 3: Do we dig it? 12% 3% 2% Direct 1% 1% Indirect 15% Assumption of line building as predominant activity Hold ◦ assignments vary across fires 11% 22% 72% ◦ many crews are not digging line Mop Up 3% n = 83 Other
Unknown Tecolote Mission Types DPCit= f(# of agency crews, total crews, mission type, ERC, maxwind, maxtemp, minhumid,IMT) 14% IA 2% 4% 7% Rehab 12% What are crews being asked to do? 10% 33% Contingency ◦ contribution to overall fire containment 13% Point Protection ◦ composition of labor by assignment type 43% n = 513 19% n = 95 Tecolote Fire Assignments by Crew Type 12
10 Current Status 8 6
Specification search 4
◦ econometric analysis 2
◦ regression exercise 0 ◦ Which model specifications are interesting?Day 1 Day 2 Day 3 Day 4 Day 5 Day 6 Day 7 Day 8 ◦ comparing variables across specifications Fontenelle Fire Assignments by Crew Type 10 agency 9 What relationships can be extracted? contract 8 Case study approach 7 6 ◦ Agency hotshot crew vs. Type II contract5 crew ◦ location, assignment, accomplishments4 3 2 1 0 Moving Forward Scale ◦ Is analyzing daily aggregate suppression resources enough? ◦ Resource placement and probability of success Spatially explicit data ◦ Better understanding of where resources are deployed ◦ Amount of fireline built that engages final fire perimeter ◦ Enhanced evaluation of firefighter exposure Substitutes? ◦ Can we demonstrate that agency and contract crews are not substitutes? ◦ Modeling mop-up activities vs. initial attack – how do we get better?
Improving wildfire management
.Reduce potential resource loss
.Decrease unnecessary exposure of wildland firefighters
.Reduce management costs
Acknowledgments Dr. Claire Montgomery, OSU College of Forestry, FERM Department, Corvallis, OR Dr. Dave Calkin, Rocky Mountain Research Station, Human Dimensions Program, Missoula, MT Dr. Matthew Thompson, Rocky Mountain Research Station, Human Dimensions Program, Missoula, MT Jon Rieck, Rocky Mountain Research Station, Human Dimensions Program, Missoula, MT
Rick Strachan Graduate Fellowship 2011-2012 Rocky Mountain Research Station Joint Venture Agreement 2013-2014
Thank you!
Mount Sentinel burning at night in July 2008 Photo Credit: Chad Harder
ECONOMIC PRODUCTION MODELS BASED ON SPATIALLY EXPLICIT DATA Improve understanding of the relative effectivenss of resources in producing fireline Amount of fireline that engages the final fire perimeters Other activities that suppression resources engage in other than fireline production Enhancing evaluation of exposure of firefighters to fireline dangers Optimal mix of agency and contract crews Are contract crews a fundamentally different type of firefighting resource? Donovan (2006) Periods of low demand (rather than high) determine the optimal number of agency crews Availability of alternative work is at least as important as fire season severity
Model is a framework for considering the tradeoffs between agency & contract crews (only one of potentially many tools) What about differences in productivity? ◦ Assumes equal productivity across crew types, across fires ◦ Need for combining productivity with uncertainty ◦ Anecdotally it seems that ICs don’t think of contract crews as particularly high efficiency units
Hot to get contractors to lower their bids- what could the FS do to get their costs down at this level? 160 Average acreage of U.S. wildfires: 1990-2012 140
120
100
80
60
40
20
0 1990 1993 1996 1999 2002 2005 2008 2011 US Forest Service Budget Breakdown
Supplemental/Emergency/Reserve Other Appropriations
Land Acquisition: LWCF State and Private Forestry
$10 Forest and Rangeland Research Capital Improvement and Maintenance
$9 Mandatory Appropriations National Forest System
$8 Wildland Fire Management
$7
$6
$5
$4 Billions (2012 dollars) (2012 Billions
$3
$2
$1
$0 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Fire days 72 Fire Assignments 513 Divisions (513)
Schultz Mission Types Saddle Complex Mission Types
Direct
Indirect
Hold
21% Mop Up 13% 4% 16% 33% Other
Unknown
14% IA
Rehab 28% 39% 10% 22% Contingency n = 106 Point Protectionn = 76 Agency Crew Fire Assignments 1% 1% 0% 0% Direct
11% Indirect 2% 22% Hold
Mop Up
12% Other Unknown
IA
9% Rehab
Contingency
42% Point Protection Contract Crew Fire Assignments 0% 1% 0% 2% Direct Indirect 8% 16% 5% Hold Mop Up Other 12% Unknown IA 7% Rehab Contingency 49% n = 86 Point Protection Schultz 2010 Teclolote 2010 Saddle Complex 2011