An important component of wildland fire behavior is rate of spread (ROS). The objectives of this research were to evaluate landscape-scale fuel and terrain controls on ROS estimates derived from airborne thermal infrared imagery (ATIR). Repetitive ATIR image sequences were collected during the 2017 Detwiler and Thomas wildfire events in California and detailed wildfire progression maps and ROS estimates were generated. Landscape sampling units (LSUs) were created to extract environmental covariates derived from pre-fire National Agriculture Imagery Program (NAIP) orthoimagery and USGS digital elevation models (DEMs). Statistical relationships between fire spread rates and landscape covariates were analyzed using (1) bivariate linear regression, (2) multiple linear regression, (3) geographically weighted regression (GWR), (4) eigenvector spatial filtering (ESF) regression, (5) regression trees (RT), and random forest (RF) regression. Directional slope is found to be the most statistically significant covariate with ROS for the five fire imagery sequences that were analyzed, and its relationship with fire spread rate is best characterized as an exponential growth function. Imaged-derived fuel covariates alone are statistically weak predictors of ROS, but when included in multivariate and machine learning models increased spread rate predictability and variance explanation compared to models with directional slope alone.
