Study on Construction Land Distr ibution in Fujian and Taiwan Provinces Based on Spatial Autocor r elation Analysis
基于空间自相关的闽台城镇建设 用地分布研究

Spatial autocorrelation is an important kind of spatial statistics, which is used to disclose the spatial structure of regional disparity. Spatial autocorrelation means self-correlation or spatial dependence among observations of a geo-referenced attribute. There are two different scales for spatial dependence: global indicators and local indicators. Based on the image data including Fujian and Taiwan provinces cutting from ASTER images in 2002 after geometric correction, the construction land patches were distilled, and the question of construction land density of regional disparity in Fujian and Taiwan was reconsidered by using spatial analysis methods. According to the analytical results, there is a significant positive spatial autocorrelation of construction land density in Fujian and Taiwan. That is to say, the relatively high (low) developed area tends to be located more often nearby other high (low) developed areas than expected by random chance, and then each area should not be viewed as an independent observation. We also investigate the spatial association between core areas and adjacent areas by computing the Local Moran I at county level. With the use of Moran's scatter plots and LISA (Local Indicators of Spatial Association) cluster maps, we can identify some different types of significant local spatial association based on the analysis of different areas in Fujian and Taiwan. As a result, the research provides an important cognition to construction land distribution in Fujian and Taiwan, and insights into the types of spatial association present in an economic region allow for more effective implementation of economic development policies in Fujian.