Study on quantile estimates of extreme precipitation and their spatiotemporal consistency adjustment over the Huaihe River basin

摘要

The quantile estimates and spatiotemporal consistency of extreme precipitation are studied by regional linear frequency analysis for Huaihe River basin in China. Firstly, the study area can be categorized into six homogeneous regions by using cluster analysis, heterogeneity measure, and discordancy measure. In the next step, we determine the optimum distribution for each homogeneous region by using two criteria of Monte Carlo simulations and the root-mean-square error (RMSE) of the sample L-moments. A diagram of L-moments ratio is used to further judge and validate the optimum distribution. The generalized extreme value (GEV), generalized normal (GNO), and generalized logistic (GLO) for 24-h duration are determined to be the more appropriate distribution based on the two criteria, L-moments ratio plot, and the tail thickness of curve in adjacent regions. A summary assessment can provide the more reasonable distribution, which avoids arbitrary results from single test. An important practical element of this study that was missing from previous works is the quantile spatiotemporal consistency analysis, which helps identify non-monotonicity among quantiles at different durations and reduces the gradient of estimates in the adjacent regions. Abnormality and spatial discontinuation can be removed by distributing the surplus of the ratio and twice different interpolation. A complete set of spatiotemporal consistent quantile estimates for various duration (24 h, 3 days, 5 days, and 7 days) and return periods (from 2 to 1000 years) can be obtained by using the abovementioned method in the study area, which are in the agreement with the observed precipitation extremes. It will provide important basis for hydrometeorological research, which is of significant scientific and practical merit.