Burning Up: Modeling Wildfire Temperature Using AVIRIS Imagery

Burning Up: Modeling Wildfire Temperature Using AVIRIS Imagery

Patrick Sullivan, University of Utah; Dr. Dar Roberts, University of California Santa Barbara; Keegan Quigley, MIT Lincoln Laboratory

Measuring active fire temperature of wildfires is important for understanding burn severity, the role of fuels and predicting fire spread, but because of the dangers presented by wildfires, in-situ temperature data are difficult to collect. With the high-resolution data collected by the Airborne Visible Infrared Imaging Spectrometer, fire temperature can be modeled accurately from the radiance within pixels of a hyperspectral image containing fire. Using Planck's Law, the measured radiance can be fit to modeled blackbody curves to determine maximum temperature within a pixel by limiting the Root Mean Squared Error (RMSE) between measured and modeled radiance. Results showed our method to provide a more accurate model, because it retrieved fire temperature with minimal error for the 2017 Sherpa and Thomas Fires in California. The model was also shown to be portable across fires with dissimilar behavior providing the rate of temperature decay within a pixel was modeled carefully.