Application On Regional Extreme Rainstorm Events Based On Multiple Variable Extreme Value Distribution Model

Based on daily precipitation of 26 observation stations in Yangtze and Huaihe river basin, we analyzed the probability distribution of regional extreme precipitation events of each station and its region in flood season (May to September) in recent 57a (1951-2007). This paper mainly contains such research contents:we sample the rainfall using AM sampling and multiple a year sampling and calculate the extreme precipitation values with single variable Gumbel distribution calculate the distribution probability and estimate the recurrence period of extreme precipitation values.We use the multiple variable Gumbel distribution to fit the regional extreme precipitation events and analyze the probability characteristics of regional extreme values.We spread two-dimensional Gumbel distribution model to GEV distribution model and derived the all kinds of two-dimensional combination distribution formula. We applied the formula into probability distribution description of regional extreme precipitation events. We used the GPD distribution to analyze the probability distribution of extreme value of precipitation. Finally we introduced the multiple variable linear GPD distribution and two-dimensional GPD distribution to analyze regional extreme precipitation events furthermore.Here are the research conclusions of this paper.(1)Although single variable Gumbel distribution can fit the probability distribution of the only rainstorm events well in Yangtze and Huaihe river basin, using the multiple variable Gumbel distribution to analyze the joint probability characteristics of rainstorm peak and total rainfall in Yangtze and Huaihe river basin can conform to reality better;(2)Only use single variable probability analysis method to confirm quantile of rainstorm peak or quantile of total rainfall may overestimate the rainstorm events.It's suitable to analyze the two main features of regional strong precipitation process with Multiple variable Gumbel distribution;(3)We spread two-dimensional Gumbel distribution model to GEV distribution model and derived the all kinds of two-dimensional combination distribution formula.It's feasible to use rainstorm stations to represent the true rainstorm areal extent of regional strong precipation process. It's feasible to use the multiple variable GEV model to analyze the joint probability distribution and multiple variable GEV may more suitable to analyze the joint probability characteristics of the variable;(4)It's feasible to analyze the probability distribution of combain variables constituted by the linear relationship between regional rainstorm peak and regional total rainfall;(5)the multiple variable GPD can also analyze the joint probability characteristics of intra-region bivariate rainstorm events well.From the statistical of acculturation, it's more suitable to analyze the probability characteristics of extreme precipitation events with multiple variable extreme values distribution model which can get better effect than the single variable distribution model we used. It can provide evidence for evaluation of meteorological disasters risk and play a role in prevention and mitigation.