不同采样尺度下土壤图和Kriging法的空间估值精度比较--以砂姜黑土典型地区的研究为例

The Comparison of Soil Map and Kriging Methods for Spatially Prediction Precision of Soil properties with Different Sample Spacings——A Case of Shajiang Black Soil Area

在河南省驻马店市西平县砂姜黑土(砂姜潮湿雏形土)区,采用0.5 km×0.5 km网格法采集248个土壤耕作层样本,其中124个作为估值数据集,124个作为验证数据集,设计了3种不同采样尺度的估值数据集抽样方案,对比了土壤图法和Kriging法对土壤有机质、全氮、全磷、缓效钾和速效钾的估值效果。结果表明土壤图法对土种内变异系数较小,但土种之间差异较大的全磷、全氮和有机质的估值精度仅稍低于Kriging方法,在空间变异格局表达方面的功能也大致相同。对土种内变异系数较大,但空间变异结构性明显的速效钾,Kriging估值效果显著优于土壤图法。降低采样密度使Kriging估值平滑效应增加,估值精度降低,且集中出现在那些内部变异系数较大的土种内。该结论揭示在土壤调查的初级阶段采集少量的样点对土种内的土壤变量的变异程度做出估计对进一步的采样方案设计和土壤估值方法选择有指导意义。

In Shajiang black soil（Sha-Aquic Cambosols） area of Xiping county,Zhumadian district of Henan Province,248 topsoil samples were collected according to 0.5 km ×0.5 km grid.Samples were evenly divided into two subgroups,one prediction group to build predict models and the other verification group to verify the predicted values.Three scenarios with different amount of samples abstracted from prediction group were used to conducting the prediction respectively by soil map and Kriging method for soil organic mater,total nitrogen,total phosphorus,slowly available potassium and available potassium.The results indicated that the prediction exactness of soil map was comparable to that of Kriging method for total phosphorus,total nitrogen,soil organic mater,those all were low variable within same soil species while were remark different among soil species.Kriging method was remarkably superior to soil map method for the prediction of soil property which showed a structural variation in space though high coefficient of variable within soil species.Smoothing effect of Kriging preiction was increased because of the decline in the density of samples,which induced the decrease of prediction precision.However those sample sites where prediction precision decreased often focused on the soil species with high coefficient of variation.It was concluded that it was useful to gauge the coefficient of variation of soil properties within same soil species in advance for further planning detailed soil investigation and selecting prediction method.