基于土壤类型和微量元素辅助信息的土壤属性空间模拟

为探讨大尺度区域土壤属性空间化的方法，以吉林省为例，研究土壤养分（pH值、有机质、速效磷、速效钾和碱解氮）和地形地貌、微量元素等变量之间的关系；并在考虑土壤类型基础上，将相关性较高的变量作为协因子进行土壤养分的Cokriging插值研究。结果表明，pH值与经度、有效铁、锰和速效氮的相关系数分别高达-0.66、-0.71、-0.70和-0.67；有机质与经度、pH值、有效钙、锰的相关系数分别为0.55、-0.58、0.56和0.52；碱解氮与经度、纬度、pH值、有效铁、锌的相关系数分别为0.57、-0.57、-0.67、0.56、0.54；速效磷与速效钾、有效锌的相关系数分别为0.67和0.64。分析发现以相关性较高的微量元素作为协因子进行Cokriging插值精度均优于采用地形变量作为协因子的Cokriging插值。交叉检验和检验站验证结果表明，与普通Kriging相比，基于土壤类型和微量元素的Cokriging插值在增加估值精度方面有所提高。

Spatial interpolation of soil properties based on soil types and trace micro-elements

To develop the methodology for spatialization of large scale soil property, we firstly investigated the relationship between soil properties and environmental factors in Jilin province located in Northeast China. Secondly, with a consideration of soil types, the relevant factors were utilized as co-factors for interpolating soil properties by using the means of Cokriging technique. The results showed that soil pH value, soil organic matter and alkali-hydrolyzable N were generally correlated with latitude and longitude. In details, soil pH value had an obviously negative correlation with longitude (-0.66), available iron (-0.71), manganese (-0.70) and active nitrogen (-0.67); Negative correlation could be found between soil organic matter and soil pH value (-0.58), while positive between soil organic matter and longitude (0.55), available calcium (0.560) and manganese (0.52); in addition, alkali-hydrolyzable N was negatively correlated with soil pH (-0.67) and latitude (-0.57), and positively correlated with longitude (0.57), available iron (0.56) and zinc (0.54); and active phosphorus had an obviously positive correlation with active potassium (0.67) and zinc (0.64). By contrast, the accuracy of Cokriging interpolations used with more relevant co-factors (i.e. trace elements) was obviously higher than the interpolations with other factors (i.e. topographic factors). In addition, the verification results of cross-validation and testing station also prove that accuracy of Cokrigng interpolations is higher than ordinary Kriging.