基于三明治抽样的市域耕地土壤重金属监测点位r优化布设方法

摘要

如何平衡成本和精度之间的矛盾是区域土壤环境监测网构建亟待解决的技术难题.以广东省江门市耕地土壤砷含量为研究对象,开展基于三明治抽样的土壤环境监测点位优化研究.按照1 km×1 km网格在研究区域均匀布设3375个点位,采用三明治抽样方法对其点位进行抽样优化,并运用Cochran公式、空间插值表征、交叉验证、独立验证等验证方法对不同期望精度下的抽样样本进行验证.结果表明,(1)期望精度为5％、10％、15％、20％的三明治抽样样本量分别为278、147、102、80个.(2)抽样结果的4种验证方法表明,期望精度为5％和10％的抽样样本能满足监测精度要求.该精度下样本量均满足利用Cochran经典公式在相应置信水平和相对误差条件下获取的合理采样数量,并且相应的半变异函数模型拟合度好、均方根误差值(RMSE)小、克里金插值代表性强.(3)期望精度为10％的抽样样本在保证监测精度不变的同时,大大降低了监测成本,其监测点位从3375个降至147个,点位数量降低了96％.总之,抽样样本的期望精度值越小,监测精度越高.在江门市市域尺度上,期望精度为10％的三明治抽样是一种高精度、低成本、代表性强的土壤环境监测点位优化方法.%It is urgent to balance the contradiction between cost and accuracy during regional soil environmental monitoring network. In this study, As content of cultivated soil in Jiangmen city in Guangdong Province was chosen to explore the optimization of soil environmental monitoring sites based on sandwich sampling method. The grid scale of 1 km×1 km was used with uniformly distributed 3375 sampling sites in the survey region. The monitoring sites were optimized using sandwich sampling method. The sampling samples under different expected accuracies were evaluated by Cochran formula, spatial interpolation, cross validation, and independent verification. The results showed that, (1) The sandwich sampling numbers of expected accuracies 5％, 10％, 15％, and 20％ were 278, 147, 102, and 80, respectively. (2) The four verification methods were used to evaluate accuracy of sampling results. It was indicated that the sampling samples under expected accuracies 5％ and 10％ achieved monitoring precision. The sample numbers of the two sampling samples met the reasonable sampling numbers on relevant confidence levels and allowable errors by Cochran formula. The good fitting degrees of semivariogram models and small root mean square errors (RMSE) of the two sampling samples were achieved. And the ordinary Kriging interpolation results had a strong representative. (3) The sandwich sampling method under expected accuracy 10％ could not only achieve the same monitoring precision, but also obviously save monitoring cost. The sampling numbers were decreased to 147 from 3375, and 96％ monitoring points were reduced. In conclusion, the smaller expected accuracy of sampling sample, the higher monitoring precision. And the site optimization method for soil environmental monitoring based on sandwich sampling method under expected accuracy 10％ has high precision, low cost, and strong representative on the scale of Jiangmen city.