Survey Of Study On Precipitation Resource In Fujian Based On GIS

Meteorological statistic data about the annual mean precipitation at 71 stations in Fujian Province dated for 1971-2000,DEM data,are used in this study.Based on Geographic Information System and the theory of spatial econometrics,the paper study the spatial distribution of precipitation of Fujian Region,by the use of spatial autocorrelation and spatial interpolation modelsSpatial econometrics model and GIS are combined to perform spatial autocorrelation analysis on spatial distribution of precipitation.Firstly,Moran's I is used to estimate the global spatial autocorrelation of precipitation in Fujian Region;secondly,Moran's LISA is used to estimate the local,spatial autocorrelation in Fujian Region.Global spatial association analysis shows that both of them have obvious spatial dependence on the whole region,while local spatial association analysis indicates that to what extent each of them exists on various local regions.Spatial econometrics model is classified as spatial lag regression model and spatial error regression model.On the basis of classification,either of models is given detailed exposition,then select suitable precipitation model by the use of model flow chart,classic multi-variable regression model of the annual mean precipitation is firstly built based on many elements,such as longitude,latitude,elevation,distance of coasting,etc.analyzing estimation results of classic multi-variable regression model,The research indicates that, spatial lag regression model is applicable to spatial autocorrelation of the prediction of annual precipitation,spatial lag regression model indicates also that the mostly elements affecting the distribution of annual precipitation are elevation,distance of coasting and some other factors which have a little clustering.Taking into account of some topography factors such as elevation,etc.different interpolation methods were used to simulate the distribution of precipitation based on geography information system(GIS)and geo-statistics technology,to analysize the precisions and applicability of different interpolation methods.The research indicated that the distribution of precipitation in fujian regions has obvious spatial dependence,being to what extent each of them exists on various local regions.Then,classic multi-variable regression model cannot meet the precision. Ordinary Kriging being a kind of method of linear and unbiased estimation,can not meet the precision of the distribution of precipitation,not taking into account of topographty elements.Compared with Ordinary Kriging,the prediction precision of Universal Kriging and Cokriging can be improved by adding some factors which have a little clustering,such as elevation and distance of coasting elements.though IDW can improve the precision,there are something wrong with the fact in some regions.Combined method is a better successive interpolation approach to simulate the distribution of precipitation, taking topographty elements and geo-statistics into consideration.