Introduction of Confidence Interval Based on Probability Limit Method Test into Non-Stationary Hydrological Frequency Analysis

Nonstationarity in hydrological variables has been identified throughout Japan in recent years. As a result, the reliability of designs derived from using method based on the assumption of stationary might deteriorate. Non-stationary hydrological frequency analysis is among the measures to counter this possibility. Using this method, time variations in the probable hydrological quantity can be estimated using a non-stationary extreme value distribution model with time as an explanatory variable. In this study, we build a new method for constructing the confidence interval regarding the non-stationary extreme value distribution by applying a theory of probability limit method test. Furthermore, by introducing a confidence interval based on probability limit method test into the non-stationary hydrological frequency analysis, uncertainty in design rainfall because of lack of observation information was quantified, and it is shown that assessment pertaining to both the occurrence risk of extremely heavy rainfall and changes in the trend of extreme rainfall accompanied with climate change is possible.

Comparative Study of Regional Frequency Analysis and Traditional At-Site Hydrological Frequency Analysis

Hydrological frequency analysis plays an indispensable role in the construction of national flood control projects. This study selects the stations with the smallest and largest discordances in the nine homogeneous regions of Sichuan Province as the representative stations, and results obtained by regional frequency analysis are compared with those obtained by traditional at-site hydrological frequency analysis. The results showed that the optimal frequency distribution of each representative station obtained by traditional at-site hydrological frequency analysis and the ones of corresponding homogeneous regions obtained by regional frequency analysis were not necessarily consistent, which was related to the site and homogeneous regions. At the same time, there were also differences between the fitting of the theoretical rainstorm frequency curve obtained by the two methods and the observation. In general, in each homogeneous region, the results obtained by regional frequency analysis and traditional at-site hydrological frequency analysis at the stations with the largest frequency analysis were quite different. The design values obtained by the two methods were also increasingly different with the increase of the return period. The study has specific reflections on the differences between regional frequency analysis and traditional at-site hydrological frequency analysis.