Back,P.-E.(2007): A model for estimating the value of sampling programs and the optimal number of samples for contaminated soil. Environ. Geol., 52, 573-585.

『汚染土壌の試料採取計画の価値と最適試料数を見積るモデル』


Abstract
 A model is presented for estimating the value of information of sampling programs for contaminated soil. The purpose is to calculate the optimal number of samples when the objective is to estimate the mean concentration. A Bayesian risk-cost-benefit decision analysis framework is applied and the approach is design-based. The model explicitly includes sample uncertainty at a complexity level that can be applied to practical contaminated land problems with limited amount of data. Prior information about the contamination level is modelled by probability density functions. The value of information is expressed in monetary terms. The most cost-effective sampling program is the one with the highest expected net value. The model was applied to a contaminated scrap yard in Goteborg(最初のoの頭に¨), Sweden, contaminated by metals. The optimal number of samples was determined to be in the range of 16-18 for a remediation unit of 100 m2. Sensitivity analysis indicates that the perspective of the decision-maker is important, and that the cost of failure and the future land use are the most important factors to consider. The model can also be applied for other sampling problems, for example, sampling and testing of wastes to meet landfill waste acceptance procedures.

Keywords: Contamination; Value of information; Data worth; Soil; Cost-effectiveness; Bayesian analysis』

Introduction
Methodology
 Decision analysis framework
 Procedure of VOIA
 Step 1: sampling program
 Step 2: prior information
 Step 3: probability estimation
  Prior probabilities
  Including sample uncertainty
  Weighing sample uncertainty with prior information
  Preposterior probabilities
 Step 4: cost estimation
 Step 5: estimation of the value of information
Application
 Wockatz scrap yard
 Sampling approach
 Prior information
 Probability estimation
 Cost estimation
Results
 Single-factor sensitivity analysis by graphs
 Sensitivity analysis by Monte Carlo simulation
Conclusions and discussion
Acknowledgments
References


戻る