Flood Susceptibility Mapping Using SVM Model and the Impact of Sabzevar City Development on Increasing Flood Peak and its Damages

Flood forecasting is considered through land use and building density in most urban designs, whereas this method has not been very successful based on the experiences. This study aimed to propose a proper method for flood forecasting in Sabzevar city, emphasizing the relationship between the parameters of hydrological models, urban development, and the scenarios considered in the comprehensive plan of 2005-2006 in Sabzevar city.

This research is applied and descriptive-analytical in terms of the aim and method, respectively. The data were collected using documentary and field methods. The flood susceptibility map of Sabzevar city was prepared using the SVM model. The model was validated using the ROC curve. The basin’s hydrology was studied using the Sabzevar Synoptic Meteorological Station data from 1358 to 1398. Urban hydrological models and Geographic Information System were used to predict the runoff during different urban development scenarios in the return periods of 2, 5, 10, and 25 years.

Due to the higher building density, the susceptibility to flooding risk vulnerabilities is higher in districts 1, 2, and 3 of Sabzevar city ROC showed the acceptable validity of the SVM model (AUC=0.9123). Surface runoff forecasts for the return periods and different scenarios indicate that the channels are not proper for runoff transmission.

It is very important to pay attention to the ratio of permeable and impermeable surfaces in comprehensive plans in order to prevent flooding. The SVM method is sufficiently compatible with Iranian conditions and can be used in similar conditions.