Flood prone areas mapping by using statistical index and weights of evidence models (Case study: Madar Soo watershed, Golestan)

The aim of this study is preparation of potential flooding maps in Madar Soo watershed in Golestan province. For this end, the statistical index and weights of evidence models, various data layers and geographic information systems were used. The results of the applied models indicated that, among the effective factors for flooding, the elevation lower than 500m, flat slope, drainage density more than 0.33 (km/km2), distance from river of 500-2000m, malicious soil type, quaternary rocks and agricultural land with the greatest weight, were the most important factors in the Madar Soo watershed. The results also showed that 49.9% of the area of the Madar Soo watershed had high and very high potential for flooding, and 50.1% of the study area, had medium and low potential flooding. Validation of the results using Receiver operating characteristic (ROC) curve showed that the statistical index with AUC equal to 83% performed slightly better than the weights of evidence model with AUC equal to 80.7%. These results represent very good accuracy for both models for identification of flood prone areas in the Madar Soo watershed in the Golestan province.