基于主成分分析的协同克里格插值模型对土壤铜含量的空间分布预测

以湖北省宜都市红花套镇的柑橘产区为例,选取土壤重金属全铜含量为研究对象,利用Pearson相关系数方法选择与土壤铜含量具有较高相关性的土壤因子(全K、全Cr、CEC、全Al、全N(P0.05))作为辅助变量,利用主成分分析(PCA)对辅助变量进行降维求总得分处理后,与协同克里格插值相结合构建土壤铜含量空间模型(COKPCA),同时构建土壤铜含量的普通克里格插值模型(OK)和以全K作为辅助变量构建协同克里格插值模型(COKK),对不同的空间模型进行模拟精度的对比和分析之后提出最优空间模型方法,进一步分析土壤铜含量在研究区域的空间分布特征。结果表明:普通克里格插值(OK)、协同克里格插值(COKK)和与主成分相结合的协同克里格插值(COKPCA)3种模型的RMSE分别为7.044、6.907和6.309,R2分别为0.716、0.743和0.852,赤池信息量准则(AIC)分别为101.591、96.908和87.203。综合比较,COKPCA具有最优的模拟插值结果,其次为COKK,而OK则相对较差。

Spatial distribution of soil heavy metal Cu content using Cokriging interpolation model combined with principal component analysis

Insufficient or excess heavy metals in soils seriously affect the healthy growth of citrus.It is necessary to establish an accurate space model which can effectively reflect the spatial distribution of heavy metals content.This paper took the citrus production areas in Honghuatao county,Yidu City,Hu-bei Province were used to study heavy metal Cu in soil.Soil factors including total K,total Cr,CEC,total Pb and total N (P <0.05)closely correlated with Cu were chosen as the spare variables through Pear-son’s correlation coefficient method.Total scores were calculated by principle components analysis (PCA)method and used as auxiliary variables to reduce the dimension and redundancy of original varia-bles.Three spatial interpolation models including ordinary Kriging (OK )model,ordinary Cokriging (COK)models which took total scores (COKPCA )and total K (COKK )as auxiliary variable were con-structed.The optimal space model was proposed by comparing the prediction accuracy of the models (OK,COKK ,COKPCA ).The most suitable model was used to predict spatial distribution of Cu in the area studied.The results showed that RMSE of OK,COKK and COKPCA models were 7.044,6.907 and 6.309, with R 2 of 0.71 6,0.743 and 0.852,and AIC of 101.591,96.908 and 87.203,respectively.Interpolation re-sults were comprehensively compared.COKPCA was found to be the best simulating model,followed by COKK but OK was poor.It will provide reference for accurately simulating soil heavy metals Cu in the small-scale area and a scientific basis for modern agricultural production and fine management of agricul-ture.