A comparison of groundwater fluxes computed with MODFLOW and a mixing model using deuterium: Application to the eastern Nevada Test Site and vicinity Rosemary W.H. Carroll * , Greg M. Pohll, Sam Earman, Ronald L. Hershey Desert Research Institute, Division of Hydrologic Sciences, 2215 Raggio Parkway, Reno, NV 89512, United States Received 9 February 2008; received in revised form 24 July 2008; accepted 7 August 2008 KEYWORDS MODFLOW; Isotope mixing model; Nevada Test Site; Deuterium Summary The primary objective of this study was to verify groundwater flows in the vicinity of the eastern Nevada Test Site (NTS) computed with a hydraulically defined flow model against groundwater dD values computed by a steady-state mixing model. The Uni- ted States Geological Survey’s Death Valley regional flow model (DVRFM) is a transient, three-dimensional, groundwater model that uses the public domain, finite-difference code MODFLOW. The mixing model (Discrete-State Compartment Model-Shuffled Complex Evolution, or DSCM-SCE) is a recently developed code that is able to autocalibrate ground- water fluxes (both magnitude and direction) to best match observed tracer concentrations in groundwater. To compare modeling approaches, DVRFM boundary conditions and cell- to-cell interactions were implemented into a previously developed 15-cell DSCM-SCE dD model of the eastern NTS. Analysis of dD and d 18 data conducted throughout the model domain suggests recharge and mixing are the dominant mechanisms for groundwater iso- topic enrichment in the downgradient direction. Therefore, evaporation, at least at the regional scale, was ignored. Model results showed that DVRFM boundary fluxes and cell outflow volumes reproduced observed groundwater dD values in nine of the 11 hydro- graphic basins that contained deuterium data. The remaining four modeled basins did not have any deuterium data, however, modeling results closely matched estimated dD for two of these basins. The isotope mixing model independently verified that most of the hydraulically defined groundwater flows simulated by MODFLOW are reasonable. Opti- mization of the DSCM-SCE was then done by adjusting groundwater fluxes between cells to improve predicted groundwater dD values. This significantly lowered the weighted root- mean-squared error and helped to identify several basins where DVRFM–simulated bound- 0022-1694/$ - see front matter ª 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jhydrol.2008.08.005 * Corresponding author. Tel.: +1 970 349 0356; fax: +1 775 673 7363. E-mail address: Rosemary.Carroll@dri.edu (R.W.H. Carroll). Journal of Hydrology (2008) 361, 371– 385 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jhydrol