Evaluation of Random Forest model for forest fire prediction based on climatology over Borneo

Abstract

Indonesia has entered an alarming condition related to forest fires. It is becoming a seasonal hazardous phenomenon in tropics. As the largest tropical forest in Indonesia, Borneo is the most susceptible area to fire especially in dry condition. Forest fires are threatened by climate, human activities, and ecosystem processes, but only climate variable can be quantified well in Borneo. This research aims to evaluate random forest model in predicting forest fires based on the climate variables and satellite data of burned area. Prediction of forest fires is expected to reduce the impact of forest fires in the future. Based on analysis of spatial and annual variability, the random forest model with all selected climate variables can represent the forest fires event over Borneo.