A new numerical ground-water-flow model (Edwards aquifer model) that incorporates important components of the latest information and plausible conceptualization of the Edwards aquifer was developed. The model includes both the San Antonio and Barton Springs segments of the Edwards aquifer in the San Antonio region, Texas, and was calibrated for steady-state (1939–46) and transient (1947–2000) conditions, excluding Travis County. Transient simulations were conducted using monthly recharge and pumpage (withdrawal) data. The model incorporates conduits simulated as continuously connected (other than being separated in eastern Uvalde and southwestern Medina Counties), one-cell-wide (1,320 feet) zones with very large hydraulic-conductivity values (as much as 300,000 feet per day). The locations of the conduits were based on a number of factors, including major potentiometric-surface troughs in the aquifer, the presence of sinking streams, geochemical information, and geologic structures (for example, faults and grabens). The simulated directions of flow in the Edwards aquifer model are most strongly influenced by the presence of simulated conduits and barrier faults. The simulated flow in the Edwards aquifer is influenced by the locations of the simulated conduits, which tend to facilitate flow.

The simulated subregional flow directions generally are toward the nearest conduit and subsequently along the conduits from the recharge zone into the confined zone and toward the major springs. Structures simulated in the Edwards aquifer model influencing groundwater flow that tend to restrict flow are barrier faults. The influence of simulated barrier faults on flow directions is most evident in northern Medina County.

A water budget is an accounting of inflow to, outflow from, and storage change in the aquifer. For the Edwards aquifer model steady-state simulation, recharge (from seepage losses from streams and infiltration of rainfall) accounts for 93.5 percent of the sources of water to the Edwards aquifer, and inflow through the northern and northwestern model boundaries contributes 6.5 percent. The largest discharges are springflow (73.7 percent) and ground-water withdrawals by wells (25.7 percent).

The principal source of water to the Edwards aquifer for the Edwards aquifer model transient simulation was recharge, constituting about 60 percent of the sources of water (excluding change in storage) to the Edwards aquifer during 1956, a drought period, and about 97 percent of the sources (excluding change in storage) during 1975, a period of above-normal rainfall and recharge. The principal discharges from the Edwards aquifer for the transient simulation were springflow and withdrawals by wells. During 1956, representing drought conditions, the change in storage (net water released from storage) was much greater than recharge, accounting for 75.9 percent of the total flow compared to 14.5 percent for recharge. Conversely, during 1975, representing above-normal rainfall and recharge conditions, recharge constituted 79.9 percent of the total flow, compared to 7.1 percent for the change in storage (net water added to storage).

A series of sensitivity tests was made to ascertain how the model results were affected by variations greater than and less than the calibrated values of input data. Simulated hydraulic heads in the Edwards aquifer model were most sensitive to recharge, withdrawals, hydraulic conductivity of the conduit segments, and specific yield and were comparatively insensitive to spring-orifice conductance, northern boundary inflow, and specific storage. Simulated springflow in the Edwards aquifer model was most sensitive to recharge, withdrawals, hydraulic conductivity of the conduit segments, specific yield, and increases in northern boundary inflow and was comparatively insensitive to spring-orifice conductance and specific storage.