Access Type
Open Access Dissertation
Date of Award
January 2023
Degree Type
Dissertation
Degree Name
Ph.D.
Department
Civil and Environmental Engineering
First Advisor
Carol J. Miller
Abstract
Groundwater plays a crucial role in sustaining ecosystems, providing drinking water, and supporting various industrial activities. As water scarcity becomes an increasingly pressing global issue, accurate groundwater level monitoring is essential for effective resource management and environmental protection. The study introduces a comprehensive methodology that integrates groundwater flow modeling, stochastic simulations, and ML algorithms to design efficient Groundwater Level Monitoring Networks (GLMNs) in regions with limited groundwater observation data. Two primary ML algorithms, K-means clustering and Relevance Vector Machine (RVM), are used to select and position observation wells strategically. The research addresses the challenge of minimizing data uncertainties and determining optimal well locations to maximize data collection while reducing the monitoring expenditure. To assess the effectiveness of the proposed approach, the study employs a real-world scenario in Metro Detroit, an area lacking comprehensive groundwater observation wells. The results demonstrate the utility of ML-based GLMNs in enhancing hydrogeological understanding, improving groundwater modeling accuracy, and managing installation budgets. The modeling performance is evaluated by using statistical error metrics to measure predictive accuracy and validate the modeling, comparing model predictions with real-observed groundwater levels. The research highlights that the groundwater models containing proposed GLMNs more accurately represent aquifer behaviors by minimizing errors, contributing to better-informed decision-making. Furthermore, the study evaluates the efficiency gains in terms of the calibration time required for model development. ML-based network design expedites the calibration process, enabling more efficient modeling and management of groundwater resources.
Recommended Citation
Teimoori, Sadaf, "Optimal Spatial Design Of Groundwater Level Monitoring Networks Using Machine Learning Methods" (2023). Wayne State University Dissertations. 3965.
https://digitalcommons.wayne.edu/oa_dissertations/3965