Predicting Burn Severity Fire Effects in the Sierra Nevada: Relative Importance of Climate, Ecology, and Topography

Recent and projected activity, frequency, and intensity of fires in the Western United States nec¬es¬sitate that land managers understand the biophysical factors that strongly affect landscape fire pat¬terns. This is a management priority for optimizing fuel treatment locations as well as regions at elevated risk. I compare the influence of variations in local environments on burn severity pat¬terns for several medium-to-large fires during the 2018 fire season, studying the Sierra Nevada area in California, USA. While plume-dominated and day-of-burn weather are important aspects of fire severity, my study attempts to predict burn severity using predictive variables related to top¬ography, ecology, and climate. Random forest (RF) models predicted just under 75.70% of the Sierra Nevada fires’ burn variance using several variables: ‘time since last fire’, fire be¬havior, surface moisture availability (SMA), slope, aspect, land use, and snow water equivalent (SWE). Models using the untrained set (Whaleback fire) predicted burn variance with a 40.41%% accur¬¬acy. While reducing accuracy with the validation dataset, user accuracy for high severity fires was just above 60%. I found accounting for spatial autocorrelation in the model portrayed correlation be¬tween biophysical predictors. My results suggest that SMA, fire beha¬vior, and slope can help predict likely burn severity patterns under many burn conditions. This can provide land managers with crucial insight as to where fuel treatment priority should be al¬lo¬cated. Additionally, it will be important to address issues regarding conservation and fire man¬age¬ment using a broader scope. I propose the creation of the Greater Sierra Nevada Ecosystem (GSNE) as well as a Coordinating Committee to shift away from a piecemeal approach at conservation.