Study On The Regional Extremal Model For Extreme Precipitation In The Different Scales

In the context of climate change,with the increasing frequency and intensity of extreme rainfall,damaging floods and sediment often emerge in rills due to the combination of landscape and intense rainfall,which put in jeopardy the development of economy and society and bring huge harm to the natural environment.To this end,how to make the risk assessment of connectivity has become more and more crucial and emergent.The incidences and intensity of extreme rainfall are association with probability statistics.Classical probability distribution is a practical and significant tool for a visualized description of statistical characteristics of extreme climate events.However,this method has a disadvantages: ignore spatial dependence among extreme values.Probability distribution usually fit a series of rainfall data of a single situation rather than that of all stations.It is likely that a single extreme event affects several locations and then the areal modeling of extremes is essential if the spatial dependence of extremes has to be appropriately taken into account.An extreme rainfall event occurred in one area,while any other extreme rainfall events happened in other neighboring area at the same time,which would cause widespread damage.So it is necessary to establish a regional model considering spatial dependence among all stations.The rainfall process,a complex geographical problem,is related not only to synoptic system,but to terrain factors.Studies of extreme rainfall,although useful from a statistical point of view,are in accordance with the geographic environment.In order to study on the extreme rainfall influenced by natural factors,this study will put some terrain factors(i.e.altitude,slope,aspect,etc.)into the regional extreme rainfall model.This study took the east area of Huaihe river basin as the study area,and made the regional extreme precipitation model based on the Max-stable process considering the geographic environment,which could make up for the short of single-value extremal model,that is,the correlation among stations.This study analyzed the optimal regional extreme precipitation model under the optimal scale based on the geographic factors in the different scales,which could provide the reference for future researches.The main contents and steps of research are as follows:Firstly,chose the three most relevant factors with extreme precipitation in the different scales used by the grey relevant method according to the feature of study area;secondly,made the regional extreme precipitation model for the study area in the different scales by the above three factors based on the Max-stable process;thirdly,extracted the values from the above established models and selected the model in the optimal scale by comparing with the actual extreme precipitation and Akaike Information Criterion;finally,analyzed the influence of different stations for the model.The main conclusions of research are as follows:(1)The aspect,altitude and NDVI are the three most relevant geographic factors with the extreme precipitation in the different scales by the grey relevant method.(2)As for the optimal regional extreme precipitation model in the different scales,according to the correlation between the actual values and the simulation values,the optimal regression models with geographical factors are M5,M5,M5,M5,M3,M5,M1,M1,M3 and M3 from 500 m to 5000 m respectively.Then the optimal regional extreme precipitation model in the different scales are made on the above regression models with geographical factors.(3)As for the optimal scale of the regional extreme precipitation models,considering the correlation between the actual values and the simulation values and relative error,3000 m is the optimal scale of the regional extreme precipitation models,and M5 is the optimal regression models with geographical factors at the optimal scale and the model based on the M5 is the optimal regional extreme precipitation model.(4)As for the number of stations affect the models,the change of number of stations have some impact on the spatial distribution of results,the trend of variations and the effects of models.With the increase of stations,the spatial distributions of extreme precipitation in the study area change from the low edge of west and high east to the high edge of west and low east patterns with some high small areas.The majority stations show a trend of rising in the first stage and then decreasing,and distribute in the middle areas,and other stations show an opposite trend.The estimation effects are different under the different numbers of models.When the number of model is 17,the estimation effect is small.