When wildfires start, firefighters and emergency personnel need detailed predictions about weather, fire spread, smoke concentration, heat release, fire-generated updrafts and downdrafts, and more to help them make critical decisions about how to deploy firefighting resources safely to protect lives and property.
Despite this urgent need, providing actionable information is a challenge. Standard weather forecasts, for example, are too coarse in resolution to capture the extremely localized winds that can drive a wildfire and they do not take into account how the fire itself can change the weather, creating abrupt shifts in the winds, pyrocumulus clouds, fire whirls, and other phenomena. Accurate predictions of fire behavior also require up-to-date inventories of area fuels, recent measurements of fuel and soil moisture, and other local-scale observations.
At NCAR, scientists are working to fill this critical information gap. Our scientists have built and are continuing to develop sophisticated wildfire behavior models that can scale a regional weather forecast with a grid spacing of 3 km down to a much finer grid spacing of about 100 meters, which is necessary to produce useful wildfire behavior forecasts. The modeling system also connects the wildfire behavior to the weather, allowing users to see how winds and other weather conditions affect the fire and, in turn, how the fire affects local weather conditions, thereby producing a much more realistic simulation of hyper-local conditions.
NCAR’s wildfire modeling system ingests observations of fire location from satellites and aircraft, data from land fuel models, and detailed information on the elevation and terrain. NCAR has also worked with end users, including the Colorado Division of Fire Prevention and Control, to deliver the results of the modeling in a format that is usable for emergency responders. The NCAR system includes such critical details as flame length, rate of spread, heat release, smoke concentration, turbulence intensity, down/updraft regions, and more.
NCAR scientists also use wildfire behavior models to simulate fires after the fact. This provides experts with a virtual laboratory to explore the relative importance of the many factors — such as fuel moisture, winds, or terrain — that contribute to a particular fire’s growth and movement.
Finally, NCAR scientists are partnering with university collaborators to develop a new computational platform to predict wildfire risks days to weeks before a blaze occurs. This information will help wildfire managers, emergency responders, and utility companies better anticipate where and when fires may ignite so they can plan and mobilize their resources in advance.