A fire severity mapping system for real time fire management application and long-term planning.

Most recent efforts to map the severity of wildland fire use satellite imagery data to create maps with three to five ordinal categories of severity (Kasischke et al. 2007).

The process involves evaluating differences in the spectral reflectance characteristics of landscape features between pre‐ and post‐fire conditions and relating that information to the severity of a fire. While this approach is useful for evaluating fire effects sometime after the fire (“burn severity”), imagery‐derived maps of “fire severity” (i.e., fire effects during or immediately following a fire event; Lentile et al. 2006) are rarely used for real‐time or short‐term wildfire operations. Smoke and cloud obfuscation, lack of sufficient image‐processing expertise, and image availability issues (new TM data are only available every 14 days) all make it nearly impossible to quickly obtain useable imagery to use in real‐time mapping.

An alternate method of mapping fire severity uses computer simulation to mechanistically model fire effects that are then used to predict fire severity (Keane et al. 2009[in press]). Computer models, such as FOFEM and CONSUME (Ottmar et al. 1993; Keane and Reinhardt 1994), have been available to fire management for over a decade to simulate fire effects. These models simulate the direct effects of a fire on the vegetation, fuels, and soils for a point in space and output these effects using biophysically‐based variables such as fuel consumption and tree mortality. By modeling direct fire effects with computer simulations, fire severity assessments can be tailored for specific applications and maps can be produced anytime during a wildfire to provide instant assessments for real‐time fire management.

Additionally, powerful new statistical models combined with satellite‐derived historical burn severity data from the Monitoring Trends in Burn Severity (MTBS, http://mtbs.gov) program provide an opportunity to infer potential severity for future fires from an understanding of recent fires. Recent studies (Holden et al. 2009, Holden et al. [in prep]) have shown that MTBS burn severity data coupled with topographic and climatic predictor variables and the Random Forest machine learning algorithm (Brieman 2001) can predict the occurrence of high severity fire with over 70% accuracy. Using similar methods, it is possible to create a map of the potential for high severity fire across broad geographic scales, such as the western United States, that could be available as an on‐the‐shelf resource to fire managers.

We are creating a Fire Severity Mapping System (FSMS) for the western United States that will deliver fire severity map products to fire management applications at relevant time and spatial scales. By integrating LANDFIRE data layers, fire effects models, and new techniques for analyzing satellite‐derived burn severity data into one comprehensive computer modeling package, we hope to make it easier for managers to run fire hazard and fire severity maps at real‐time or short‐term time frames and over a range of spatial scales. This FSMS will be composed of a suite of digital maps, simulation models, and analysis tools that can be used to create fire severity maps for: 1) real‐time forecasts and assessments in wildfire situations, 2) wildfire rehabilitation efforts, and 3) long‐term planning. This FSMS would NOT replace the suite of fire severity products currently used by fire management (e.g., BARC and BAER severity maps); rather, it would complement them to provide a more comprehensive suite of fire severity mapping products. The blend of many fire severity mapping approaches that are incorporated into this system should help meet fire management demands for rapid but accurate assessment of spatial fire severity given their time, funding, and resource constraints.

For additional details about the satellite-derived mapping efforts for FIRESEV please see the Study plan for developing burn severity potential map.

For additional details about simulating fire effects and burn severity please see the Study plan for developing fire severity keys from simulation modeling.

PRINCIPAL INVESTIGATOR
Robert E. Keane, Research Ecologist and Deputy Program Manager for the Fire, Fuel, and Smoke (FFS) Science Program; Director of the Fire Modeling Institute; Supervisory Research Ecologist

Staff
Greg Dillon, Ecologist; Pamela Sikkink, Research Ecologist
Penelope Morgan, University of Idaho; Jason Herynk, Systems for Environmental Management

GOALS AND OBJECTIVES

1) To create a map of the Landscape Potential for Severe Fire (LPSF map) that predicts potential for high severity fire in the western United States based on site conditions and over 20 years of satellite imagery-derived burn severity data.

2) To create a Simulated Potential Fire Severity map (SPFS map), using LANDFIRE data layers and the FOFEM model to use in real-time wildfire or wildland-use applications.

3) To integrate the two products above with several existing fire analysis tools so that managers can chose between different fire severity classifications to evaluate either (1) the fire severity potential on a landscape or (2) estimate actual fire severity from immediately after to a month after a fire event.

PROJECT STATUS

Initiated spring, 2009

FUNDING ORGANIZATIONS

Joint Fire Sciences Program Project # 09-1-07-4
NIFFT, Fire & Aviation Management Washington Office

REFERENCES CITED

Breiman, L (2001). Random forests. Machine Learning. 45: 5-32.

Holden ZA, Crimmins M, Luce C, Heyerdahl E, Morgan P, and Wood AW. (In prep.) Climate-Burn Severity Relationships in the Pacific Northwestern US (1984-2005).

Holden ZA, Morgan P, Evans JS. (2009). A predictive model of burn severity based on 20-year satellite-inferred burn severity data in a large southwestern US wilderness area. Forest Ecology and Management 258: 2399 - 2406.

Kasischke ES, Hoy EE, French NHF, and Turetsky MR (2007). Post-fire evaluation of the effects of fire on the environment using remotely-sensed data. In Gitas IZ and Carmona-Moreno C (Eds). Proceedings of the 6th International Workshop of The EARSeL Special Interest Group On Forest Fires, September 27-29, 2007 (Thessaloniki, Greece).

Keane RE, Reinhardt E (1994). FOFEM -- A First Order Fire Effects Model for predicting the immediate consequences of wildland fire in the United States. In Proceedings of the 12th Conference of Fire and Forest Meteorology, October 26-28, 1993 (Jekyll Island, GA).

Keane RE, Karau, EC, Drury, S, Hessburg P, and Reynolds, K. (2009 [in press]) Evaluating wildfire benefits by integrating fire and ecosystem simulation models in a spatial domain. Ecological Modelling.

Lentile LB, Holden ZA, Smith AMS, Falkowski MJ, Hudak AT, Morgan P, Lewis SA, Gessler PE, Benson NC (2006). Remote sensing techniques to assess active fire characteristics and post-fire effects. Int. J.Wildland Fire. 15: 319-345.

Ottmar RD, Burns MF, Hall JN, Hanson AD (1993) ‘CONSUME users guide’. USDA Forest Service General Technical Report Pacific Northwest Research Station GTR-PNW-304. (Portland, OR)