Modeling ignition probability and wildfire emissions in the boreal forest

Abstract

Wildfires are an important disturbance factor driving the biodiversity and carbon balance of forest ecosystems. Boreal forests, rich in forest resources and carbon stocking, play an important role in maintaining biodiversity and resisting climate change but receive less attention than other regions on Earth. Rapid warming at higher latitudes, boreal forests are facing an unprecedented wildfire crisis, with an increase in the frequency, intensity and burned area of wildfires. The prediction and prevention of boreal wildfires have become a major challenge for human response to climate change. However, the domain drivers of wildfire activities in the boreal forest remain to be explored, prediction models based on traditional algorithms with low accuracy, and the predictions of wildfire burned area, fire season and emission in the boreal forest are lacking. To address these concerns, this study first explored the spatio-temperal patterns of boreal wildfires; identified the driving factors responsible for ignition and emission; developed models of wildfire activities based on traditional and machine learning algorithms; and generated the wildfire susceptibility and emission maps in the boreal forest.