A0224
Title: Robust fitting of the generalized Pareto distribution for extreme precipitation modeling: A case study in Japan
Authors: Muhammad Aslam Mohd Safari - Universiti Putra Malaysia (Malaysia) [presenting]
Tosiyuki Nakaegawa - Meteorological Research Institute (Japan)
Nurulkamal Masseran - Universiti Kebangsaan Malaysia (Malaysia)
Abstract: The probability integral transform estimator (PITE) is presented, a robust-efficient method for estimating the parameters of the generalized Pareto distribution (GPD). Designed to handle outliers commonly encountered in extreme event modeling, PITE enhances robustness while maintaining computational simplicity. Its properties, including efficiency and robustness, are analyzed through score function and breakdown point assessments, confirming its resilience to data contamination. Monte Carlo simulations show that PITE consistently outperforms traditional estimators, particularly in high-variability and contaminated data scenarios. Applied with the peaks over threshold method, PITE effectively models the tail behavior of extreme precipitation events. The method is utilized on daily precipitation data from 12 meteorological stations in southern Japan, a region vulnerable to typhoon-induced rainfall. GPD parameters are estimated, and return levels for 5-, 10-, 25-, 50-, 100-, and 200-year periods are computed. The results highlight regional variability in extreme precipitation, with stations such as Naze and Kumamoto exhibiting particularly high return levels, underscoring their susceptibility to intense rainfall. The robust application of PITE provides valuable insights for flood risk management in southern Japan.
