History and Perspectives of Hydrologic Frequency Analysis in Japan

Abstract

Hydrologic frequency analysis provides basic information for the planning, design, and management of hydraulic and water resources systems for promoting the river basin quality and human health. It uses meteorological/hydrological extreme-value data and probability distribution functions to estimate T-year events (quantiles). Reviewing the history of hydrologic frequency analysis in Japan, this study describes goodness-of-fit criteria such as the standard least-squares criterion and Akaike information criterion and their applications. The jackknife and bootstrap methods are introduced as useful resampling methods for bias correction and quantile variability estimation. As future directions, this study proposes the incorporation of (1) partial duration series or peaks-over-threshold series, if available, instead of the annual maximum series; (2) a nonparametric method using empirical distributions for larger samples with more than 100-year observation period; and (3) probable maximum precipitation or probable maximum flood into frequency analysis.