• 关键词：补充采样  克里格  方差  置信度  不确定性

• 基金项目：国家自然科学基金项目（No.41271339）；北京市科技计划项目（No.D16110900470000）；中国科学院重点部署项目（No.ZDRW-ZS-2016-5-5）；国家重点研发计划（No.2016YFD0800906-1）

• 陶欢
• 1. 北京农业信息技术研究中心, 北京 100097;2. 中国科学院地理科学与资源研究所, 北京 100101

• 廖晓勇
• 中国科学院地理科学与资源研究所, 北京 100101

• 阎秀兰
• 中国科学院地理科学与资源研究所, 北京 100101

• 赵丹
• 中国科学院地理科学与资源研究所, 北京 100101

• 李尤
• 中国科学院地理科学与资源研究所, 北京 100101

• 杨坤
• 中国科学院地理科学与资源研究所, 北京 100101

• 李存军
• 北京农业信息技术研究中心, 北京 100097

• 摘要：污染场地中的污染区域较小，但污染物的空间变异性大，污染土体较深，因而在场地调查中通常需采集场地钻井数据进行三维模拟.为了准确估计场地土壤污染物的空间分布和需修复的污染土方量，提出了场地污染调查补充采样钻井点位布设的优化方法.在初始采样的基础上，首先利用克里格方法预测场地中土壤污染物的克里格方差，基于克里格方差估算置信度，再结合场地置信度和场地污染物含量的空间分布确定场地的最大不确定性点，并将该点作为下一阶段补充采样点，以实现逼近式动态追补钻井点位的布设.最后，以某硝基苯污染的工业场地为例对该方法进行验证.结果表明：场地中硝基苯的空间变异性大，增加采样点会显著减少基台值，提高研究区总体的置信度，同时减少总体的不确定性.在项目经费约束条件下，不确定性减少65%时，需要至少增加采样点10个.10个样点的增加使得总体估值方差降低了40%，同时总体的估计精度提高了22.4%.案例结果验证了该方法在单一类有机污染场地中的可行性和适用性，能够在保证场地污染调查精度的同时，显著降低样本量，实现场地调查的成本-效益最大化.

• Abstract：Given great spatial variability of contaminant and depth of contamination soil in the relatively small contamination area, a three-dimensional modelling with drilling data is employed in contaminated site investigation. In order to accurately estimate the spatial distribution of soil pollution and the volume of contaminant remediation, this study presented an optimization strategy for drilling design in contaminated site investigation. The variance of soil contaminant and the confidence coefficient were predicted using Kriging interpolation. The drilling sample with maximum uncertainty was then determined based on the spatial distribution of confidence coefficient and the detected contamination concentration. A dynamically adding samples strategy for drilling design was then developed. Subsequently, this method was applied and verified at a nitrobenzene contaminated site. Results showed that nitrobenzene in the site had great spatial variability. Increasing number of drilling samples significantly decreased sill values, increased population confidence coefficient within the site, and reduced the population uncertainty. Under the constraint of project expenditure, at least 10 additional samples were needed to reduce the uncertainty by 65% and the population variance by 40% while increase the population estimation accuracy by 22.4%. This case study proved the feasibility and adaptability of the proposed strategy in a site with a single type of contaminant. This strategy could dramatically reduce the sampling population within the contaminated site while ensuring investigation accuracy and maximizing the cost-benefit in site investigation.

• Key words：additional sampling  kriging  variance  confidence coefficient  uncertainty

• 
  摘要点击次数： 1279 全文下载次数： 1389