LEVERAGING DATA-DRIVEN AND PROCESS-BASED MODELING TO ESTIMATE GROUNDWATER BUDGETS FOR AQUIFER SYSTEMS SUPPORTING IRRIGATED AGRICULTURE

Abstract

ABSTRACTThe future of major aquifer systems supporting irrigated agriculture is threatened due to unsustainable groundwater pumping. Metering of pumping is key for implementing robust groundwater management, but metering is limited in most aquifers. Although machine learning methods have been used to estimate pumping over certain regions, these studies have not fully demonstrated the data quantity and input parameter requirements to accurately estimate regional groundwater pumping. Chapter two of this dissertation determined the data quantity required and identified relevant features to develop Random Forests-based annual groundwater pumping estimates (2008-2020) over the Kansas High Plains aquifer. Pumping is predicted at two spatial scales, i.e., point (well) and grid (2 km). Data requirements are evaluated using a combination of different training splits against a constant test set to understand the performance of the models. Summing predicted pumping over a 2 km grid was made possible with knowledge of crop irrigation area. This knowledge also decreased the uncertainty observed in linking individual wells with irrigated areas and further improved the spatial and temporal pumping estimates. At the 2 km scale, it is observed that a model trained on 10% of the total available data had coefficient of determination (R2) values of 0.98 and 0.75 for training and testing, respectively. These results show reasonable estimates of irrigation pumping are possible at the 2 km scale when 10% of irrigation wells are metered and if the irrigated area is known. This finding has significant implications for groundwater management in many heavily stressed aquifers. Groundwater recharge is a vital component for water budget and water availability assessments but is challenging to measure due to its complex dynamics and subsurface occurrence. Process-based modeling tools are often used to simulate recharge. Analyzing the numerous, interacting factors that control recharge is time consuming and computationally expensive. In chapter three, machine learning (ML) models are developed that predict recharge and use explainable AI (XAI) as a diagnostic tool to understand the drivers of groundwater recharge, capture critical threshold values, and identify high recharge zones. ML models are trained and tested using calibrated SWAT+ recharge values as the response variable. Two ML models are developed: (i) a surrogate model that captures the physics from the SWAT+ simulator by utilizing selected model inputs with a directly comparable definition of distributed parameter values; and (ii) a stand-alone predictor that utilizes commonly available open-source data to predict recharge and has annual temporal and 500 m spatial resolution for the period of 2002 – 2015. The study is implemented in the semi-arid Lower Arkansas river basin in Colorado, USA which contains an alluvial aquifer that is used for agricultural irrigation. Both the surrogate and stand-alone models demonstrated high predictive accuracy with R2 values of 0.98 and 0.91, respectively, when evaluated against test data. Explainable AI assessments show that average diffuse groundwater recharge increases significantly primarily when precipitation exceeds approximately 500 mm/year. Irrigation return flow is also identified as a major source of recharge where cultivated land cover is dominant. Unlike the surrogate model, which correctly identifies key drivers of recharge (e.g., precipitation, snow melt), XAI results from the stand-alone model identify some predictor variables that are correlated with recharge (e.g., high slope and certain LULC classes) but don’t physically drive recharge. These results highlight the effectiveness of ML surrogate models to meaningfully analyze physics-based simulation results. Stand-alone ML models may have high predictive accuracy but limited explanatory power. The methods adopted in this study provide a valuable approach for investigating recharge dynamics in unconfined aquifers. Remote sensing-based aquifer storage loss analysis has rendered relevant insight into groundwater systems supporting irrigated agriculture. Interferometric Synthetic Aperture Radar (InSAR) is a common tool for analyses of confined aquifer storage loss from fine compressible sediments. However, in aquifer systems where a significant amount of storage loss comes from unconfined aquifers, the application of InSAR to determine aquifer storage loss could limit our understanding of total storage loss. In chapter four, aquifer water budget from two groundwater flow models are analyzed, storage loss from unconfined and confined aquifers, and compared subsidence estimates from Interbed Storage Package (IBS) and InSAR. The study used groundwater flow models developed for a portion of the Diamond Flow system in Central Nevada, US. Model results show up to 52 m of drawdown for the irrigated region between 1955 and 2024. Storage loss from fine-grained consolidation, modeled with the IBS package, is not significant (i.e., less than 3 percent of the water storage loss from aquifer units) while the study area shows cumulative (2014 to 2024) subsidence of 600 mm and 800 mm estimates from IBS and InSAR. These results provide a counter-example to recent studies showing consolidation of fines contributes significantly (40-50%) to total storage loss in primarily confined aquifers, and highlights the differing magnitudes of storage loss mechanisms in unconfined aquifers. It also suggests use of remotely-sensed subsidence as a complementary, rather than sole estimation approach, for storage loss in unconfined aquifers.