Multi-scale spatial autocorrelation analysis of cultivated land quality in Zhejiang province

Abstract: Understanding the spatial characteristics of cultivated land quality at different spatial scales is the base to formulate appropriate management countermeasures and targeted cultivated land construction. The purpose of this paper was to explore the spatial characteristics of cultivated land quality at the provincial, municipal and county-level scale. Zhejiang province was taken as an example. A combination of spatial autocorrelation models represented by Moran's I index and coefficient of variation method was used for the study. In 2012, data of cultivated land information in 2011 in Zhejiang province were collected from Yellowbook and investigation from local government. Analysis of global spatial autocorrelation of cultivated land quality showed that the Moran's I value was higher than 0.5, showing strong positive autocorrelation. The Moran's I value of economical land index was the highest, followed by the land use index and land natural index. For 11 cities in Zhejiang, the Moran's I value of land natural index was higher in Hanzhou, Shaoxing, Ningbo, Huzhou and Jiaxing, that of land use index was higher in Hanzhou, Shaoxing, Ningbo and Huzhou, and that of economical land index in Hanzhou was higher than the provincial level. Furthermore, among 85 counties, the Moran's I value of global spatial autocorrelation was the highest in Xiaoshan (0.87) and the lowest in Gongshu (?0.16) for the natural index, the highest in Xiuzhou (0.87) and the lowest in Gongshu (?0.22) for the land use index, and the highest in Xiuzhou (0.88) and the lowest in Gongshu (?0.13) for the economical land index, respectively. The local spatial autocorrelation analysis showed that the high-high area of natural index was clustered in the north of Zhejiang including Xiaxing, Huzhou, Hanzhou et al., the low-low area was in counties such as Dongyang, Yongkang, Wuyi, et al., and the low-high and high-low areas were mainly distributed in the center of Zhejiang such as Yiwu. The high-high area of land use index was concentrated in Xiaxing, Shaoxing, Hanzhou, et al., the low-low area of that was in Fuyang, Lin'an, et al., and low-high and high-low areas were dispersedly distributed. The high-high area of economical index was even scattered mainly in parts of Huzhou, Jiaxing and Shaoxing and so on. The high Moran's I values of the 3 indexes appeared mainly in the north of Zhejiang, sparsely along the coast and the low values were shown in the northwest and center of Zhejiang. In sum, the spatial distribution of cultivated land quality was sensitive to spatial scale. The larger scale indicated high autocorrelation and agglomeration. Among the 3 indexes, the natural index was greatly affected by the scale change, followed by land use index and economical index. Therefore, these characteristics should be taken into account in the cultivated land management and partition.