Numerical modeling and hydrologic analysis of floodplain hydrogeology, bank storage interactions, and underflow in the Colorado River alluvial aquifer

A comprehensive hydrogeological investigation was conducted within the Colorado River alluvial aquifer of south-central Texas. The study characterized bank storage interactions caused by irrigation and flood releases from the Highland Lake Reservoirs above Austin, and evaluated the underflow criterion for managing water resources in the Colorado River Basin. The hydrodynamics of the bank storage effect were studied in an integrated tracer and modeling investigation. Variations in dissolved solids between Colorado River water and ground-water created ideal conditions for analysis of stream/aquifer interaction during natural and artificial floodwaves. Two sites were instrumented with observation well and stream-stage recorders. At each site, head, temperature, and chemical data were collected during low-flow, mid-flow, and peak-flow conditions. Significant variations in stream stage and reversals of hydraulic gradient in the adjoining alluvial aquifer were created by releases of water from upstream reservoirs both in response to storm events and for irrigation by downstream users. Aquifer head responded relatively rapidly to changing stream stage during controlled reservoir releases, but tracer (both solute and heat) breakthrough curves were generally not observed even for observation wells very close to the Colorado River. The water table response lagged behind the average head response in the observation wells. These results are postulated to occur as a result of elastic head response in the shallow, unconfined aquifer to rapid changes in stream stage, analogous to the delayed yield effect in an unconfined aquifer during the early stages of pumping. The coupled-process numerical model HST3D was used to develop confined and unconfined aquifer models to simulate the stream/aquifer response. Although the aquifer is clearly recognizable as an unconfined aquifer, the confined aquifer model gave a superior simulation. Both modeling and field data support the hypothesis of elastic head recovery response in the unconfined aquifer. The underflow criterion was evaluated as a legal provision to prevent stream depletion of the Colorado River by high capacity alluvial wells. Study results indicate that the Colorado River alluvial aquifer is not an underflow-dominated system. Even so, site-specific investigations showed that temporally varying underflow zones sometimes exist very close to the river. In more than 90% of the floodplain, data are not available to identify these zones. Beyond the underflow zones, the hydraulic connection between the Colorado River and the alluvial aquifer may extend thousands of feet into the floodplain. Therefore, the potential for stream depletion by wells distant from the underflow zone(s) is great. Study results suggest that the underflow criterion will not be effective as a provision to prevent stream depletion of the Colorado River as the floodplain undergoes agricultural and urban development.