Ned Nikolov, Phillip Bothwell, John S. Snook
Year Published:

Cataloging Information

Fire Behavior
Simulation Modeling

Record updated: June 29, 2022
NRFSN number: 24703

The National Predictive Services (NPS) asked the USFS Rocky Mountain Center for Fire-Weather Intelligence (RMC) as a part of the Fire, Fuel, and Smoke Science Program (FFS) at the USFS Rocky Mountain Research Station (RMRS) to assist with the development of a system of statistical models for predicting the ignition probability (chance of start) & growth potential of wildfires on a national grid using NWS meteorological forecast data as input. The development of this gridded system of predictive equations was envisioned to proceed in 3 stages (phases).

Phase 1 (Oct. 2017 – Dec. 2018) developed and partially validated a high-resolution spatial model for forecasting the probabilities of cloud-to-ground lightning flashes as a function of 3-D fields of atmospheric parameters stretching from the surface to the tropopause. Lightning is known to account on average for 60% of the wildland-fire area burned annually in the USA. Hence a lightning-forecast model is a prerequisite to developing a robust fire-ignition prediction application.

Phase 2 (2019-2021) completed the verification of the lightning forecast model using independent meteorological data from NARR and GFS for 2017 and 2018. Also developed were statistical (logistic) equations for predicting the probabilities of ignition of lightning-caused, human-caused and any-cause wildfires on a uniform grid across the lower 48 states. The ignition-probability forecast model was validated against independent fire-occurrence data from 2017 and 2018 as well. This addressed the need of Predictive Services and the fire-management community for forecasting the chance of wildfire start at high spatial resolution across Conterminous USA (ConUS) going out to 7- 10 days based on long-term climatology and current forecast weather conditions.

The output from Phase 2 of this project is a set of monthly logistic equations capable of predicting daily probabilities of one or more wildfire ignitions due to either lightning, human factors, or any- cause on a national 20-km grid up to 10 days in advance using NWS numerical weather forecasts as input. Real-time operational wildfire-ignition forecasts are now available at the RMC Website (click on “7-Day Wildfire Start Forecast” under the Forecast Applications Panel).


Nikolov N, Bothwell P, and Snook J. 2022. Developing a Gridded Model for Probabilistic Forecasting of Wildland-Fire Ignitions Across the Lower 48 States: USFS-CSU Joint Venture Agreement Phase 2 (2019-2021), Final Report, 33p.

Access this Document