Bayesian uncertainty assessment in multicompartment deterministic simulation models for environmental risk assessment
Bayesian uncertainty assessment in multicompartment deterministic simulation models for environmental risk assessment
复制标题
DOI：
10.1002/env.590
复制
发表时间：
2003-06-01
期刊：
影响因子：
1.7
通讯作者：
Raftery, AE
中科院分区：
文献类型：
--
作者：
Bates, SC;Cullen, A;Raftery, AE
我们使用贝叶斯融合的一个特殊情况下，从确定性模型的推断，同时考虑到模型输入的不确定性。该方法使用所有可用的信息，基于数据和专家知识，并通过使用可用数据更新模型来扩展当前的“不确定性分析”方法。我们扩展的方法与序贯多室模型使用。我们介绍了这些方法在土壤和蔬菜中多氯联苯（PCB）浓度确定性模型中的应用。结果是土壤和蔬菜中浓度的后验分布，它解释了所有可用的证据和不确定性。不考虑模型的不确定性。版权所有（C）2003约翰威利父子有限公司。
We use a special case of Bayesian melding to make inference from deterministic models while accounting for uncertainty in the inputs to the model. The method uses all available information, based on both data and expert knowledge, and extends current methods of 'uncertainty analysis' by updating models using available data. We extend the methodology for use with sequential multicompartment models. We present an application of these methods to deterministic models for concentration of polychlorinated biphenyl (PCB) in soil and vegetables. The results are posterior distributions of concentration in soil and vegetables which account for all available evidence and uncertainty. Model uncertainty is not considered. Copyright (C) 2003 John Wiley Sons, Ltd.