Expert Systems for Parameter Estimation in Pollutant Transport-and-Fate Modeling

这项提案概述了一项雄心勃勃的努力，即在一个具有国家后果的实际问题的背景下，研究最先进的建模技术。毒理学、卫生政策、环境管理和相关领域的风险管理决策往往由复杂的数学模型支持。这些模型旨在预测每种可供选择的行动方案的后果。模型预测取决于潜在用户指定的值。然而，这些值往往不能确切地知道。如果不能反映参数说明中的不确定性，这些模型可能会导致不适当的风险管理决策。研究人员建议应用一套分层贝叶斯方法来表征预测地表水中合成有机化学品浓度的模型中的不确定性。最终目标是创建一个专家系统，以帮助用户选择风险管理模型中的参数值。这类问题很重要，但只有几个调查人员研究过。研究团队训练有素，精通相关分析工具。有充分的理由相信他们会成功地完成他们提出的项目。

This proposal outlines an ambitious effort to investigate state- of-the-art modeling technology in the context of a practical problem with national consequences. Risk management decisions in toxicology, health policy, environmental management, and related fields are often supported by complex mathematical models. These models are intended to predict the consequences of each alternative course of action. The model predictions depend on values specified by the potential users. However, these values are often not known with certainty. In failing to reflect uncertainty in parameter specification, the models may lead to inappropriate risk management decisions. The investigators propose to apply a set of hierarchical Bayesian methods to characterize uncertainty in models which predict the concentration of synthetic organic chemicals in surface waters. The ultimate goal is the creation of an expert system to assist users in selecting values of parameters in risk management models. The general class of problems is an important one, but one that has only been worked on by a few investigators. The research team is well trained and well versed in the relevant analytic tools. There is every reason to believe they will complete the project they propose successfully.