FLOOD RISK ESTIMATION GIS-BASED SUPPORT VECTOR MACHINE

Abstract

Support Vector Machine (SVM) is statistical technique widely used in natural hazard management assessment. The main objective to assess flood risk with spatial area factors. Which indicates the predictive ability of each factor used analysis on Geographic Information system (GIS)- Based Support Vector Machine in Mueang Ubon Ratchathani District and Warin Chamrap District. A flood susceptibility map was produced by mapping the flood locations. Which was divided into training and testing datasets using random selection. The spatial database was constructed using physical factors cause flooding of the study area. To compare forecasting performance of support vector machine. Flood Susceptibility Index (FSI) introduced to compare flood risk area and validate of model. Additionally, the area under the curve (AUC) was used to validate the resulting flood risk map. The validate results demonstrated that the prediction rate curves for flood risk maps generated by the SVM was 0.8209 while success rate was 0.8945. The result demonstrated that all physical factors cause flooding of the study area have reasonably positive ability to analyze. Whereas river distance was a highest positive performance factor. The next were altitude, drainage density, flow accumulate and slope, respectively. It can be concluded that SVM technique is an effective tool for identify and trend of flood. Which risk communication from study area to government sector or private sector comes into role in prevention and flood relieve.