Semi-parametric Geographically Weighted Regression Modelling for Flood Damage Assessment

Climate change has been a global issue since the 19th century. The increase in rainfall variability, which covers the increase in the earth’s total precipitation, will definitely lead to frequent and more severe flood disasters. As the damage increases year after year with floods as the most costly disaster among these hazards, Korea has to improve its technological responses and countermeasures. This study aims to develop a semi-parametric geographically weighted regression which can estimate a flood damage of Gunsan City. The model include parameters like flood depth, flood damage, flood duration, inundated area, family income and land price. This study collects flood depths, flood duration from GIS based flood inundation map and flood damages of local buildings from damages report collected by local government after flooding on August, 2012 in Gunsan City. Flood damage estimation of residential, commercial and agricultural facilities was done by Ordinary Least Squares and Geographically Weighted Regression using collected data. Considering the building and GIS-based spatial information, flood damage by GWR is more appropriate than OLS.