利用信息融合提升水污染数值溯源性能

王寒涛; 杨睿意; 平扬; 李鸿鸣; 陈圆; 姜继平

doi:10.13205/j.hjgc.202512010

利用信息融合提升水污染数值溯源性能

doi: 10.13205/j.hjgc.202512010

王寒涛¹,
杨睿意²,
平扬¹,
李鸿鸣³,
陈圆^2,4, ,,
姜继平²

1. 中电建生态环境集团有限公司, 广东深圳 518101;
2. 南方科技大学, 广东深圳 518000;
3. 中电建水环境科技有限公司, 广东深圳 518101;
4. 深圳市智薯环保科技有限公司, 广东深圳 518000

基金项目:

国家自然科学基金面上项目（51979136）；中电建研发课题城市智慧排水系统构建关键技术与装备研发（ST-ZB-ZC-JY-JS-2023-15）

详细信息

作者简介:
王寒涛（1980—），男，工程师，主要研究方向为水环境治理。wanghant-shj@powerchina.com

通讯作者:
陈圆（1997—），男，工程师，主要研究方向水环境水质水动力数值模拟。cheny@zsest.cn

计量
- 文章访问数: 7
- HTML全文浏览量: 2
- PDF下载量: 0
- 被引次数: 0
出版历程
- 收稿日期: 2024-09-29
- 录用日期: 2024-11-18
- 修回日期: 2024-11-02
- 网络出版日期: 2026-01-09

Enhancing water pollution numerical source tracking performance using information fusion

1. Power China Eco-Environmental Group Co., Ltd., Shenzhen 518101, China;
2. Southern University of Science and Technology, Shenzhen 518000, China;
3. Power China Water Environmental Technology Co., Ltd., Shenzhen 518101, China;
4. Shenzhen Zhishu Environmental Protection Technology Co., Ltd., Shenzhen 518000, China

摘要

摘要: 水污染溯源是环境监管的一个重要场景，基于数值法的污染溯源技术近年来得到了广泛关注。然而现有的数值法溯源往往只考虑利用一种污染物信息来溯源，忽略了同时排放的不同类别污染物，完全可以有效利用不同的污染物信息来提升数值溯源的可靠性和精度。为此，以AM-MCMC贝叶斯推理溯源模型为基准数值溯源算法，模拟了3种污染物同时同地排放的情景，设计了2类污染物信息融合的路径，结合数据同化进行数值溯源性能提升效果的研究。结果表明：信息融合显著提升了排放时刻和位置2个误差较大的源项参数的反演性能，特别是采用马尔科夫链加权的信息融合路径；排放位置反演结果的不确定性区间（95%置信区间）平均从10%提升到1%误差范围，精度（相对平均误差）从9.6%~18.6%提升到2.9%。信息融合结合数据同化的框架，不增加监测负担，只需要增加现有的监测的水质参数就可极大提升溯源的稳健性，具有较高实践价值。
- 水环境污染 /
- 数值溯源 /
- 数据同化 /
- 信息融合 /
- AM-MCMC
Abstract: Water pollution source tracing is an important aspect of environmental monitoring. In recent years， pollution source tracing technologies based on numerical methods have gained widespread attention. However， existing numerical source tracing methods often consider only the information of a single pollutant， neglecting the simultaneous discharge of different types of pollutants. It is entirely possible to effectively utilize information from various pollutants to enhance the reliability and accuracy of numerical source tracing. To this end， this study used the AM-MCMC Bayesian inference source tracing model as the baseline numerical tracing algorithm and simulated a scenario of three pollutants being discharged simultaneously and at the same location. Two pollutant information fusion paths were designed， and the effect of combining data assimilation on the performance enhancement of numerical source tracing was investigated. The results showed that information fusion significantly improved the inversion performance of two source parameters， the emission time and location， which have relatively large errors. In particular， the information fusion path weighted by the Markov chain yielded significant improvements. For example， the uncertainty range （by 95% confidence interval） of the emission location inversion result was reduced from an average of 10% to 1%， and the accuracy （relative mean error） improved from 9.6%~18.6%， to 2.9%. The framework of information fusion combined with data assimilation didn't increase the monitoring burden. It only required additional existing water quality parameters to significantly enhance the robustness of the source tracing， demonstrating high value in practice.
- water environmental pollution /
- numerical source tracking /
- data assimilation /
- information fusion /
- AM-MCMC

HTML全文

参考文献(27)

[1]	TANG W，PEI Y，ZHENG H，et al. Twenty years of China's water pollution control：experiences and challenges［J］. Chemosphere，2022，295：133875.
[2]	LI S C，AN Y. Environmental incidents in China from November to December 2022［J］. Journal of Safety and Environment，2023，23（3）：971-974. 李生才，安莹. 2022年11—12月国内环境事件［J］. 安全与环境学报，2023，23（3）：971-974.
[3]	LIU R，JIANG J，GUO L，et al. Screening of pollution control and clean-up materials for river chemical spills using the multiple case-based reasoning method with a difference-driven revision strategy［J］. Environmental Science and Pollution Research，2016，23（11）：11247-11256.
[4]	LIU X D，WANG J. Research progress on identification methods of surface water pollution sources［J］. Advances in Water Science，2020，31（2）：302-311. 刘晓东，王珏. 地表水污染源识别方法研究进展［J］. 水科学进展，2020，31（2）：302-311.
[5]	JIN Z Q，CHEN X Q. Solving the source term control inverse problem of the convection-diffusion equation using pulse spectrum optimization method［J］. Journal of Hohai University（Natural Science Edition），1992，20（2）：1-8. 金忠青，陈夕庆. 用脉冲谱-优化法求解对流-扩散方程源项控制反问题［J］. 河海大学学报（自然科学版），1992，20（2）：1-8.
[6]	LIU X D，YAO Q，XUE H Q，et al. Research progress on inverse problems in environmental hydraulics［J］. Advances in Water Science，2009，20（6）：885-893. 刘晓东，姚琪，薛红琴，等. 环境水力学反问题研究进展［J］. 水科学进展，2009，20（6）：885-893.
[7]	ZHOU W G. Research on inverse problems in environmental hydraulics based on adjoint assimilation method［D］. Wuhan：Wuhan University，2019. 周武刚. 基于伴随同化法的环境水力学反问题研究［D］. 武汉：武汉大学，2019.
[8]	MAZAHERI M，SAMANI J M V，SAMANI H M. Mathematical model for pollution source identification in rivers［J］. Environmental Forensics，2015，16（4）：310-321.
[9]	ZEUNERT S，MEON G. Influence of the spatial and temporal monitoring design on the identification of an instantaneous pollutant release in a river［J］. Advances in Water Resources，2020，146：103788.
[10]	BARATI M M，MAZAHERI M，MOHAMMAD V S J. Inverse modeling of contaminant transport for pollution source identification in surface and groundwaters：a review［J］. Groundwater for Sustainable Development，2021，15：100651.
[11]	YANG R，JIANG J，PANG T，et al. Crucial time of emergency monitoring for reliable numerical pollution source identification［J］. Water Research，2024，265：122303.
[12]	WANG J，ZHAO J，LEI X，et al. An effective method for point pollution source identification in rivers with performance-improved ensemble Kalman filter［J］. Journal of Hydrology，2019，577：123991.
[13]	CHO K H，PACHEPSKY Y，LIGARAY M，et al. Data assimilation in surface water quality modeling：a review［J］. Water Research，2020，186：116307.
[14]	WANG J，ZHAO J，LEI X，et al. New approach for point pollution source identification in rivers based on the backward probability method［J］. Environmental Pollution，2018，241：759-774.
[15]	WIKLE C K，BERLINER L M. A Bayesian tutorial for data assimilation［J］. Physica D：Nonlinear Phenomena，2007，230（1-2）：1-16.
[16]	LI X，LIU F. A mathematical comparison of data assimilation and machine learning in Earth system state estimation from a Bayesian inference viewpoint［J］. Information Geography，2025，1（1）：100001.
[17]	ZHU Z，LI Y，SUN Y，et al. A decision support framework for pollution source detection via coupled forward-inverse optimization and multi-information fusion［J］. Water Resources Research，2023，59（6）：e2022WR032866.
[18]	DE SMEDT F，BREVIS W，DEBELS P. Analytical solution for solute transport resulting from instantaneous injection in streams with transient storage［J］. Journal of Hydrology，2005，315（1）：25-39.
[19]	CROMPTON E J. Traveltime for the Truckee River between Tahoe City，California，and Vista，Nevada，2006 and 2007［R］. Reno：U.S. Geological Survey，2008：1-42.
[20]	JIANG J P，DONG F J，LIU R T，et al. Bayesian pollution source tracing based on river tracer experiment：algorithm parameters，influencing factors，and comparison with frequency method［J］. China Environmental Science，2017，37（10）：3813-3825. 姜继平，董芙嘉，刘仁涛，等. 基于河流示踪实验的Bayes污染溯源：算法参数、影响因素及频率法对比［J］. 中国环境科学，2017，37（10）：3813-3825.
[21]	RIVORD J. Modeling contaminant spills on the upper Truckee River in California and Nevada［D］. Reno：University of Nevada，Reno，2008.
[22]	LI Z，MAO X Z，LI T S，et al. Estimation of river pollution source using the space-time radial basis collocation method［J］. Advances in Water Resources，2016，88：68-79.
[23]	MARSHALL L，NOTT D，SHARMA A. A comparative study of Markov chain Monte Carlo methods for conceptual rainfall-runoff modeling［J］. Water Resources Research，2004，40（2）：W02503.
[24]	KHALEGHI B，KHAMIS A，KARRAY F O，et al. Multisensor data fusion：a review of the state-of-the-art［J］. Information Fusion，2013，14（1）：28-44.
[25]	YU J，XU L，XIE X，et al. Contamination event detection method using multi-stations temporal-spatial information based on Bayesian network in water distribution systems［J］. Water，2017，9（11）：865-878.
[26]	ZHE Z. A decision support framework for pollution source detection via coupled forward-inverse optimization and multi-information fusion［Z］. 2022：1-38.
[27]	JIANG J，HAN F，ZHENG Y，et al. Inverse uncertainty characteristics of pollution source identification for river chemical spill incidents by stochastic analysis［J］. Frontiers of Environmental Science& Engineering，2018，12（5）：1-10.