**5. Case study of wildfire risk**

112 Novel Approaches and Their Applications in Risk Assessment

were assigned in a group workshop format as part of a broader wildfire risk assessment conducted for the Lewis and Clark National Forest in Montana, United States. Aspen stands are highly valued because they provide habitat for a broad diversity of wildlife, and due to their relative rarity across the landscape. Aspens are reliant on wildfire for natural regeneration, and so are modeled as experiencing substantial benefit from fire at low to moderate intensities, with minor loss expected high intensity fires. For high investment infrastructure (e.g., developed campgrounds, cabins, ranger stations), damages are expected from any interaction with fire, and are expected to increase in severity as fire intensity

Fig. 9. Response functions plotting relative NVC (y-axis) against fire intensity level (FIL; x-

Figure 10 displays additional response functions identified as part of the wildfire risk assessment for the Lewis and Clark National Forest. This figure highlights use of an additional geospatial variable, in this case moisture conditions on the site, to further refine fire effects estimates to old growth (OG) forest stands. Dry forests typically have evolved with and tend to receive a benefit from low to moderate intensity fires. At extreme intensities, high levels of mortality and damage are expected irrespective of site moisture

axis), for stands of aspen and high investment infrastructure.

increases.

conditions.

In this section we briefly review a recently published example of integrated, national-scale wildfire risk assessment (Thompson et al., 2011b). The effort leveraged tools, datasets, and expertise of the Fire Program Analysis (FPA) system, a common interagency strategic decision support tool for wildland fire planning and budgeting (http://www.fpa.nifc.gov). We aggregated results according to eight geographic areas organized largely for purposes of incident management and mobilization of suppression resources: California (CA), Eastern Area (EA), Great Basin (GB), Northern Rockies (NR), Northwest (NW), Rocky Mountain (RM), Southern Area (SA), and Southwest (SW).

Fig. 10. Response functions plotting relative benefit/loss (y-axis) against fire intensity level (FIL; x-axis), for the Old Growth (OG) HVRA, sorted by dry/wet site.

To map wildfire hazard we used wildfire simulation outputs (burn probability and intensity) from the FSim large fire simulator (Finney et al., 2011a), mapped at on a pixel basis (270m x 270m). In cooperation with the FPA Executive Oversight Group we identified seven key HVRA themes: residential structure locations, municipal watersheds, air quality, energy and critical infrastructure, federal recreation and recreation infrastructure, firesusceptible species, and fire-adapted ecosystems. Table 2 delineates the major HVRA themes along with identified sub-themes. We engaged ten fire and fuels program management officials from the Forest Service, National Park Service, Bureau of Land Management, Fish and Wildlife Service, and the Bureau of Indian Affairs to facilitate response function assignments.

The Science and Opportunity of Wildfire Risk Assessment 115

categories: Moderate, High, and Very High. HVRAs assigned to the Very High category related to human health and safety, specifically concerns regarding air quality, water quality, and communities at risk. We then aggregated TCE results into a single weighted risk metric (wTCE) by assuming that the ranking of value categories maintained a simple proportional relationship. With this framework a (1, 3, 9) weight vector means that HVRAs assigned in the Very High value category are 3 times as important as resources in the High value category, which in turn are 3 times as important as resources in the Moderate value category. Clearly decision-makers can experiment with alternative value category and weight vector assignments, but our purposes were primarily to illustrate joint application of

Table 3 summarizes TCE values by HVRA, value category, and geographic area. In the Moderate value category the Southern Area (SA) presents the greatest risk, largely to Class I areas and concerns about air quality. Across all geographic areas fire-adapted ecosystems expect to see a benefit from fire, which on balance tend to outweigh losses to other HVRAs, leading to positive values for NVC. Within the High value category fire-susceptible species were the largest contributors to risk. The Southern Area contained the largest overall area of risk to energy infrastructure, with relatively low loss expected elsewhere. Low density built structures similarly had relatively low TCE values, with higher losses associated with the Southern Area, California, and the Southwest. Within the Very High value category nonattainment areas were by far the largest contributors to risk, and especially in California. Overall California presents the largest risk in the Very High value category, followed by the Southern Area. Lastly the bottom row presents weighted TCE (wTCE) values using the (1, 3, 9) weight vector. Consistent with results from the Very High value category, California and the Southern Area appear most susceptible to wildfire-related losses. The Great Basin ranks third, due primarily to extensively mapped sage grouse habitat. Thompson et al. (2011b) present additional results including sensitivity analysis of assigned relative importance weights, and refinements regarding the temporal effects of air quality degradation and the

> **Geographic area CA EA GB NR NW RM SA SW**

**-530.11 -25.12 -118.68 -47.82 -68.12 -32.82 -163.41 -67.86** 

Moderate -0.58 0.62 3.18 1.53 5.34 0.63 -4.86 2.54 High -9.91 -1.72 -32.99 -12.70 -19.40 -7.44 -9.54 -5.57 Very High -55.53 -2.29 -2.54 -1.25 -1.70 -1.24 -14.44 -5.97

Table 3. Total change equivalent (TCE) in thousands of hectares for each Geographic area

In summary, the case study briefly explored here demonstrates application of quantitative wildfire risk assessment. The approach is scalable, in that the same integration of burn probability maps, geospatial identification of HVRAs, and resource response functions can

multi-criteria decision analysis and risk assessment.

spatial extent of mapped habitat.

**HVRA Value Category** 

> **wTCE Totals (1, 3, 9)**

and Value Category.

Response functions indicated percentage NVC according to fire intensity category, as measured by flame length. As a consistent measure of NVC across HVRAs we used an areabased proxy called Total Change Equivalent (TCE). TCE effectively measures the equivalent area lost (or gained) for a particular HVRA. Since mapped pixels can support multiple HVRA layers, generation of risk estimates entailed geospatial computations for each pixel-HVRA layer combination.


Table 2. HVRA layers used in national risk assessment Modified from (Thompson et al., 2011b).

Although calculating TCE in a common area-based measure does facilitate integration of multiple HVRAs and the evaluation of alternative mitigation strategies on the basis of costeffectiveness, TCE does not capture management priorities across HVRAs. To better integrate TCE calculations we turned to multi-criteria decision analysis techniques to assign each HVRA an importance weight. First, we adopted a categorical approach using input from the fire and fuels program management officials consulted for assistance with fire effects analysis. With guidance from the experts we assigned each HVR to one of three value

Response functions indicated percentage NVC according to fire intensity category, as measured by flame length. As a consistent measure of NVC across HVRAs we used an areabased proxy called Total Change Equivalent (TCE). TCE effectively measures the equivalent area lost (or gained) for a particular HVRA. Since mapped pixels can support multiple HVRA layers, generation of risk estimates entailed geospatial computations for each pixel-

**HVRA Category HVRA Layer HVRA Value Category** 

Moderate and high density built

Non-attainment areas for PM 2.5

Air quality Class I areas Moderate

Oil and gas pipelines Power plant locations Cellular tower locations

FS campgrounds FS ranger stations BLM recreation sites and

campgrounds

campgrounds

Fire-adapted regimes

NPS visitor services and

Designated critical habitat National sage-grouse key habitat

National scenic and historic trails National alpine ski area locations

Table 2. HVRA layers used in national risk assessment Modified from (Thompson et al.,

Although calculating TCE in a common area-based measure does facilitate integration of multiple HVRAs and the evaluation of alternative mitigation strategies on the basis of costeffectiveness, TCE does not capture management priorities across HVRAs. To better integrate TCE calculations we turned to multi-criteria decision analysis techniques to assign each HVRA an importance weight. First, we adopted a categorical approach using input from the fire and fuels program management officials consulted for assistance with fire effects analysis. With guidance from the experts we assigned each HVR to one of three value

FWS recreation assets

structures

and Ozone

Energy infrastructure Power transmission lines

Low density built structures High

6th order Hydrologic Unit Codes Very High

Very High

Very High

High

High

High

Moderate

HVRA layer combination.

Residential structure

location

Municipal watersheds

Recreation infrastructure

Fire-susceptible

Fire-adapted ecosystems

species

2011b).

categories: Moderate, High, and Very High. HVRAs assigned to the Very High category related to human health and safety, specifically concerns regarding air quality, water quality, and communities at risk. We then aggregated TCE results into a single weighted risk metric (wTCE) by assuming that the ranking of value categories maintained a simple proportional relationship. With this framework a (1, 3, 9) weight vector means that HVRAs assigned in the Very High value category are 3 times as important as resources in the High value category, which in turn are 3 times as important as resources in the Moderate value category. Clearly decision-makers can experiment with alternative value category and weight vector assignments, but our purposes were primarily to illustrate joint application of multi-criteria decision analysis and risk assessment.

Table 3 summarizes TCE values by HVRA, value category, and geographic area. In the Moderate value category the Southern Area (SA) presents the greatest risk, largely to Class I areas and concerns about air quality. Across all geographic areas fire-adapted ecosystems expect to see a benefit from fire, which on balance tend to outweigh losses to other HVRAs, leading to positive values for NVC. Within the High value category fire-susceptible species were the largest contributors to risk. The Southern Area contained the largest overall area of risk to energy infrastructure, with relatively low loss expected elsewhere. Low density built structures similarly had relatively low TCE values, with higher losses associated with the Southern Area, California, and the Southwest. Within the Very High value category nonattainment areas were by far the largest contributors to risk, and especially in California. Overall California presents the largest risk in the Very High value category, followed by the Southern Area. Lastly the bottom row presents weighted TCE (wTCE) values using the (1, 3, 9) weight vector. Consistent with results from the Very High value category, California and the Southern Area appear most susceptible to wildfire-related losses. The Great Basin ranks third, due primarily to extensively mapped sage grouse habitat. Thompson et al. (2011b) present additional results including sensitivity analysis of assigned relative importance weights, and refinements regarding the temporal effects of air quality degradation and the spatial extent of mapped habitat.


Table 3. Total change equivalent (TCE) in thousands of hectares for each Geographic area and Value Category.

In summary, the case study briefly explored here demonstrates application of quantitative wildfire risk assessment. The approach is scalable, in that the same integration of burn probability maps, geospatial identification of HVRAs, and resource response functions can

The Science and Opportunity of Wildfire Risk Assessment 117

as a function of contemporary management, land use patterns, vegetative succession and

The authors wish to recognize and thank Julie Gilbertson-Day, Mark Cochrane, Anne Birkholz, Jon Rieck, Joe Scott, and Don Helmbrecht for various contributions to figures and tables presented in this figure. The lead author is grateful for support of the Rocky

Agee, J.K. & Skinner, C.N. (2005). Basic principles of forest fuel reduction treatments. *Forest Ecology and Management,* Vol. 211, No. 1-2, pp. 83-96, ISSN 0378-1127 Ager, A.A., Finney, M.A., Kerns, B.K., & Maffei. H. (2007). Modeling wildfire risk to

*Forest Ecology and Management*, Vol. 246, No. 1, pp. 45-56, ISSN 0378-1127 Ager, A.A., Vaillant, N.M., & Finney, M.A. (2010). A comparison of landscape fuel treatment

Ager, A.A., Vaillant, N.M., & Finney, M.A. (2011). Integrating Fire Behavior Models and

Planning. *Journal of Combustion*. Doi: 10.1155/2011/572452, ISSN 2090-1976 Ager, A.A., Finney, M.A., Reger, A., & Buonopane, M. (Submitted) Wildfire exposure to

Andrews, P., Finney, M., & Fischetti, M. (2007). Predicting Wildfires. Scientific American

Brillinger, D.R., Autrey, B.S., & Cattaneo, M.D. (2009). Probabilistic risk modeling at the

Bruins, R.J.F., Munns, Jr., W.R., Botti, S.J., Brink, S., Cleland, D., Kapustka, L., Lee, D.,

Calkin, D., Ager, A.A., & Thompson, M.P. (2011). A comparative risk assessment

Agriculture, Forest Service, Rocky Mountain Research Station. 63 p.. Carmel, Y., Paz, S., Jahashan, F., & Shoshany, M. (2009). Assessing fire risk using Monte

northern spotted owl (*Strix occidentalis caurina*) habitat in Central Oregon, USA.

strategies to mitigate wildland fire risk in the urban interface and preserve old forest structure. *Forest Ecology and Management,* Vol. 259, No. 8, pp. 1556-1570, ISSN

Geospatial Analysis for Wildland Fire risk Assessment and Fuel Management

analysis on the national forests in the Pacific Northwest, USA. *Risk Analysis*, ISSN

wildland urban interface: the 2003 Cedar Fire. *Environmetrics* Vol. 20, pp. 607-620,

Luzadis, V., McCarthy, L.F., Rana, N., Rideout, D.B., Rollins, M., Woodbury, P., & Zupko, M. (2010). A New Process for Organizing Assessments of Social, Economic, and Environmental Outcomes: Case Study of Wildland Fire Management in the USA. *Integrated Environmental Assessment and Management* Vol. 6, No. 3, pp. 469-483,

framework for wildland fire management: the 2010 cohesive strategy science report. Gen. Tech. Rep. RMRS-GTR-262. Fort Collins, CO: U.S. Department of

Carlo simulations of fire spread. *Forest Ecology and Management* Vol. 257, No. 1, pp.

Mountain Research Station and the National Fire Decision Support Center.

disturbance, and, importantly, climate change.

**7. Acknowledgement** 

0378-1127

0272-4332

ISSN 1180-4009

ISSN 0730-7268

370-377, ISSN 0378-1127

Vol. August, pp. 47-55, ISSN 0036-8733

**8. References** 

be applied at project-level to regional to national analyses. A number of improvements can and are being pursued, such as refining the fire simulation outputs, identifying a larger and more representative set of HVRAs, introducing more structure and engaging more experts to define response functions, and using more complex multi-criteria decision analysis methods to articulate relative importance across HVRAs.
