Modeling Forest Fire Spread in Southwestern Canada Using MODIS Remote Sensing and Integrated Data
Date:
Abstract: This research highlights the importance of accurately predicting forest fire spread to enhance fire management, proactive planning, and the judicious use of resources. The study is centered on wildfires in Alberta’s Fort McMurray and Slave Lake areas. It utilized a modified fire propagation model that incorporated MODIS data, including land surface temperature, land cover/use, and climate information. Pixels were categorized as burned or unburned based on the 2011 Slave Lake fire and the first 16 days of the 2016 Fort McMurray fire, using specific starting points and datasets. The 2011 Slave Lake fire simulation was accurate, with weighted average precision, recall, and F1-scores of 0.989, 0.986, and 0.987 respectively. Additionally, macro-averaged scores were 0.735 for precision, 0.829 for recall, and 0.774 for F1-score. For the 2016 Fort McMurray fire, the simulation employed a phased analysis, dividing the initial 16 days into three distinct periods. This method resulted in average precision, recall, and F1 scores of 0.958, 0.933, and 0.942. Macro-averaged scores for these periods were 0.681 for precision, 0.772 for recall, and 0.710 for F1-score. Segmenting the simulations into phases may improve the model’s adaptability to dynamic factors such as changing weather conditions and varying firefighting strategies. The study’s approach could significantly bolster wildfire management practices.
Conference page: https://wildlandfirecanada.com/