Identification of the Most Important Environmental Variables in Spatial Prediction of Flood Prone Areas using the Maximum Entropy Model in Parts of Golestan Province

Flood is a destructive natural hazard over the recent years. In the current study, the maximum entropy model as an advanced data mining model was used to model and provide the spatial maps of flood prone areas in Saliantapeh Watershed, Golestan Province, with an area of 4515.47 km2. For this purpose, the flood inventory map was prepared based on available reports and field surveys. Then, 13 effective variables including the digital elevation model, slope percent, slope direction, rainfall, distance from drainage network, land use, lithology, soil texture, plan curvature, profile curvature, topographic moisture index, drainage density and flow capacity index were identified and introduced to the model. After that, three different series of  flood risk points (i.e. ds1, ds2, and ds3) including 70% for training and 30% for validation of the model were randomly prepared to evaluate the accuracy and robustance of the model based on the ROC Index. The results showed that the maximum entropy model with high accuracy (above 90%) predicted flood prone areas. Moreover, in this study, the degree of importance of the variables was investigated by the model and the results demonstrated that the two factors of the drainage density (about 49% of importance) and distance from the streamflow (about 15% of importance) were detected as the most important environmental factors affecting flood in the studied area.