Remotely sensed dry matter productivity and soil moisture content as potential predictors of arid rangeland wildfires: A case study of Kgalagadi District, Botswana

Fire is a critical tool for managing rangeland ecosystems; however, the increasing wildfire occurrence poses considerable danger to ecosystem continuity. Predicting fire and mapping in highly flammable rangelands. To identify potential remotely sensed variables prediction, this study employed Random Forest (RF) classifier using selected environmental assess their possible use prediction Kgalagadi District, Botswana. The used 107,883 active points from Visible Infrared Imaging Radiometer Suite (VIIRS) sensor 2015 2021. Datasets of Dry Matter Productivity (DMP), Soil Moisture (SM), Land Surface Temperature (LST), Live Fuel Content (LFMC), Dead (DMFC) were analysed ArcMap 10.7 Esri©. RF model developed gave an Out Bag (OOB) error 9.91% overall accuracy 90.15% classifying fires non-fire test dataset. results also showed Kappa coefficient 0.803, with 88.25% 91.76% producer user accuracies, respectively, points. DMP was most significant variable Mean Decrease Accuracy (MDA)= 1,055.20 Gini (MDG)= 9.328.62), followed by SM (MDA= 828.39 MDG= 15,745). LFMC DMFC found be weak detecting fires. It recommended that field studies carried out area calibrate these improve contribution accurate as literature shows they are prediction.