研究报告
苏晗,邹锐,蒋青松,叶瑞,梁中耀,马文静,陈岩,刘永.流域负荷削减的主观深度不确定性分析及稳健决策[J].环境科学学报,2017,37(7):2777-2785
流域负荷削减的主观深度不确定性分析及稳健决策
- Subjective deep uncertainty and robust decision making for load reduction at the watershed scale
- 基金项目：国家水体污染控制与治理科技重大专项（No.2013ZX07102-006）
- 作者
- 单位
- 苏晗
- 北京大学环境科学与工程学院, 水沙科学教育部重点实验室, 北京 100871
- 邹锐
- 1 Tetra Tech, Inc. 10306 Eaton Place, Ste 340, Fairfax, VA 22030;2 云南省高原湖泊流域污染过程与管理重点实验室, 昆明 650034
- 蒋青松
- 北京大学环境科学与工程学院, 水沙科学教育部重点实验室, 北京 100871
- 叶瑞
- 南京智水环境科技有限公司, 南京 210012
- 梁中耀
- 北京大学环境科学与工程学院, 水沙科学教育部重点实验室, 北京 100871
- 马文静
- 南京智水环境科技有限公司, 南京 210012
- 陈岩
- 环境保护部环境规划院, 北京 100012
- 刘永
- 北京大学环境科学与工程学院, 水沙科学教育部重点实验室, 北京 100871
- 摘要：深度不确定性影响着流域负荷削减及其决策. 针对决策过程中主观层面产生的深度不确定性,本文提出了包含3个部分的稳健性决策方法：①确定衡量主观因素的权重的合理范围,采取适当采样方法确定每组合理权重下的最优决策；②稳健性指标计算；③以稳健性指标为目标结合进化算法求出稳健决策. 将上述方法应用到江西八里湖的总氮（TN）负荷削减决策之中,该决策问题的相关模型考虑到了污染负荷与水质的响应关系及客观不确定性. 以此为基础,在八里湖流域TN 277~297 t的年负荷削减水平下分析主观深度不确定性的影响. 结果显示,即便结构相似的权重,细微的改变也能导致不同权重下的最优决策发生比较大的变化,而通过上述方法可以识别出污染负荷削减的稳健决策,该决策以点源及特定子流域的面源控制为主. 在削减率上下波动5%（2.5%）的范围内,上述方案能够在83.5%（40.3%）的合理权重情况下保持最优,说明该决策能够满足绝大部分情况下对于不同站点之间的权衡考虑,可以为最终决策提供支持.
- Abstract：Deep uncertainty will affect the decision making in watershed load reduction. A strategy for robust decision making was proposed in this study, aiming at dealing with the deep uncertainty from subjective aspect. It includes three steps: ① determining the reasonable range of weights that used to measure subjective factors and solve optimal decision for each sampled weight from reasonable range; ② calculating robustness index; and ③ solving the optimization using evolutionary algorithm which takes robustness index as objectives. The proposed method was applied to the decision making on watershed total nitrogen (TN) load reduction in Lake Bali, Jiangxi Province. The basic model was established to explore the response between load reduction and water quality changes considering objective uncertainty. Analysis of subjective deep uncertainty was made on the annual reduction 277~297 t. The results showed that small change in certainty weight would lead to obvious change in optimal decisions. The proposed strategy could identify robust decision, where the reduction of point source pollution and some specific non-point source pollution should be in the priority. Considering the reduction rate fluctuating over 5% (2.5%), the robust decision maintained optimality over 83.5% (40.3%) reasonable weights indicating that it could satisfy most consideration of different monitoring stations and support the decision making.
摘要点击次数： 1068 全文下载次数： 2091
