Applied Machine Learning in Development of Geospatial Information Tools for Sustainable Groundwater Management
Groundwater plays an important role in sustainable water resource management. Globally, the development of groundwater resources for irrigation provides a stable water source and enhances food security. However, developing groundwater resources is difficult, requiring the collection, synthesis, analysis, and dissemination of information about groundwater in an accessible, effective manner so decision makers have the knowledge and tools to implement sustainable management strategies. In this dissertation, I present solutions to assist with the paucity of groundwater data. The first problem I explore is extending the Palmer Drought Severity Index to near present day. This dataset is crucial to current groundwater imputation methods, and is useful for new methods, as there is a moderate to strong correlation between drought and groundwater usage. Next, I present two novel methods for imputing groundwater level values. The first approach combines in situ groundwater data with Earth observation data using inductive bias to predict long-term trends using machine learning; this approach is called satellite imputation. The second approach relies on identifying geospatial trends and using observations from nearby wells to correct, or refine, initial imputation results; this is referred to as iterative refinement. The tools developed permit stakeholders at all levels to assess and interpret groundwater data with minimal computation infrastructure. These tools are developed by analyzing aquifers in the United States but are intended to be used in the unique conditions of West Africa, where groundwater data are particularly scarce and computer infrastructure, bandwidth, and support are limited.
College and Department
Ira A. Fulton College of Engineering and Technology
BYU ScholarsArchive Citation
Ramirez, Saul Gallegos, "Applied Machine Learning in Development of Geospatial Information Tools for Sustainable Groundwater Management" (2023). Theses and Dissertations. 9828.
Sparse Time Series Imputation, Machine Learning, Groundwater, Time series imputation, Earth observation