Predicting extreme precipitation events using analogues: Henan, China

This study focuses on predicting extreme precipitation events in Henan Province, China, by leveraging climate analogs to improve forecasting accuracy in the context of limited data. Daily precipitation records from Henan (1961–2022) and neighboring regions with similar climatic profiles, such as Anhui and Hubei, were analyzed. The Weibull hazard function was applied to assess the recurrence intervals of extreme precipitation, particularly at the 95th and 99th percentiles. Using generalized Pareto distributions, the return periods of extreme events were estimated. Additionally, a long short-term memory neural network was used to predict daily precipitation, showing strong consistency with observed data for 2023. The findings suggest that Henan Province encounters extreme precipitation at the 99th percentile approximately every 3.49 years, offering crucial insights for flood control and disaster preparedness. The integration of climate analogs and advanced machine learning techniques presents a valuable strategy for forecasting rare events in data-scarce regions.