空间插值模型对土壤Cd污染评价结果的影响  

Impact of spatial interpolation methods on the estimation of regional soil Cd

作　　者：谢云峰[1,2] 陈同斌[1] 雷梅[1] 郑国砥[1] 宋波[1] 李晓燕[1] 

机构地区：[1]中国科学院地理科学与资源研究所环境修复中心,北京100101 [2]中国科学院研究生院,北京100049

出　　处：《环境科学学报》2010年第4期847-854,共8页Acta Scientiae Circumstantiae

基　　金：中国科学院知识创新工程重大项目(No. KZCX1-YW-06-03);资源与环境信息系统国家重点实验室开放基金;"十一五"国家科技支撑计划课题(No. 2007BAC28B04)~~

摘　　要：分析和比较了不同插值模型(反距离加权、局部多项式、普通克里格和径向基函数)对土壤Cd空间插值、污染面积估算和污染区空间分布的影响,结果表明,插值方法对极大值都存在不同程度的平滑,交叉验证的Cd极大值插值的相对误差高于45.1%.不同插值模型估算的污染区域面积在0～2.12%之间.根据样点超标率估算的污染面积比插值模型估算的面积高1.53%～3.65%.污染评价结果不确定性较大的区域主要是在局部极大值区域和高值向低值的过渡区域.因此,土壤重金属的空间插值精度对污染评价结果有很大影响,开展土壤重金属污染调查时应注意在浓度过渡区域加大采样密度.The distribution of Cd in surface soil of the Tongzhou district,Beijing（605 km^2）was interpolated with 4 spatial interpolation methods：Inverse Distance Weighted,Local Polynomial,Ordinary Kriging and Radial Basis Function. The accuracy of every method was compared according to the extreme value prediction,the impact on area estimation and the spatial distribution of high Cd area. The cross-validation results showed that all of the methods underestimated the extreme value of Cd,and the mean relative error （MRE） was higher than 45.1%. The cadmium-pollution areas calculated by different methods are from 0 to 2.12%,which were from 1.53% to 3.65% less than those estimated by sample statistics. The differences （or uncertainties） of the interpolation results among the models were mainly located in the region of extreme high concentration and the transitional region from high Cd concentration to low Cd concentration. Therefore,a proper interpolation model has a significant impact on the results of the spatial distribution estimation of soil heavy metals,and that of soil pollution evaluation. While surveying the soil heavy pollution,density sampling in the soil heavy metal concentration transitional region is advised.

关 键 词：CD 空间插值 污染评价 极大值 面积估算 区域土壤环境质量 

分 类 号：X144[环境科学与工程—环境科学]