| Title |
Computation of Criterion Rainfall for Urban Flood by Logistic Regression |
| Authors |
김현일(Kim, Hyun Il) ; 한건연(Han, Kun Yeun) |
| DOI |
https://doi.org/10.12652/Ksce.2019.39.6.0713 |
| Keywords |
도시홍수;한계강우량;유전자 알고리즘;로지스틱 회귀 Urban flood;Rainfall criterion;Genetic algorithm;Logistic regression |
| Abstract |
Due to the climate change and various rainfall pattern, it is difficult to estimate a rainfall criterion which cause inundation for urban drainage districts. It is necessary to examine the result of inundation analysis by considering the detailed topography of the watershed, drainage system, and various rainfall scenarios. In this study, various rainfall scenarios were considered with the probabilistic rainfall and Huff's time distribution method in order to identify the rainfall characteristics affecting the inundation of the Hyoja drainage basin. Flood analysis was performed with SWMM and two-dimensional inundation analysis model and the parameters of SWMM were optimized with flood trace map and GA (Genetic Algorithm). By linking SWMM and two-dimensional flood analysis model, the fitness ratio between the existing flood trace and simulated inundation map turned out to be 73.6 %. The occurrence of inundation according to each rainfall scenario was identified, and the rainfall criterion could be estimated through the logistic regression method. By reflecting the results of one/two dimensional flood analysis, and AWS/ASOS data during 2010~2018, the rainfall criteria for inundation occurrence were estimated as 72.04 mm, 146.83 mm, 203.06 mm in 1, 2 and 3 hr of rainfall duration repectively. The rainfall criterion could be re-estimated through input of continuously observed rainfall data. The methodology presented in this study is expected to provide a quantitative rainfall criterion for urban drainage area, and the basic data for flood warning and evacuation plan. |