394799 - The evaluation of nonstationary population index flood model based on regionalization
Wednesday, June 6
2:00 PM - 3:30 PM
Location: Greenway IJ
Sunghun Kim, Seoul, South Korea – Yonsei University; Tae-Ho Jung, Seoul, South Korea – Yonsei University; Jun-Haeng Heo, Seoul, South Korea – Yonsei University
Regional frequency analysis is widely used to estimate more reliable and accurate quantiles of extreme hydrological events than at-site frequency analysis. In this approach, the observed data are assumed to be stationary. However, the hydrologic and meteorological data has become nonstationary due to the influence of the climate change. Therefore, nonstationary regional frequency analysis (NS-RFA) should be applied to these nonstationary data. The NS-RFA has been researched and some types of the index flood model which include time-varying parameters have been used. The aim of this study is to compare the performance of nonstationary population index flood (NS-PIF) model with other nonstationary index flood (NS-IF) models. For this purpose, several regions composed of 12 sites based on stationary and/or nonstationary generalized extreme value distributions were assumed. Using Monte Carlo simulation, the regional average values of the relative root mean square error and relative bias were calculated to evaluate the performance of nonstationary index flood models. As the results, NS-IF model with the time-independent index flood and time-dependent growth curve showed the best performance for the regions where all sites have the same trend signs. For the regions where the trend signs of sites are not identical, NS-PIF model showed the best performance.