Application of remote sensing and machine learning algorithms for forest fire mapping in a Mediterranean area

Forest fire disaster is currently the subject of intense research worldwide. The development accurate strategies to prevent potential impacts and minimize occurrence disastrous events as much possible requires modeling forecasting severe conditions. In this study, we developed five new hybrid machine learning algorithms namely, Frequency Ratio-Multilayer Perceptron (FR-MLP), Ratio-Logistic Regression (FR-LR), Ratio-Classification Tree (FR-CART), Ratio-Support Vector Machine (FR-SVM), Ratio-Random (FR-RF), for mapping forest susceptibility in north Morocco. To end, a total 510 points historic fires inventory map 10 independent causal factors including elevation, slope, aspect, distance roads, residential areas, land use, normalized difference vegetation index (NDVI), rainfall, temperature, wind speed were used. area under receiver operating characteristics (ROC) curves (AUC) was computed assess effectiveness models. results conducting proposed models indicated that RF-FR achieved highest performance (AUC = 0.989), followed by SVM-FR 0.959), MLP-FR 0.858), CART-FR 0.847), LR-FR 0.809) fire. outcome prediction risk areas can provide crucial support management Mediterranean ecosystems. Moreover, demonstrate these novel increase accuracy studies approach be applied other areas.