Researchers at Aalto University have developed a neural network model that accurately predicts the occurrence of wildfires in peatlands. They then used it to assess different management strategies and identified interventions that could prevent three-quarters of the wildfires. The same approach could be used to predict and prevent wildfires in other contexts, say the researchers.
The team trained the model using data from 2002-2019 in the Central Kalimantan province of Borneo in Indonesia. Based on variables such as the type of land cover and pre-fire indices of vegetation and drought, the model predicts the likelihood of a fire at each spot on the map, producing an expected distribution of fires for the year. The team then tested how different interventions affected the predictions and identified strategies that would lead to a 50-76% reduction in fires.
Although the model in this study was trained specifically for peatland fires in Indonesia, the researchers say similar AI models could be developed for wildfires in other regions to help countries around the world deal with increasingly intense fire seasons.
‘Our goal with the model is to quantify how different prevention strategies would work,’ says Associate Professor Matti Kummu, who led the study team, noting that any prevention strategies would have to balance risks, benefits and costs. ‘Our system doesn’t provide direct solutions, but policy-makers can use it to make informed decisions about the best strategies.’
