Evaluation of Shannon entropy in order to flood probability and susceptibility mapping at Haraz catchment

Abstract

One of the main measures in order to financial and life damages reduction is a production of flood susceptibility mapping. In the current research, Shannon entropy has been studied to determine the flood probability and susceptibility mapping. In this research 211, flood locations were prepared. 211 flood locations are divided into two categories based on random selection: 70 % for training and 30% for validation of final map. In the next step, 10 flood conditioning factors such as slope, plan curvature, distance from the river, altitude, rainfall, stream power index, topographic wetness index, lithology, land use and NDVI was selected. By using the Shannon entropy the weight of each factor in flooding have been determined. ROC curve has been drowned and area under the curve (AUC) was determined for validation of flood susceptibility mapping by success rate and prediction rate. The result showed that AUC for success rate in entropy model was 92.53% with a prediction accuracy of 91.42%. The high accuracy of the Shannon entropy model showed that this model is proper for flood susceptibility mapping.

Keywords

• shannon entropy
• haraz
• flood susceptibility
• auc