Monday, April 29, 2013: 5:20 p.m.
Regency East 3 (Hyatt Regency San Antonio)
The groundwater availability models available for evaluating management strategies in the Gulf Coast Aquifer System are constructed with large vertical layers that represent a single aquifer. Because of such thick model layers at the surface, potential problems can arise with adequate numerical representation of shallow, intermediate and deep flow systems. The relative importance of these potential problems was investigated by the development and construction of a regional model that covered a 10-county area and overlaps the coverage provided by two existing groundwater availability models for the Gulf Coast. The 10-county regional model (known as the LCRB model) was constructed to simulate pumping impacts for a proposed LCRA-SAWS water project and it benefited from a well-funded four-year period that included the development of an extensive site conceptual water budget model. The LCRB model was developed with significantly finer vertical grid spacing, including a model layer to specifically represent shallow groundwater flow. Comparison of the water budgets from the LCRB model (with fine vertical descritization) and from the two groundwater availability models (with coarse vertical descritization) show order of magnitude differences in the water budgets for a shared calibration period from 1980 to 1999. The model predictions differ the most in the source of the water for pumping. Whereas the primary source for pumped water for the LCRB model is recharge from precipitation, the primary source for the pumped water for one availability model is river water and the primary source for the pumped water for the other availability model is aquifer storage. The lessons learned from this comparison of regional models demonstrate that the numerical construction of a regional model has a major impact on the capability of the model to properly simulate water budgets and to accurately represent recharge from precipitation and groundwater-surface water interaction.