Application Of Extreme Value Statistics In The Analysis Of Water Level Of The Yamzho Yumco Lake

Extreme value statistics mainly studies the events that happen rarely but have great influences. Recently, in order to reduce the risks caused by extreme risky events, the theory of extreme value statistics are applied to predict and assess these extreme events by more and more researchers. The theory has been applied to many fields, especially in finance, hydrology and meteorology. In this paper, we introduce the theory of generalized extreme value and apply it into analysis the distributions of annual highest and lowest water level of Yamzho Yumco Lake. Then we predict the water level of the lake. This paper consists of three parts.In Part One, we introduce the resesrch background and the development of extreme value statistics.In Part Two, it refers to the theory of generalized extreme value model, including modeling, maximum likelihood estimation of parameters and the model diagnostic by probability plot, quantile plot, return level plot and density function plot.In Part Three, we use the annual highest and lowest water level data of the Yamzho Yumco Lake from1974to2010to establish the generalized extreme value models. The models show that the Yamzho Yumco Lake’s annual highest and the lowest water level tend to fall year by year; In2015-2033, the annual lowest water level changes around15meters of ecological restrictive level. However, the annual highest water level will decline to this standard during2037to2055; The lowest and the highest level will decline to the power warning level8.5meters about2090and2113initially. The research results will remind the related departments to strengthen the protection of Yamzho Yumco Lake and provide them with valuable references.