Parameter Estimation On Curve-fitting Methodology In Hydrologic Frequency Analysis

Due to the impact of climate change, intensive activities of human beings and the development of society and economy, flood disaster would occur more frequently, and the loss would be more serious. So human might face more and more severe challenge of flood control safety. It also gives rise to a new challenge in the study of improving the precision of design flood. Hydrologic frequency analysis is one of the key parts and dominates on the rationality of design flood. Based on the thread of plotting position and curve-fitting criteria, the theory analysis frame and statistical test frame of curve-fitting method and the numerical computation of order statistics expected value has been discussed. Using statistical test and observed data, several curve-fitting methodologys has been evaluated. The main contents and results could be concluded as follows:Firstly, on the study of hydrologic stochastic simulation and statistical test, by comparison of the uniformity and statistical character of the stochastic simulation result, based on Monte-Carlo method and Quasi-Monte Carlo method, the GLP point set is the best series not only for uniformity and unbias but effectiveness. As for result evaluation of simulation sample based on every statistical test method, GLP-based packet simulation technique shows better than others.Secondly, on the study of plotting position by curve fitting method, the necessity and feasibility of using unbiased plotting position has been analysed. With extension generalized order statistics theory, a systemic study was executed on difference and relation between three kinds of order statistics. Then how to select plotting position has been discussed, and comprehensive assessment was studied by statistical testing. Also, based on double exponential formulas for numerical integration, a high speed and high precision computation framework for unbiased plotting position has been constructed.Thirdly, on the study of optimization criterion by curve fitting method, based on the order statistics theory, a systemic study has been made on unbias and effectiveness of several optimization curve fitting criterias. According to statistical test, the results of parameter estimation and design value were evaluated. A optimization curve fitting criteria allowing for unbias and effectiveness was constructed on basis of analysis above.Above all, a hydrologic frequency analysis framework considering unbias and effectiveness by curve-fitting method was constructed, which was applied to Huai River Basin. The main results of parameter estimation methodology show significant improvements in flood prevention and disaster reduction of hydraulic engineering, in timely exploring potential flood storage of reservoirs, in mitigating water resources shortage and fully development and utilization of water resources.