Analyzing the effects of impervious surface area and rainfall variables on flood hazard in Tainan City, Taiwan

Flood, as known as the most frequent natural hazard in Taiwan, has induced severe damages of re sidents and properties in urban areas. The flood risk is even more severe in Tainan since 1990s, with the significan t urban development over recent decades. Literatures have indicated that the characteristics and the vulnerability o f flood are affected by the increase of impervious surface area (ISA) and the changing climate condition. There for e, this study is basically conducted under the assumption that land cover and climate change could be two major fa ctors that influence the occurrence of flood hazard. Tainan City is selected as the study area, not only because of it s significant urban development, but also the high flood vulnerability over the city. To clarify the linkage between urban development, rainfall variables (total rainfall, rainfall duration and rainfall in tensity) and the likelihood of flood occurrence, this study used logistic regression to analyze the relationship betw een rainfall variables, ISA and historical flood events. Specifically, rainfall records of 18 rainfall events from 2004 to 2014 associated with flood hazards (2004-014) were collected and mapped, and Landsat images of year 2004, 2007 and 2014 were used to generate the ISA with SVM (support vector machine) classifier. The result shows that rainfall variables and ISA are effectively and significantly correlated to the flood occurrence in Tainan City. Acco rding to the derived logistic function, the study is expected to present the future flood potential of Tainan City, if reliable future projections of ISA and rainfall estimates can be obtained.