A combination method for multicriteria uncertainty analysis and parameter estimation: a case study of Chaohu Lake in Eastern China
A combination method for multicriteria uncertainty analysis and parameter estimation: a case study of Chaohu Lake in Eastern China
复制标题
多准则不确定性分析与参数估计的组合方法——以华东巢湖为例
DOI：
10.1007/s11356-020-08287-1
复制
发表时间：
2020-04-06
影响因子：
5.8
通讯作者：
Cheng, Jilin
中科院分区：
文献类型：
--
作者：
Wang, Yulin;Chen, Haomiao;Cheng, Jilin
富营养化模型非常重要，是制定政策和立法的宝贵工具。然而，参数的不确定性和大量的计算成本导致决策困难，特别是对于多指标的复杂模型。提出了一种多准则不确定性分析和参数估计（MUAPE）方法，该方法结合Pareto优势选择行为参数，同时利用最大似然概念和核密度估计获得建模所需的可接受值。将该方法应用于巢湖富营养化模型的不确定性分析和参数估计，该方法不需要设定阈值和权重。使用不同的标准集，相对误差（RE）和均方根误差（RMSE）的行为参数的结果进行了比较，结果表明，在边缘概率密度所代表的参数不确定性的影响方面，差异不大。与藻类动力学相关的参数的不确定性（即，BMR，PM和KESS）小于营养和温度相关参数（即，KDN、Nitm、KTB和KTHDR）。然而，当使用RE时，两个参数的联合不确定性的减少比使用RMSE时更大。通过RE准则，得到了巢湖富营养化模型关键参数的可接受值。结果与实测值吻合较好，参数可用于模型预测。这一结果表明，该组合方法不仅是实用的，以减少参数的不确定性，但也有助于确定参数值。该方法为富营养化模型中的多准则不确定性分析和参数估计提供了基础。
Eutrophication models are of great importance and are valuable tools for the development of policy and legislation. However, the parameter uncertainty and substantial computational cost lead to difficulties in decision-making, especially for complex models with multiple indicators. A multicriteria uncertainty analysis and parameter estimation (MUAPE) method, which selected behavioral parameters combined with Pareto domination and simultaneously obtained acceptable values for modeling by the maximum likelihood concept and kernel density estimation, was shown. This method, which did not assign thresholds and weights, was applied to analyze the uncertainty of the Chaohu Lake eutrophication model and estimate parameters. The results of the behavioral parameters were compared using different criterion sets, the relative error (RE) and the root mean square error (RMSE), and the results showed little discrepancy in terms of the effects on parameter uncertainty represented by the marginal probability density. The uncertainties of the parameters related to algal kinetics (i.e., BMR, PM, and KESS) were smaller than those of nutrient- and temperature-related parameters (i.e., KDN, Nitm, KTB, and KTHDR) for both sets of criteria. However, the reduction in the joint uncertainty of the two parameters was greater when RE was used than when RMSE was used. The acceptable values for the key parameters of the Chaohu Lake eutrophication model were also obtained by the RE criterion. The results strongly agreed with the observed values, and parameters could be applied for model prediction. This result indicated that the combination method was not only practical for reducing parameter uncertainty but also useful for determining parameter values. This method provides a basis for multicriteria uncertainty analysis and parameter estimation in eutrophication modeling.