Understanding the spatial distribution of total copper, available copper, and the spatial non-stationary relationships between available copper and relevant environmental factors is important for the delineation of soil risk areas and the development of related control measures. This study was conducted in Zhangjiagang County of Jiangsu Province, China. The risk status for soil copper was assessed based on 357 soil samples in the study area. The effects of soil type and land-use type on the concentration of available soil copper were discussed first. Then, ordinary kriging was adopted to map the spatial distribution patterns of the total soil copper and available soil copper, and the spatial distribution map of the copper availability ratio (i.e., available copper/total copper) was also developed for the study area. The risk areas for soil copper were delineated based on the spatial distribution patterns of available soil copper and the copper availability ratio. Finally, a new spatial local regression technique, geographic weighted regression (GWR), was used to explore the local spatial regression relationships between available copper and its three main impact factors (i.e., total soil copper, soil pH, and SOM). Results showed that both soil type and land-use type had some effect on the concentration of available soil copper. The copper availability ratio had a strong spatial heterogeneity, with the higher values mainly in the northeast, southeast, and northwest of the study area and the lower values mainly in the middle and southwest of the study area. The range of the copper availability ratio is 13.56% to 29.15%. The results of the comparison of the traditional ordinary least squares regression (OLSR) and GWR showed that the GWR model had higher fitting accuracy than the OLSR model[i.e., a larger decision coefficient R2, and smaller corrected Akaike information criteria (AICc) and the sum of squares of residuals] in modeling the relationships between available copper and its three main impact factors. The GWR analysis showed that the effect of soil factors on the concentration of soil available copper was non-stationary. The GWR could effectively reveal the spatial non-stationary influence of the related soil factors on the concentration of available soil copper, and the results could explain the reasons for the accumulation of available soil copper in local areas. Potential risk areas for available soil copper were delineated based on the copper availability ratio and the concentration of available soil copper in the study area. The results should be crucial data for developing specific control measures for soil copper at a regional scale.
Keywords: available copper; geographic weighted regression; regulation; soil; spatial nonstationary.