Approach For Exploring Geospatial Subregions With Consistent Homogeneity And Its Application

Spatial heterogeneity is ubiquitous in geographical objects.Spatial heterogeneity and spatial dependence are critical intrinsic characteristics of geographic phenomena and spatial processes.Therefore,in geographical analysis,it is inevitable to face the spurious regression and analysis caused by spatial heterogeneity.To present,majority of the investigation on spatial heterogeneity focuses on the detection method and description of spatial heterogeneity.However,these investigations are so weak that the complete detection theoretical system of spatial heterogeneity has been not yet formed.The aim of this paper is attempt to provide the detection method of spatial heterogeneity.Based on the idea of geospatial spatial scale,the region with spatial heterogeneity is considered to be combined with several sub-region with spatial homogeneity in view point of different geographic spatial scales.The region with spatial heterogeneity is partitioned into several sub-regions with spatial homogeneity so as to detect spatial heterogeneity.The geospatial homogeneous region can be regarded as spatially stationary region from perspective of different geographical spatial scales.Therefore,it is very important theoretical significance and practical valuable in spatial analysis how to spatially partition heterogeneous regions into homogeneous sub-regions.In this paper,the methodology of theoretical analysis is applied to explore the theory and the way of geospatial homogeneous regions and the application case of the real data is utilized to verify the proposed approach is reliable.Firstly,the relationship between spatial heterogeneity and spatial homogeneity is revealed.And then,the approaches and key techniques are proposed to realizing that geospatial homogeneous region is partitioned into spatial homogeneous sub-regions.Next,the proposed method is adopted in the study of NDVI in Eastern China and tries to explore geospatial homogeneous regions in which correlation among NDVI,precipitation and air temperature is consistent.Finally,discussion in detail is made and comparative results with other researchers are presented.The main research topics are as follows:(1)The relationship between spatial heterogeneous region and spatial homogeneous region was disclosed.In base of the basic concepts of spatial heterogeneity,a preliminary classification of spatial heterogeneity is made,and the relationship between spatial heterogeneity and geospatial scale is analyzed,and the relationship between spatial heterogeneity and spatial homogeneity is explored.(2)Approach for exploring geospatial sub-regions with consistent homogeneity was proposed.Based on geographic spatial scales,the idea of exploring geospatial subregions with consistent homogeneity was proposed.The relationship between the proposed method and the Fisher optimal dissection method for one-dimensional series was uncovered and the Fisher optimal dissection method was extended and utilized to find the spatial homogeneous region from spatial heterogeneous region.The sum over residual squares of linear fitting substitutes the similarity and distance measure in classical K-means clustering algorithm,and the approach of intra-stratal dichotomy was proposed.In the end,the method of find the spatial homogeneous region from spatial heterogeneous region is put forward,and the approach realization is given.(3)The example verification of the partition approach of geospatial regions with spatial heterogeneity is taken.The data in this experiment include monthly mean temperature,monthly precipitation,DEM data and NDVI data in Eastern China in 2017.The air temperature,precipitation and DEM as the independent variables and NDVI as the dependent variable,the spatial regions with the synchronous period and the lagged period were partitioned,respectively,and the discussion of our results are made in detail,and comparison with other researchers are conducted carefully.The investigation results show that our proposed method is verified to divide spatially heterogeneous regions into several homogeneous sub-regions.Our partitioning results are basically consistent with existing studies,and are more subtle.