Evaluation of stream capture related to groundwater pumping, Lower Humboldt River Basin, Nevada

The Humboldt River Basin is the only river basin that is contained entirely within the State of Nevada. The effect of groundwater pumping on the Humboldt River is not well understood. Tools are needed to determine stream capture and manage groundwater pumping in the Humboldt River Basin. The objective of this study is to estimate capture and storage change caused by groundwater withdrawals in the lower Humboldt River Basin that can provide the Nevada State Engineer with data and information needed to manage groundwater and surface-water resources.

A numerical groundwater flow model was developed for the purpose of estimating stream capture from pre-2016 and future pumping as well as for any location of potential future pumping within the lower Humboldt River Basin. This model was developed using MODFLOW-NWT to represent the lower Humboldt River Basin hydrologic system, including Humboldt River; Rye Patch Reservoir; groundwater evapotranspiration; pumping from municipal, agricultural, mining, and domestic wells; and agricultural drains. Aquifer properties were calibrated using results from numerous single- and multi-well aquifer tests (Nadler, 2020) and through the process of model calibration.

Historical capture was estimated for 1960–2016 and predictive capture for the system was projected 100 years into the future (2017–2116) based on historical pumping patterns. Stream capture and drain capture are relatively low for the historical and predictive periods. During the historical period, increased pumping during dry years caused increased connections with capture sources and less water sourced to wells from aquifer storage. Storage and groundwater levels generally recovered during subsequent wet years. Overall, storage change has been the main source of water to wells in the lower Humboldt River Basin, followed by groundwater evapotranspiration capture. During the predictive period, pumping is projected to remain constant and capture 9 percent of stream water after 100 years.

Capture and storage change maps were created to visualize spatial variability in potential capture and storage change through time and to provide a database of results that can be used to manage groundwater and surface-water resources. These maps show that potential stream capture would be a minor source of water to wells located across most of the simulated area, except for locations close to the Humboldt River and Rye Patch Reservoir. Drains also would be a minor potential source of water to wells except for those directly adjacent to the drains. In general, the potential supply of water to wells is storage-dominated and over time groundwater evapotranspiration-dominated in the agricultural area.

Capture difference maps were generated to visualize where potential capture results might have greater limitations associated with nonlinear flow processes, such as head-dependent boundary conditions. Higher capture differences indicate larger capture map bias and therefore greater capture map uncertainty due to the inability of capture maps to account for nonlinear flow processes. Stream capture differences are highest directly adjacent to the river but are otherwise minimal. Drain capture differences are highest in the region of the agricultural drain network but are otherwise minimal. The Humboldt River, Rye Patch Reservoir, and drains introduce very little nonlinearity to the model, and their associated capture map bias is minimal. Potential groundwater evapotranspiration capture introduces a fair amount of nonlinearity to the model and has the potential to result in significant, localized groundwater evapotranspiration capture map bias over time. Groundwater evapotranspiration capture differences are the result of higher pumping rates lowering the water table below the root zone faster than lower pumping rates and essentially removing groundwater evapotranspiration as a potential source of capture faster than lower pumping rates. Wells that can no longer source their supply through groundwater evapotranspiration capture then generally source more of their water from storage. Thus, storage change bias increases over time as well.

Capture prediction uncertainty due to parameter estimation was evaluated using a covariance matrix adaptation-evolution strategy. One hundred Monte Carlo realizations of model parameters were applied to the model to assess capture uncertainty at 13 grid cell locations within the model domain. In general, results indicated that greater capture uncertainty for a given source (river, drains, or evapotranspiration) is associated with proximity of a pumping well to that source. The magnitude of maximum capture fraction uncertainties after 100 years of pumping for stream capture, drain capture, groundwater evapotranspiration capture, and storage change were plus or minus (±) 0.17, ±0.10, ±0.20, and ±0.22, respectively.