基于空间异质分区的残差IDW插值方法

Residual Inverse Distance Weighting Spatial Interpolation Method Based on Spatial Heterogeneity Sub-region

空间插值可以利用已有观测数据修补缺失的观测数据,也可以利用离散数据构建连续的表面数据,但现有的空间插值方法没有充分考虑空间数据的异质性。该文提出一种基于空间异质分区的残差反距离加权插值方法(RRIDW)。首先根据采样点属性值对研究区域进行空间异质分区;为了进一步去除不同子区域内的空间趋势,对每个子区域计算趋势面,进而计算得到采样点属性值的异质分区残差,利用属性值残差进行反距离加权插值;最后结合趋势计算得到待求点处的空间插值结果。实验采用两组实际PM2.5浓度数据和降雨量数据,运用交叉验证方法对RRIDW方法与其他常用空间插值方法进行对比分析,验证了该方法的优越性和可行性。

Spatial interpolation methods,on one hand,can utilize the observation data to complete the missing data,and on the other hand,can also convert the discrete data into a continuous surface.However,the existing spatial interpolation methods lack an adequate consideration of the effect of spatial heterogeneity in spatial data.Therefore,the residual inverse-distance weighting spatial interpolation method based on spatial heterogeneity sub-region was presented in the paper.First of all,according to the sample point attribute value,the research area is divided into spatial heterogeneity sub-regions.And then calculate the trend surface to obtain the residuals of the sample points′attributes,which are used for inverse distance weighted interpolation in the next step.Finally,the spatial interpolation result is estimated by combining the trend surface values and the interpolated residuals.At the end of this paper,two sets of actual PM2.5concentrations and rainfall data are adopted to verify the effectiveness of the method proposed by comparing RRIDW with the commonly used spatial interpolation methods.