Qurban, Ibraheem A., authorGates, Timothy K., advisorBailey, Ryan T., committee memberGrigg, Neil S., committee memberIppolito, James A., committee member2024-09-092024-09-092024https://hdl.handle.net/10217/239208The Lower Arkansas River Valley (LARV) in southeastern Colorado has been a source of significant agricultural productivity for well over a century, primarily due to extensive irrigation practices. Mirroring trends seen in other semi-arid irrigated areas globally, however, irrigated agriculture in the LARV has resulted in several challenges for the region. In addition to the emergence of waterlogging and soil salinization, leading to decreased crop yields, elevated levels of nutrients and trace elements have appeared in the soil and water. Among these constituents, uranium (U), along with co-contaminants selenium (Se) and nitrate (NO3), has shown particularly high concentrations in groundwater, surface water, and soils. These heightened concentrations pose environmental concerns, impacting human health and the well-being of aquatic life such as fish and waterfowl. Careful monitoring and management practices are crucial to prevent potential harm to water resources. The main goal of this research is to develop a comprehensive numerical model for assessing U pollution in a stream-aquifer system within a large irrigated area. To achieve this, a computational model is built and tested that can predict with reasonable accuracy how U, along with Se and NO3, are mobilized and move within a coupled system of streams and groundwater. The approach combines two key modeling components: a MODFLOW package, which handles the simulation of groundwater and stream flow dynamics, and an RT3D package, which addresses the reactive transport of U, Se, and nitrogen (N) species in both groundwater and interconnected streams. RT3D relies on the simulated flows generated by MODFLOW to track the movement of U, Se, and N species between streams and the aquifer in the irrigated landscape, updating daily to adequately capture changes over time. This integrated model provides an understanding of how these contaminants behave and interact within the stream-aquifer system, aiding in effective pollution assessment and providing insights valuable to the planning of management strategies. The coupled MODFLOW-RT3D flow and reactive transport model is applied to a 550 km² area within the LARV, stretching from Lamar, Colorado, to the Colorado-Kansas border and spanning a period of 14 years. The flow package is compared with observations of groundwater hydraulic head and stream flow, along with estimates of return flow along the Arkansas River. The reactive transport package is assessed by comparing predicted U, Se, and NO3 concentrations against data collected from groundwater monitoring wells and stream sampling sites along with estimates of solute mass loads to the river. To calibrate and refine the model, the PESTPP-iES iterative ensemble smoother (iES) software is employed. This calibration process is dedicated to enhancing the model's accuracy in predicting both flow and transport dynamics. PESTPP-iES addresses calibration uncertainty by establishing prior frequency distributions for key model parameters based on data and expertise, then iteratively adjusts these parameters during calibration to align model predictions with observed data. Post-calibration, posterior distributions reflect updated parameter values and reduced uncertainties. Demonstrating a strong alignment with concentrations of CNO3, CSe, and CU values found in groundwater, streams, and the mass loading entering the Arkansas River, outcomes of the model-based simulations reveal a substantial violation of the Colorado chronic standard (85th percentile = 30 μg/L) for CU throughout the study region. On average, simulated CNO3, CSe, and CU values for groundwater in non-riparian areas in the region are 3.6 mg/L, 41 µg/L, and 126 µg/L, compared to respective averages of 4 mg/L, 53 µg/L, and 112 µg/L observed in monitoring wells. When considering the 85th percentile of simulated CNO3, CSe, and CU values, the figures for non-riparian groundwater are 6 mg/L, 50 µg/L, and 218 µg/L, respectively. Groundwater in riparian areas shows lower average simulated CNO3, CSe, and CU values of 3 mg/L, 26 µg/L, and 72 µg/L, respectively, and 85th percentile values of 5 mg/L, 41 µg/L, and 152 µg/L. Additionally, simulated average mass loading rates for NO3, Se, and U along the river are 8.8 kg/day per km, 0.05 kg/day per km, and 0.27 kg/day/km respectively, compared to stochastic mass balance estimates of 9.2 kg/day per km , 0.06 kg/day per km , and 0.23 kg/day per km. The simulated 85th percentile CNO3, CSe, and CU values in the Arkansas River are 1 mg/L, 11 μg/L, and 87 μg/L, respectively. Notably, the simulated U levels in groundwater exceed the chronic standard across 44% of the region. Along the Arkansas River, concentrations consistently surpass the chronic standard, averaging 2.9 times higher. Predicted Se concentrations also show significant exceedances of the chronic standard, while NO3 violations are slight to moderate. The varying pollutant levels across the region highlight specific areas of concern that require targeted attention, indicating potential contributing factors to these hotspots. Findings outline how serious and widespread the problem is in the LARV, providing a starting point for comparing potential pollution reduction from alternative water and land best management strategies (BMPs) to be explored in future applications of the calibrated model.born digitaldoctoral dissertationsengCopyright and other restrictions may apply. User is responsible for compliance with all applicable laws. For information about copyright law, please see https://libguides.colostate.edu/copyright.Text