The Technical Efficiency of Wildfire Suppression in Alberta, Canada: A Stochastic Frontier Analysis

Abstract

Wildfire management agencies are increasingly interested in the efficiency of wildfire suppression as they work to protect human lives and communities from wildfire damages under constrained management budgets. In Alberta, climate change is expected to increase the length of the wildfire season and increase annual area burned over the coming decades. These anticipated changes in the wildfire season reinforce the need for efficient use of suppression resources. This study uses stochastic frontier analysis to determine the efficacy of suppression resources and quantify the technical efficiency of wildfire suppression in Alberta’s boreal forest zone. At the individual fire scale, we define the output of suppression resources, and therefore dependent variable of the suppression production function, as a held wildfire perimeter. Geospatial wildfire progression perimeter data from 34 wildfires were used to calculate the length of held perimeter for each observed day across all 34 wildfires. Suppression resources are included as production function inputs working to generate held perimeter while weather and fuel variables were included as factors effecting the level of suppression efficiency. Values at risk were included in the stochastic frontier models by using a satellite imagery machine learning dataset to identify inhabited structures within 30 km of active wildfire perimeters. Model results suggest ground equipment is the most effective suppression resource and positively contributes to average daily held perimeter. Average technical efficiency across all 34 wildfires was 26.89% and median technical efficiency was 23.61%. Alberta has no previous wildfire suppression efficiency research to establish a frame of reference for comparison of the efficiency estimates. The average suppression efficiency of 26.89% is lower than what was estimated in a similar study from the western United States which likely reflects Alberta’s boreal forest’s wildfire regime and coniferous fuels that are conducive to high intensity crown fires that are often difficult to contain. High drought codes and high percentages of coniferous fuels decrease the technical efficiency of containment. Technical efficiency of suppression was qualitatively higher for wildfires with nearby values at risk compared to wildfires with no nearby values at risk. Without considering values at risk, large wildfires had a low median technical efficiency of suppression that was similar to the smallest sample wildfires. After weighting the output to include values at risk, the technical efficiency of suppression for the largest wildfires was higher than the smallest wildfires in the sample. Results were robust across multiple model specifications and suggest there are opportunities to increase the efficiency of wildfire suppression in Alberta.