How DNR built the state hazard and risk maps: FSIM, LANDFIRE and county-level scaling
July 16, 2026 | Board Council Commission Agencies , Executive, Washington
This article was created by AI summarizing key points discussed. AI makes mistakes, so for full details and context, please refer to the video of the full meeting. Please report any errors so we can fix them. Report an error »
Washington DNR technical staff laid out the modeling and calibration steps behind the draft hazard and risk maps.
Matthew Deer, the lead wildland fire meteorologist for Washington DNR, said the team used FSim (a large-fire simulator originally developed by the U.S. Forest Service) and real weather and fire-history data to simulate thousands of fire seasons. "We run the fire simulation for 10,000 fire seasons," Deer said. From those simulations the team extracts annual burn probability and conditional fire intensity distributions for each pixel on the landscape.
Deer explained how the outputs translate to hazard categories: modeled annual burn probability values are classed so that 1% annual probability corresponds to the "high" hazard threshold and 2% corresponds to "very high." The team also computes conditional fire intensity (flame-length distributions) and applies a structure-focused response function that maps flame length to expected percent loss to produce an annual expected net value-change metric for each pixel.
On fuels and base data, the DNR team uses LANDFIRE as a starting dataset and then adjusts fuel models where LANDFIRE misclassifies human-modified or agricultural areas (for example, irrigated pasture or orchards that do not burn like wildland fuels). Presenter remarks noted there are roughly seven primary fuel categories and about 53 distinct fuel models underlying the analysis; local verification and fuel edits were used around areas such as Ellensburg, the Palouse and Wenatchee.
To make the product useful at the local level, Deer said the team currently scales risk relative to each county (pixel-level risk normalized within county) so planners in counties with generally low statewide hazard can still identify comparatively higher-risk areas locally; an option to view state-relative risk is under consideration. "The risk map is not tied to regulation," Deer said, adding the team was asked to build a useful planning product that local governments can adopt as needed.
DNR staff emphasized calibration to observed data: model runs are iteratively adjusted so that simulated large-fire counts and typical burned acres align with the last 20–30 years of observations for each fire-occurrence area. The team invited technical feedback on county-scaling choices and other methodological decisions during the public comment window.
Next technical steps include completing State Fire Marshal consultation on valuable assets and publishing technical documents and downloadable data alongside the final mapping tool.
Matthew Deer, the lead wildland fire meteorologist for Washington DNR, said the team used FSim (a large-fire simulator originally developed by the U.S. Forest Service) and real weather and fire-history data to simulate thousands of fire seasons. "We run the fire simulation for 10,000 fire seasons," Deer said. From those simulations the team extracts annual burn probability and conditional fire intensity distributions for each pixel on the landscape.
Deer explained how the outputs translate to hazard categories: modeled annual burn probability values are classed so that 1% annual probability corresponds to the "high" hazard threshold and 2% corresponds to "very high." The team also computes conditional fire intensity (flame-length distributions) and applies a structure-focused response function that maps flame length to expected percent loss to produce an annual expected net value-change metric for each pixel.
On fuels and base data, the DNR team uses LANDFIRE as a starting dataset and then adjusts fuel models where LANDFIRE misclassifies human-modified or agricultural areas (for example, irrigated pasture or orchards that do not burn like wildland fuels). Presenter remarks noted there are roughly seven primary fuel categories and about 53 distinct fuel models underlying the analysis; local verification and fuel edits were used around areas such as Ellensburg, the Palouse and Wenatchee.
To make the product useful at the local level, Deer said the team currently scales risk relative to each county (pixel-level risk normalized within county) so planners in counties with generally low statewide hazard can still identify comparatively higher-risk areas locally; an option to view state-relative risk is under consideration. "The risk map is not tied to regulation," Deer said, adding the team was asked to build a useful planning product that local governments can adopt as needed.
DNR staff emphasized calibration to observed data: model runs are iteratively adjusted so that simulated large-fire counts and typical burned acres align with the last 20–30 years of observations for each fire-occurrence area. The team invited technical feedback on county-scaling choices and other methodological decisions during the public comment window.
Next technical steps include completing State Fire Marshal consultation on valuable assets and publishing technical documents and downloadable data alongside the final mapping tool.
Don't Miss a Word: See the Full Meeting!
Go beyond summaries. Unlock every video, transcript, and key insight with a Founder Membership.
✓
Get instant access to full meeting videos
✓
Search and clip any phrase from complete transcripts
✓
Receive AI-powered summaries & custom alerts
✓
Enjoy lifetime, unrestricted access to government data
30-day money-back guarantee
