基于特征代表性的土壤环境质量监测点布局优化方法

为提升土壤监测点代表性以更准确获取土壤信息并实施有效管理措施,本文提出一种基于特征代表性的土壤环境质量监测点布局优化方法。该方法基于多个与监测目标变量相关性强的辅助变量的属性分布构建特征空间,采用条件拉丁超立方体方法将特征空间分层并进行编码,并通过计算特征空间分层影响度以确定抽样顺序,逐点抽样优化获得高代表性的布样方案。本研究以北京顺义区为例,以土壤类型、土壤质地、土地利用类型和坡度作为辅助变量进行监测点布局优化,并与简单随机采样方法、空间分层采样方法、条件拉丁超立方体采样方法(cLHS)进行比较。结果显示:相较于其他3种方法,本文提出的方法的特征空间覆盖率平均提高15%左右,耗时远小于cLHS,略高于简单随机采样和空间分层采样,所获取的监测点布设方案的不确定性明显低于其他3种方法,重金属含量分布特征与总体数据更为接近。综上,本研究提出的方法能够显著提升监测点在特征空间的代表性,可有效反映调查区域土壤属性总体分布特征,能为后续调查监测土壤信息提供参考手段。

Sampling optimization method for soil environmental quality monitoring based on feature representativeness

For improving the representativeness of monitoring sites to accurately obtain soil information and effectively implement management measures, this paper proposed a sampling optimization method for soil environmental quality monitoring based on feature representativeness. In this method, the feature space based on soil auxiliary variables was first built, and a sample sequence was established with the stratified impact of the feature space, and then a highly representative sampling scheme was obtained by the point-by-point sampling pattern. An example which optimizes monitoring sites using soil type, soil texture, land use type, and slope as auxiliary variables was taken in this study, and the results of the proposed method, simple random sampling, spatial stratified sampling, and conditional Latin hypercube sampling methods(cLHS) was compared. It showed that the suggested method improves the feature space representativeness by approximately 15% on average and takes much less time than the cLHS but slightly higher than the simple random sampling method and spatial stratified sampling method. The sampling distribution features of heavy metal concentration are more in line with the total, and the sampling distribution has substantially less uncertainty than that with the other three methods. The proposed method can significantly improve the representativity of monitoring sites in the feature space, which can effectively reflect the overall distribution characteristics of the survey area soil attributes, and it provides a reference means for the effective investigation and monitoring of soil information.