Scientific Investigations Report 2007–5009
U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2007–5009
Regional and local models of ground-water flow in the northeastern San Joaquin Valley were developed as part of a study of the transport and fate of agricultural chemicals by the U.S. Geological Survey’s (USGS) National Water-Quality Assessment Program. The northeastern San Joaquin Valley, near the city of Modesto, is a semi-arid region dominated by irrigated agriculture. The aquifer system is predominantly alluvial. An unconfined to semiconfined aquifer overlies a confined aquifer in the southwestern part of the study area; these aquifers are separated by the lacustrine Corcoran Clay. Ground water is a key source of water for irrigation and public supply, and exploitation of this resource has altered the natural flow system. Primarily lateral flow from the mountain front to the San Joaquin River has been overprinted with down- ward flow driven by enhanced recharge from irrigation and discharge from irrigation pumpage.
A regional-scale, 16-layer steady-state model of ground-water flow in the unconfined to confined aquifer system underlying the Modesto area was developed to provide boundary conditions for an embedded 110-layer steady-state local-scale model of part of the aquifer system overlying the Corcoran Clay along the Merced River. The purpose of the local-scale model was to develop a better understanding of the aquifer system and to provide a basis for simulating reactive transport of agricultural chemicals.
Lateral boundary conditions for the regional model were no-flow along the Sierra Nevada foothills and general-head elsewhere. The water table was a free surface, and the lower boundary, far below the depth of wells, was no-flow. Recharge and agricultural pumpage were estimated using a water-budget approach, and bare-soil evaporation was simulated where the water table was shallow. Agricultural pumpage and recharge from irrigation return flow were the dominant components of the simulated water budget for the regional model.
Lateral boundary conditions for the local model, embedded within the regional, were specified fluxes from the regional model. The upper boundary was a free surface, and the lower boundary was general-head, allowing flow through the Corcoran Clay. Recharge from irrigation return flow was estimated on the basis of crop type and irrigation method for each field. Ground water generally was not used in the local model area. The dominant components of the simulated local budget were recharge from irrigation return flow and discharge to the Merced River.
The heterogeneity of aquifer materials was explicitly incorporated into the regional and local models using information from geologic and drillers’ logs of boreholes. Aquifer materials were differentiated in the regional model by the percentage of coarse-grained sediments in a cell. This percentage was estimated for most cells in the model by three-dimensional kriging and by downward extrapolation where kriging did not generate a value owing to data scarcity. Aquifer materials were differentiated in the local model by four hydrofacies (sand, silty sand, silt, and clay). Multiple realizations of the distribution of these hydrofacies were estimated using a transition-probability geostatistical technique that honored high-quality geologic logs and was conditionally influenced by data from drillers’ logs.
Different calibration methods were used for the regional and the local models. The regional model was calibrated by exploring a range of end-member coarse- and fine-grained hydraulic conductivities used to generate horizontal and vertical conductivities for each cell using texture-weighted arithmetic and harmonic means, respectively. The local model was calibrated primarily using the parameter estimation capabilities within MODFLOW-2000.
The calibrated horizontal hydraulic conductivities of the coarse-grained materials in the zone above the Corcoran Clay in the regional model and of the sand hydrofacies of the local model were about equal (30–80 m/d [meter per day]). The vertical hydraulic conductivity of the same zone in the regional model (median of 0.012 m/d), which is dominated by the finer-grained materials, was about an order of magnitude less than that for the clay hydrofacies in the local model.
Data used for calibration of both models included long-term hourly water-level measurements in 20 short-screened wells installed by the USGS in the Modesto and the Merced River areas. Additional data used for calibrating the regional model included water-level measurements in 11 wells upslope and 17 wells downslope from these areas. The root mean square error was 2.3 m (meter) for all wells in the regional model and 0.8 m for the USGS wells; the associated average errors were 0.9 m and 0.3 m, respectively. The root mean square error for the 12 USGS wells in the local model area was 0.08 m; the average error was 0 m. Particle tracking was used for the local model to estimate the concentration of an environmental tracer, sulfur hexafluoride, in 10 USGS wells near the Merced River that were sampled for this constituent. Measured and estimated concentrations in the mid-depth and deepest wells, those most sensitive to errors in hydraulic conductivity estimates, were consistent. The combined results of particle tracking and sulfur hexafluoride analysis suggest that most water sampled from the transect wells was recharged less that 25 years ago.
U.S.
Department of the Interior | U.S. Geological Survey
Persistent URL: https://pubs.water.usgs.gov/sir20075009
Page Contact Information: Publications Team
Page Last Modified: Thursday, 01-Dec-2016 19:40:18 EST
