Recently, the temporal and spatial patterns of flood disasters have become more complex and unpredictable due to the effects of climate change and abnormal weather. It is necessary to prepare a flood hazard map that includes information such as flood risk level and flood coverage as unstructured measures to prevent flood damage caused by such weather changes. In fact, in order to create a high-precision flood risk map, digital information such as topography, geology, weather, and various database related to social economic are required. Information such as flood area and depth of flooding is acquired through modeling of rainfall-runoff-flood analyses. However, in some areas, especially in developing countries, such data is lacking or inaccessible, making flood hazard mapping impossible or its accuracy is very low. Therefore, in this study, A total of eleven single and mixed topographical factors were calculated for each watershed using only DEM topographic data such as ASTER or SRTM in large flood areas that occurred in five developing countries including North Korea Geumya River in 2007, Padma River, Jamuna River and Meghna River in Bangladesh in 2007, CitarumRiver in Indonesia in 2010, Chao Phraya River in Thailand in 2011 and Irrawaddy River in Myanmar in 2015 and in comparison with actual flood watershed, optimal topographic factors were derived using linear binary classifiers and ROC(Receiver Operation Characteristics)analysis, and the accuracy of predicted flood watershed and actual flood watershed was verified.
In this study, the flood risk mapping method based on the topographical factor of ROC analysis method has a great advantage in that it can easily distinguish the flood risk areas using only DEM for the ungauged watershed. It is relatively accurate with flood watershed. However, non-flooded watersheds that are similar to actual flood watersheds in the target watershed, they may be predicted as flood watersheds and may be overestimated from actual flood watersheds. Also, predictive uncertainty can be increased if the topography of the actual flood watershed has different geomorphological features from those of the flood watershed derived from the optimal topographic factors when creating a flood map using only a single optimal topographic factor. Therefore, it is considered that sufficient simulations should be performed for various regions in order to apply the flood risk analysis to flood analysis by effectively dividing the target watershed.