Parameter Estimation Of Generalized Pareto Distribution And Application In Hydrological Field

Generalized Pareto Distribution(GPD)was first proposed by Pickands in 1975,after which many scholars have done further research.GPD is widely applied in many fields,such as: weather,flood,precipitation,earthquake,life,finance,reliability and many other fields.GPD research nowadays has become very hot in extreme value theory area.Since,it not only takes the maximum or minimum values of data into account,but also focuses more on a certain kind of data which is beyond the threshold.Those data approximately subject to GPD model,so the GPD model are widely accepted and recognized by most scholars.In this thesis,we studied the parameter estimation problem of GPD model,using maximum likelihood estimation,moment estimation,maximum likelihood estimation,probability estimation and moment estimation/least squares estimation as different estimation methods.First,we used matlab to generate the GPD random number.Then,the parameters could be estimated by different methods.At last,we analysed the estimated parameter value to compare different estimation methods and summarized the merits,dismerits and inclusiveness of each method.At the end of this thesis,we applied GPD and GEV distribution to hydrological to give prediction of the maximum water level in the next T years,by which we could make preparation for flood control measures to reduce the impact of natural disasters.