| description abstract | One of the most remarkable factors affecting the flood susceptibility in watersheds is the change in land use and land cover (LULC). Moreover, machine learning (ML) methods showed reliable efficiency in predicting LULC and preparing flood susceptibility maps. Therefore, this study aims to predict LULC of the future using land change modeler (LCM), and then investigate the effects of LULC changes on the flood susceptibility map using ML algorithms in the Talar watershed, Iran. In this regard, LCM model based on artificial neural network (ANN) and Markov chain analysis used to predict LULC in the future. Then, the effect of LULC changes on flood susceptibility was investigated using some ML algorithms, including ANN, logistic regression (LR), random forest (RF), and support vector machine (SVM). These four algorithms were used for the first time simultaneously to generate flood susceptibility maps in the present and future. Nine influencing factors, including altitude, slope, distance from rivers, land use and land cover, topographic wetness index, stream power index, profile curvature, rainfall, soil type, and 200 flood and nonflood locations, were utilized to create the flood susceptibility map. The selected ML models were assessed using root mean square error (RMSE) and coefficient of determination (R2). RMSE values showed high performance of SVM, LR, RF, and ANN models with values of 0.125, 0.175, 0.1920, and 0.2834, respectively. Furthermore, R2 values of those models were computed as 0.9373, 0.8773, 0.8667, and 0.8525, respectively. Overall, the findings demonstrated the performance of the LCM method in forecasting the LULC map and the capability of the selected ML algorithms to produce flood susceptibility maps in the study area. In cities and watersheds all over the world that are dealing with LULC changes and a possibility of flooding, the findings and algorithms used in this study can be applied. This aids officials and decision makers to make more effective and efficient decisions in flood prevention and management. | |
