<![CDATA[Forest and land fires are a major environmental concern in Banjar Regency, South Kalimantan, with a burned area of 1,812.80 ha recorded in 2023. This study aims to model fire susceptibility levels using the Random Forest algorithm based on remote sensing data. The data utilized include Landsat 8 imagery from 2023 with extracted spectral indices such as NDVI, NBR, NDWI, MSI, and BAI, along with fire hotspot data from the Banjar Regency Disaster Management Agency. The model was trained using data from June to mid-September and validated with data from mid-September to November 2023. Results indi-cate that the northern and central areas of Banjar Regency exhibit the highest fire susceptibility. The susceptibility map was categorized into five zones based on fire probability. Accuracy assessment using a confusion matrix yielded an overall accuracy of 71.64% and a Kappa coefficient of 40.81%. These findings demonstrate that the Random Forest method is effective in iden-tifying fire-prone areas with high efficiency and minimal input data. This model provides a valuable tool for spatially targeted fire prevention and mitigation planning.]]>
Copyrights © 2025
