A comparative study of flood sensitive areas modeling (Isfahan Province)

One of the basic solutions to control and reduce the destructive effects of floods is to identify flood prone areas in the regions. Identifying flood sensitive points is one of the best methods for planning and identifying areas affected by floods. For this reason, determining flood-sensitive areas plays an important role in flood management in natural resources. For this reason, the current research is trying to determine the flood-prone areas in Isfahan province by using two methods of random forest machine learning and support vector machine and 3327 flood occurrence points. Environmental factors in four main groups including topographical factors (altitude, slope direction, steepness of slope), climatic factors (rainfall, relative humidity, wind, temperature), biological factors (vegetation and soil moisture) and man-made factors (distance from areas residential, distance from road, distance from agricultural land, distance from waterway) were prepared. The accuracy of the used models was evaluated using the area under the graph (AUC) and cross evaluation statistics. Examining the AUC index showed that both models had good accuracy, although the random forest model (AUC = 0.97) had higher accuracy than the support vector machine model (AUC = 0.86). According to the results of the random forest model, about 41% are in the high risk class and about 20% are in the low flood risk class. Also, in the support vector machine model, about 29% is in the high risk class and about 30% is in the low risk class.