克里雅绿洲浅层地下水与土壤特征的局部空间关系

通过优选的空间插值方法、网格单元法、地理加权回归(GWR)与最小二乘法(OLS)等方法的综合运用,系统研究典型干旱区绿洲的浅层地下水特征(埋深、电导率)与表层土壤特征(含水率、电导率)的空间分异与局部空间关系。结果表明:空间拟合方法上,表层土壤电导率与含水率的空间关系回归适用OLS模型,其余指标间空间关系采用GWR模型更优;空间关系上,土壤电导率与含水率的全局相关系数为极显著的0.85,局部相关性上绿洲西部明显高于东部;土壤电导率与地下水电导率的全局相关系数极显著(R=0.602),但局部相关性上同时具有正相关与负相关特征;土壤含水率与埋深、地下水电导率与埋深均呈全局负相关,但在局部出现正相关;地下水埋深由绿洲西北至东南逐渐增加,地下水电导率由西南至东北依次呈现低—高—低的特征,表层土壤电导率与含水率由西至东均呈现低—高—低—次高的特征; GWR较OLS方法能够反映更多的空间异质特征,通常在总体相关水平下,局部可能出现相关性相反或大小不同的相关性。

The localized spatial relation between shallow groundwater and soil properties in Keriya Oasis

It is of great significance to reveal the spatial distribution of shallow groundwater and surface soil properties, especially, the spatial relationship in local scale between them. Through the integrated methods of the interpolation methods, the grid cells method, the geographical weighted regression (GWR), and ordinary least squares (OLS), we studied the spatial distribution and relation of shallow groundwater characteristics (level and conductivity) and surface soil properties (water content and conductivity) in a typical arid oasis (Keriya Oasis). The results showed that interpolation prediction method using RBF on groundwater level and soil conductivity had the best fit compared to other methods. Meanwhile, using IDW on groundwater conductivity and using Ordinary Kriging method on surface soil moisture were more suitable. The OLS model was used to find the spatial relationship between conductivity and water content of surface soil, and GWR model was used for finding the spatial relationship between other indexes. Compared with the overall correlation (0.85) between soil conductivity and water content, the spatial relationship in the local scale showed that the correlations were higher in western than eastern regions. With the overall correlation (0.602) between soil conductivity and groundwater conductivity, the spatial relationship in the local scale appeared inconsistent results both positive and negative correlations. Compared with the overall negative correlation between water content of soil and groundwater level, and between groundwater conductivity and level, the spatial relationship in the local scale appeared positive correlation in some areas. The groundwater level increased from the northwestern to the southeastern, and groundwater conductivity showed low-high-low from the southwestern to the northeastern. Meanwhile, the soil conductivity and water content illustrated low-high-low-high trend in the area. GWR model could reflect more spatial heterogeneity than the OLS model, especially, might be opposite local relations or different level of relationship compared with the overall relationship.