The Peak Over Threshold Flood Series Frequency Analysis And Calculation

The conventional flood frequency analysis choose annual maximum flood series as the sample. However, as most hydrological stations were built only for decades, so the length of observation flood series are not sufficient, When we calculated the flood in a probability of 1%or 0.1%, the flood series is too short to calculate accurate design flood. POT flood frequency analysis firstly choose a threshold value, and then select the POT sample series which is great than threshold value, So POT frequency analysis can increase the length of sample and improve presision of design flood.The thesis consulted literatures of POT flood which was in the recent years. It included a system research of three aspects:POT samples selection,POT flood models optimization and POT flood frequency analysis. Based on the other researcher results,the thesis also applied and studied goodness-of-fit test theory,conventional parameters estimation,intelligent parameters estimation,multivariate Compound Extreme Value theory in POT flood frequency analysis, and finally developed a software of POT flood frequency analysis. The major results of the thesis is as follows:(1)Summarized the POT flood's research status and latest development trend. Introduced application of Normal distribution type,Γdistribution type, Extreme value distribution type, Wakeby distribution type and Logistic distribution type in POT flood frequency analysis, and estimated the parameters of different distributions by conventional parameter estimation methods,intelligent parameter estimation methods.(2)Employed five goodness-of-fit tests, such as the Kolmogorov-Smirnov (K-S) statistic, Cramer-von Mises (C-M) statistic, Aderson-Darling (A-D) statistic,modified weighted Watson (M-W-W)statistic, Liao and Shimokawa (L-S) statistic. Also applied the goodness-of-fit test to inspect five types distributions. By compiling computer cddes, calculated the critical values of different goodness-of-fit tests of different distributions.(3)According to the multivariate Compound Extreme Value theory, applied Possion distribution to combined with five type distributions, getting the formulas of annual maximum POT flood of different distributions, and then compiled computer codes to implement calculation of annual maximum POT flood.(4)Based on the flood data of twelve hydrological stations of Northern Shaanxi, Applied nine distributions to calculate POT flood frequency analysis for every hydrological station. calculated the parameters of different distributions and carried on the goodness-of-fit test for every distribution for every station. Finally, determined the optimal POT flood frequency analysis model and parameters estimation method for northern Shaanxi regional.