Extreme Rainfall Analysis and Return Period Estimation Based on Extreme Value Theory

Climate change has resulted in frequent extreme weather events such as heavy rainfall and heat waves in Japan, making accurate forecasting and countermeasures an urgent issue. Therefore, it is urgently required to analyze the statistical characteristics of extreme rainfall events using the extreme value theory (EVT). The generalized extreme value (GEV) distribution, a core model for EVT, was applied in this study to rainfall data collected in Kakunodate, Akita Prefecture, Japan, spanning May 1976 to December 2023. The analysis results confirm the presence of extreme rainfall events. Through model fitting, the GEV parameters representing location, scale, and shape were accurately estimated. The model demonstrated a good fit, particularly for moderate-intensity rainfall. However, notable uncertainties emerged in the prediction of the most extreme events. Return period analysis results indicated that extreme rainfall events occur at intervals ranging from 2 to 100 years, suggesting the necessity of incorporating safety margins into long-term forecasting frameworks. Considering the increasing frequency of such events, cross-validation with alternative statistical methods and the potential adoption of non-smooth GEV models are recommended to enhance predictive reliability. Overall, the results of this study highlight the need for adaptive and flexible revisions to infrastructure design criteria in response to evolving patterns of extreme weather.