Use of environmental tracers to estimate parameters for a predevelopment-ground-water-flow model of the Middle Rio Grande Basin, New Mexico

The question of the availability of ground water as a long-term resource in the Middle Rio Grande Basin of central New Mexico has been addressed recently by the development of ground-water-flow models by the U.S. Geological Survey. An initial model constructed in 1994 was updated by Kernodle and others (1995), and then calibrated by Tiedeman and others (1998) using nonlinear regression methods and additional hydrologic observations. A new model was constructed using some of the results from the Middle Rio Grande Basin initiative by McAda and Barroll (2002). This report documents the use of 14C activities and the location of hydrochemical zones to constrain parameter values used in a predevelopment ground-water-flow model of the Middle Rio Grande Basin. The universal inverse modeling code, UCODE, was used to help estimate hydraulic conductivities of hydrogeologic units and current and past recharge along the basin margins and tributary rivers. The water levels in the basin were simulated using MODFLOW, and travel times to wells and source-area delineation were simulated using MODPATH. A three-dimensional geologic model was discretized into a three-dimensional MODFLOW grid of the basin. Major hydrogeologic units in the geologic model included volcanic rocks, and several units that represent the Santa Fe Group sediments, including ancestral gravels from the Rio Grande and some finer grained units that represent the middle and lower Santa Fe Group. The MODFLOW grid represented the hydrogeologic units with nine layers of variable thickness totaling up to 12,000 feet in places, and a uniform horizontal grid resolution of one square kilometer (0.386 square miles). The bottom of the model was considered to be the base of the poorly to semiconsolidated basin-fill sediments as defined by geophysical observations. Observations that were used to calibrate a steady-state predevelopment model, and then a transient paleohydrologic model, included 200 water levels and 200 14C activities. Observed water levels were compared with simulated water levels, and observed 14C activities were compared with simulated 14C activities based on travel times to individual wells. In addition, the distributions of ground water that originated from the Rio Grande and Rio Puerco were also used as constraints by comparing the percentage of river water in certain hydrochemical target regions with the percentage is simulated river water. The 14C activities were adjusted for chemical reactions along the flow paths and for long-term variation in atmospheric input. Hydraulic conductivities estimated for the model using the inverse procedure were similar to values that had been estimated in the previous models. The best-fit value of hydraulic conductivity of the Rio Grande alluvium and the volcanic rocks averaged about 30 feet per day, which is in agreement with field tests and earlier models. The best-fit hydraulic conductivity of a silty layer identified in the geologic model was estimated to be about 0.4 feet per day, which is also in agreement with field tests. The ratio of horizontal to vertical hydraulic conductivity was estimated for 12 different regions of the basin, with the best-fit ratios for the different regions ranging from 230:1 to 3,400:1. Basin-margin and tributary recharge estimates were lower than estimates used in previous models. The 1995 ground-water-flow model assigned total margin and tributary recharge values of 138,600 acre-feet per year, based primarily on previous estimates using the water-budget method. The 1998 version of the model estimated this external recharge to be 95,500 acre-feet per year, based on inverse modeling using primarily water levels. The 2002 version of the model used a combination of estimates from previous sources and those from a chloride mass-balance study to arrive at a recharge of 67,500 acre-feet per year. The present study estimates recharge at 35,700 acre-feet per year, based on inverse modeling that includes 200 ground-water ages and the distribution of river waters within the basin. The water-budget methods used to estimate recharge in the earlier models do not account for runoff that enters the Use of environmental tracers to estimate parameters for a predevelopment ground-water-flow model Rio Grande, or evapotranspiration of runoff once it enters the subsurface. In addition, recharge estimates for the mountain fronts on the eastern side of the basin have been made independently using the chloride mass-balance method. Estimates by the chloride method were used in the 1998 model and are close to the estimates made in the present study. The lower recharge estimates from the current model are also consistent with the simulated water levels and source-area delineation. A ground-water trough is simulated west of the Rio Grande that is partially occupied by ground water that is derived from the Rio Grande. A ground-water trough and the Rio Grandederived ground water have been observed using water levels and hydrochemistry, respectively. The 1995 model with the greatest recharge did not reproduce these features, and the 1998 and 2002 models used hydraulic conductivity zones or barriers to produce the trough. In addition to the steady-state predevelopment model, a transient paleohydrologic model was calibrated to determine if the 14C activities could indicate whether recharge rates had changed during the past 30,000 years. Paleolimnological evidence from central New Mexico has indicated that the climate in the region was wetter during the last glacial maximum (20,000 to 25,000 years ago). The paleohydrologic simulation involved a period of 30,000 years, with an separate value of recharge estimated every 2,500 years. These paleorecharge values were estimated simultaneously with the parameters from the original steady-state model. The transient, paleohydrologic simulation suggests that recharge to the basin during the last glacial maximum was 7 to 15 times greater than that at present, and after the end of the Ice Age was as little as half that at present. However, substantial uncertainties are associated with these paleorecharge estimates.