Probabilistic Hotspot Prediction Model Based on Bayesian Inference Using Precipitation, Relative Dry Spells, ENSO and IOD

Increasing global warming can potentially increase the intensity of ENSO and IOD extreme phenomena in future, which could potential for wildfires. This study aims to develop a hotspot prediction model Kalimantan region using climate indicators such as precipitation its derivatives, IOD. The was developed Principal Model Analysis (PMA) initial basis. overall performance is evaluated concept Cross-Validation. Furthermore, model’s will be improved Bayesian Inference principle so that average increases from 28.6% 61.1% based on coefficient determination (R2). character each year development process also cross validation. Since indicator we used integrated with index, strongly influenced by phenomena. To obtain better when estimating future forest fires (related El Niño positive IOD), years high number hotspots coinciding occurrence are early (PMA). However, tends overestimate value, especially lower strength Therefore, low hotspots, normal La Niña, improvement stage (Bayesian Inference) correct overestimation.