Integrating urban recharge uncertainty into standard groundwater modeling practice: A case study on water main break predictions for the Barton Springs segment of the Edwards Aquifer, Austin, Texas

Abstract

Groundwater models serve as integral tools for understanding flow processes and informing stakeholders and policy makers in management decisions. Historically, these models tended toward a deterministic nature, relying on historical data to predict and inform future decisions based on model outputs. This research works toward developing a stochastic method of modeling recharge inputs from pipe main break predictions in an existing groundwater model, which subsequently generates desired outputs incorporating future uncertainty rather than deterministic data. The case study for this research is the Barton Springs segment of the Edwards Aquifer near Austin, Texas. Researchers and water resource professionals have modeled the Edwards Aquifer for decades due to its high water quality, fragile ecosystem, and stakeholder interest. The original case study and model that this research builds upon was developed as a co-design problem with regional stakeholders; the model outcomes are generated specifically for communication with policy makers and managers. Recently, research in the Barton Springs segment demonstrated a significant contribution of urban, or anthropogenic, recharge to the aquifer, particularly during dry periods, using deterministic data sets. Due to social and ecological importance of urban water loss to recharge, this study develops an evaluation method to help predicted pipe breaks and their related recharge contribution within the Barton Springs segment of the Edwards Aquifer. To benefit groundwater management decision processes, the performance measures captured in the model results, such as springflow, head levels, storage, and others, were determined by previous work in elicitation of problem framing to determine stakeholder interests and concerns. Through additional modeling processes, this study compares the results of the previous deterministic model and the stochastic model to determine gains to stakeholder knowledge.