Multi-scale spatial autocorrelation analysis of cultivated land quality in China's southern hillside areas: A case study of Lichuan County, Jiangxi Province

Cultivated lands are among the most fundamental resources for national food security and sustainable social development. Scientific analysis of the spatial distribution patterns and evolutionary characteristics of cultivated land quality is significant for the protection and layout optimization of cultivated lands. Analysis of the spatial distribution of cultivated land quality at different scales is the basis of cultivated land quality improvement and strengthening of cultivated land protection. This paper was based on cultivated land quality monitoring experimental unit in Lichuan County in Jiangxi Province and then used weighted average, variation coefficient, and spatial autocorrelation to analyze spatial disparity characteristics of cultivated land quality. Multi-scale spatial autocorrelation analysis of cultivated land quality research is a hot research area. The innovation in this paper was the introduction of natural land grade index, use of the grade index and economic grade index as the space variable to separately explore and discuss the degree of spatial correlation and spatial disparity of cultivated land quality at county-scale, township-scale and village-scale in GIS environment. The results of the research showed that:1) Cultivated land quality index was high in the south and north and low in the west and east in Lichuan County. When the threshold distance was 400 m, there was a significant spatial autocorrelation in cultivated land quality. The Moran's I value of natural land grade index was highest, followed by economical land grade index, and land use grade index was the lowest. The Moran's I value of cultivated land quality from county to township and then to village scales decreased systematically. Multi-scale spatial autocorrelation analysis of cultivated land quality exhibited a significant aggregation of spatial distribution in Lichuan County. 2) It was found that different types of cultivated land quality indexes had remarkably different spatial correlations at different spatial scales. For the influence degree of spatial scale, land use grade index was greatly affected by spatial scale, followed by natural land index, and the economic land grade index was the weakest. For the coefficient of variation of cultivated land quality index, fluctuation in Moran's I value for cultivated land at village-scale was far greater than that at township scale. While at the same spatial scale, the coefficient of variation of natural land index was strongest, followed by economic land index and then land use index. 3) The results based on local indicators of spatial association (LISA) showed that positive spatial autocorrelation of cultivated land quality, included the high-high type and the low-low type, emerged as the shape of the cluster and in the form of group, while the negative spatial autocorrelation contained the high-low type and low-high type was fragmented in space. The results of the study showed that spatial disparity in cultivated land quality was sensitive to spatial scale. Therefore, for cultivated land quality improvement and protection, there was the need to pay more attention to spatial disparity of cultivated land quality at town scale and village scale. Also based on the difference in spatial correlation degree among the natural conditions, utilization conditions and economic benefits, it was possible to explore cultivated land quality improvement and protection measures that met actual ground situations in the study area. The results of the study provided the needed references for the construction of high-standard basic farmlands, land reclamation, regional cultivated land quality monitoring, cultivated land protection, partitioning and management of cultivated lands, cultivated land quality improvement and spatial optimization of cultivated land quality.