Abstract:
Forest fires are a global problem that often occurs in Indonesia, especially during the dry season. One indication of forest fires is the occurrence of hotspots. Monitoring of hotspots is part of the effort to make early warnings of the dangers of forest fires. The island of Kalimantan is an island where most of the land consists of peatlands and is prone to hotspots. The hot, dry season due to sunlight and the lack of rainfall can make peatlands a fire-prone area. This study uses the Zero-Inflated Poisson Regression model to predict the number of hotspots on peatlands by considering climatic factors, namely sun exposure and rainfall. This research was conducted in four square areas (Area-1, Area-2, Area-3, Area-4) with the Tjilik Riwut Meteorological Station as the center point. The study uses hotspot monitoring data from the Terra Satellite owned by NASA (the National Aeronautics and Space Administration) and climate data in the form of data on sun exposure and rainfall from the BMKG (Meteorology, Climatology, and Geophysics Agency). The results show that the Zero-Inflated Poisson Regression model can model the four observed areas quite well with an RMSE of 12.69 in Area-1, Area-2 with an RMSE value of 10.05, RMSE of 11.53 in Area-3, and finally Area-4 with an RMSE value of 16.40.
