Edwards Aquifer Drought Model: Brackish Water, Springs and Uncertainty (USGS)

The Edwards aquifer's recharge, freshwater, brackish-water transition and saline zones. USGS.

Key findings

• Uncertainty analysis showed how reliable the model's inputs and results are.
• Brackish water is not expected to move appreciably toward production wells near the transition zone if the drought of record (1950–56) recurs — and the model is reliable for that prediction.
• The model is not reliable for predicting water levels at Bexar County index well J-17 or flows at Comal and San Marcos Springs.
• Many aquifer properties, and recharge, remain highly uncertain even after calibration.

The aquifer

The Edwards aquifer supplies residents, businesses and ecosystems in south-central Texas. It is a complex karst system — soluble limestone and dolomite in faulted, fractured Cretaceous rock that can carry enormous amounts of water fast. It has three water-quality zones: freshwater, a brackish transition zone to the south, and saline water farther downdip.

Sustained droughts in 2011–15 raised the worry that a repeat of the 7-year drought of 1950–56 could push the transition zone toward the freshwater zone — making wells near it pump saltier water — and cut flows at Comal and San Marcos Springs.

The model

In 2010, the USGS and the San Antonio Water System began a study using SEAWAT, a USGS variable-density flow and transport code that better simulates how dissolved solids affect water density and flow. It divided the aquifer into eight model layers — far more than earlier models — to represent its structure.

1. History matching: calibrating to real water levels, spring flows and dissolved-solids data for 1999–2009, adjusting thousands of parameters with the PEST software. The fit was good, with little bias.
2. Prediction: rerunning 1950–56 rainfall with recent (1999–2009) pumping — much heavier than in the 1950s, when far fewer people lived there — to predict dissolved-solids change at 25 production wells, total flow from the two springs, and the level at well J-17.

What uncertainty analysis showed

Parameters: comparing uncertainty before and after calibration shows what the data taught the model. Many hydraulic properties stayed highly uncertain, and recharge — the driver of flow — was barely constrained at all. Data are rich for the freshwater zone but sparse for the transition and saline zones.

Predictions:

• Wells near the transition zone: even at the upper 95-percent bound, only a minimal increase in dissolved solids is expected during a repeat drought — and this prediction is well informed by the data, so it is reliable.
• Springs and well J-17: the upper bounds were 10 times larger than the calibrated prediction, so these predictions are not reliable, even though the model reproduced past spring flows and levels well. They depend on many uncertain properties across a large upgradient area, compounded by uncertain recharge.

Average annual groundwater withdrawals in 1999–2009 and the production wells studied near the transition zone. USGS.

Sources

Based on New Insights into the Edwards Aquifer—Brackish-Water Simulation, Drought, and the Role of Uncertainty Analysis, by Linzy K. Foster and Jeremy T. White, USGS Fact Sheet 2016–3002, Texas Water Science Center, U.S. Geological Survey (DOI), summarizing Brakefield and others (2015), USGS Scientific Investigations Report 2015–5081; a work of the United States government in the public domain. The maps come from the fact sheet's PDF; two photographs used by the USGS with a private photographer's permission are left out.