Spatialization Of Historical Wheat Area Statistics In Anhui Province Based On GIS

Because of rough positioning and low spatial resolution of the traditional statistical data, it is difficult to analysis social statistical data and natural ecological data together. Agricultural statistics data have great value because the agricultural is an important industry of our country. But these data mainly use the administrative units as statistics and release unit, which can not reflect the internal difference of the spatial distribution of every administrative unit. In the actual application, however, we often need analysis specific geographical area. This limits the applications of agricultural statistical data. Traditionally, the way to spatialize statistics indexes of an area was assigning the grids in a regional unit the same value. The thematic maps we got by this method have two features:the statistical index of every grid in same administrative unit has the same value; the indexes of girds in different administrative units differ from each other. In fact, the distribution of the social and economic statistics data of real state was not consistent. Also, the values of indexes in different units do not change so great. For the planted area of wheat, spatialization of wheat area has important significance, and the study of the spatialization method is imperative.The goal of this study is to find a suitable method for the spatialization of wheat planting area in Anhui province. It used statistical data of wheat area, DEM, slope, temperature, precipitation and ETM+images in Anhui province for the main data source, selected two commonly analysis methods used in multi-source data fusion, the multiple regression analysis method and gray relational model, analyzed these factors affected wheat cultivation Compared the results of these methods with traditional evenly distributed to determine the best one of these spatialization methods. By using the statistical data, DEM, slope, temperature, precipitation and ETM+images data, based on the multiple regression analysis method and grey correlation model method, structure regression model and the grey relation model of spatialization. The value of statistical data of wheat planting area was divided into1km*1km grids. And so we can obtain the wheat growing area of each grid. Finally, by supervised classification based on high-resolution ETM+images, we got the more exact wheat growing region. Then, we compared it with the results got by multiple regression analysis method and grey correlation model method by pixel, we got the spatialization precisionFrom the results we can know that compared to traditional evenly distributed statistics up to the individual grid cells, the precision of the method of multiple regression analysis and gray relational model has improved a lot, and accuracy from about70%of the average allocation method to nearly90%. And through the three counties of verificaion results may be found, the precision of spatialization of wheat planting area and the complexity of the study area has a great relationship. If the study area is a complex case, the accuracy will be lower than the study area, a single place. The results show that the accuracy of the multivariate regression model and gray relational model are high, the spatial accuracy of the gray relational model is slightly higher than the spatial accuracy of the multiple regression model. It states that the multivariate regression model and gray relational model can effectively overcome the shortcomings of the traditional regional indicators uniformly distributed and different regions have great change. Achieving the purpose of add spatial information to statistical data. Gray relational model method is applied to the historical and statistical data for1985and1995, found results consistent with the actual situation.The innovation of this paper lies in spatialization of statistical data of Anhui wheat area. The administrative boundaries of the statistical data limit our application of statistical data. Through the spatialization, we can get statistics on each grid size, and greatly expanded the scope of application of statistical data.