A three-dimensional numerical model of predevelopment conditions in the Death Valley regional ground-water flow system, Nevada and California

In the early 1990's, two numerical models of the Death Valley regional ground-water flow system were developed by the U.S. Department of Energy. In general, the two models were based on the same basic hydrogeologic data set. In 1998, the U.S. Department of Energy requested that the U.S. Geological Survey develop and maintain a ground-water flow model of the Death Valley region in support of U.S. Department of Energy programs at the Nevada Test Site. The purpose of developing this "second-generation" regional model was to enhance the knowledge an understanding of the ground-water flow system as new information and tools are developed. The U.S. Geological Survey also was encouraged by the U.S. Department of Energy to cooperate to the fullest extent with other Federal, State, and local entities in the region to take advantage of the benefits of their knowledge and expertise.

The short-term objective of the Death Valley regional ground-water flow system project was to develop a steady-state representation of the predevelopment conditions of the ground-water flow system utilizing the two geologic interpretations used to develop the previous numerical models. The long-term objective of this project was to construct and calibrate a transient model that simulates the ground-water conditions of the study area over the historical record that utilizes a newly interpreted hydrogeologic conceptual model. This report describes the result of the predevelopment steady-state model construction and calibration.

The Death Valley regional ground-water flow system is situated within the southern Great Basin, a subprovince of the Basin and Range physiographic province, bounded by latitudes 35 degrees north and 38 degrees 15 minutes north and by longitudes 115 and 118 degrees west. Hydrology in the region is a result of both the arid climatic conditions and the complex geology. Ground-water flow generally can be described as dominated by interbasinal flow and may be conceptualized as having two main components: a series of relatively shallow and localized flow paths that are superimposed on deeper regional flow paths. A significant component of the regional ground-water flow is through a thick Paleozoic carbonate rock sequence. Throughout the flow system, ground water flows through zones of high transmissivity that have resulted from regional faulting and fracturing.

The conceptual model of the Death Valley regional ground-water flow system used for this study is adapted from the two previous ground-water modeling studies. The three-dimensional digital hydrogeologic framework model developed for the region also contains elements of both of the hydrogeologic framework models used in the previous investigations. As dictated by project scope, very little reinterpretation and refinement were made where these two framework models disagree; therefore, limitations in the hydrogeologic representation of the flow system exist. Despite limitations, the framework model provides the best representation to date of the hydrogeologic units and structures that control regional ground-water flow and serves as an important information source used to construct and calibrate the predevelopment, steady-state flow model.

In addition to the hydrogeologic framework, a complex array of mechanisms accounts for flow into, through, and out of the regional ground-water flow system. Natural discharges from the regional ground-water flow system occur by evapotranspiration, springs, and subsurface outflow. In this study, evapotranspiration rates were adapted from a related investigation that developed maps of evapotranspiration areas and computed rates from micrometeorological data collected within the local area over a multiyear period. In some cases, historical spring flow records were used to derive ground-water discharge rates for isolated regional springs.

For this investigation, a process-based, numerical model was developed to estimate net infiltration. This result can be used to represent recharge by assuming that water that infiltrates past the “root zone” ultimately becomes ground-water recharge. Net infiltration based on this water-balance approach, however, suggests that recharge in the region exceeds the estimated ground-water discharge for the basin by almost a factor of four. The net infiltration model contains some assumptions and was limited in regard to soil moisture, bedrock permeability, and precipitation distributions that result in these higher than expected values. Despite this limitation, the net infiltration model does provide a good starting point for estimating regional ground-water recharge in the ground-water flow model.

The Death Valley regional ground-water flow system was simulated using a three-dimensional steady-state model that incorporated a nonlinear least-squares regression technique to estimate selected model parameters. The numerical modeling program MODFLOW-2000 was used in creating a finite-difference model consisting of 194 rows, 160 columns, and 15 layers. The grid cells were oriented north-south and were of uniform size, with side dimensions of 1,500 meters. The layers span thicknesses of 50 to 300 meters. The model grid encompasses about 70,000 square kilometers.

The required initial model parameter values were supplied by discretization of the three-dimensional hydrogeologic framework model and digital representations of the remaining conceptual model components. The three-dimensional simulation and corresponding sensitivity analysis supported the hypothesis of interactions between a relatively shallow local and subregional flow system and a deeper dominant regional system controlled by the carbonate aquifer.

During calibration of the model, techniques available in MODFLOW-2000 allowed for estimation of a series of parameters that provided a best fit to observed hydraulic heads and flows. Numerous conceptual models were evaluated to test the validity of various interpretations about the flow system. Only those conceptual model changes contributing to a significant improvement in model fit, as indicated by a reduction in the sum of squared errors, were retained in the final optimized model. The final model was evaluated to assess the likely accuracy of simulated results by comparing measured and expected quantities with simulated values. Evaluation of the model indicates that although the model is clearly an improvement on previous representations of the flow system, there is an indication of important uncertainties and model error.