Related Experiment Video
Updated: Jun 23, 2026
11:53
Spatial Multiobjective Optimization of Agricultural Conservation Practices using a SWAT Model and an Evolutionary Algorithm
Published on: December 9, 2012
13.0K
Machine learning for groundwater pollution source identification and monitoring network optimization
Yiannis N Kontos1,2, Theodosios Kassandros2, Konstantinos Perifanos3
1School of Civil Engineering, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece.
Neural Computing & Applications
|July 5, 2022
Summary
Identifying groundwater pollution sources requires optimized monitoring networks. This study uses machine learning and simulations to develop effective strategies for pinpointing contaminant origins, adaptable to various budgets.
Area of Science:
- Environmental Science
- Hydrogeology
- Data Science
Background:
- Effective groundwater pollution management hinges on accurately identifying contaminant sources.
- Optimizing groundwater monitoring networks is crucial for solving the inverse modeling problem of source identification.
- Current methods often lack the adaptability needed for diverse budget constraints.
Purpose of the Study:
- To develop and evaluate a methodology for optimizing groundwater monitoring networks to identify pollution sources.
- To apply machine learning algorithms to solve the inverse modeling problem in groundwater pollution studies.
- To create adaptable monitoring strategies suitable for various budget limitations.
Main Methods:
- Simulations of a theoretical confined aquifer with pumping wells and potential pollution sources.
- Generation of synthetic datasets using different groundwater pollution modeling approaches.
- Training of classification (random forests, multilayer perceptron) and computer vision (convolutional neural networks) algorithms on formulated datasets.
- Feature selection and trial-and-error tests for optimizing monitoring well placement and sampling frequency.
Main Results:
- The proposed methodology successfully trains machine learning models to identify groundwater pollution sources.
- Demonstrated ability to generate effective, albeit sub-optimal, monitoring strategies tailored to different budget scenarios.
- Feature selection and algorithmic approaches proved effective in optimizing monitoring network design.
Conclusions:
- Machine learning, combined with simulation-based synthetic data, offers a powerful approach for groundwater pollution source identification.
- The developed methodology provides a flexible framework for designing cost-effective groundwater monitoring networks.
- This research contributes to improved environmental management through enhanced understanding of contaminant transport and source localization.
