Evaluation of flood susceptibility mapping using logistic regression and GIS conditioning factors

This paper investigates the application of logistic regression model for flood susceptibility mapping in southern Gaza Strip areas. At first, flood inventory maps were identified using Palestinian Water Authorities data and extensive field surveys. A total of 140 flood locations were identified, of which 70% were randomly used for data training and the remaining 30% were used for data validation. In this investigation, six causing flood variables from the spatial database were prepared, which are digital elevation model (DEM), topographic slope, flow accumulation, rainfall, land use/land cover (LULC), and soil type. Then, comprehensive statistical analysis techniques including Pearson’s correlation, multicollinearity, and heteroscedasticity analyses were used, to ensure that the regression assumptions are not violated. The uniqueness of the current study is its inclusiveness of influential causing flood parameters and vigorous statistical analyses that led to accurate flood prediction. Quantitatively, the proposed model is robust with very reasonable accuracy. The prediction and success rates are 76 and 81%, respectively. The practical and unique contribution of this investigation is the generation of flood susceptibility map for the region. This is a very useful tool for the decision makers in the Gaza Strip to reduce human harm and infrastructure losses.