Relating Streamflow Depletion to Groundwater Pumpage in the Context of Uncertainty

Abstract:

Interaction between groundwater and surface-water systems is an inherently complex process, especially in the context of relating groundwater use to local- scale stream flow characteristics. An integrated hydrologic model (MODFLOW with SWR process) is being developed for the lower San Antonio River Basin in Texas to assess how water use affects the groundwater contribution to stream flow under hypothetical and observed climate scenarios. The basin traverses several dipping aquifer systems and water from these systems is relied on for residential, recreational, industrial, and agricultural uses, as well as oil and gas activities. The current (2015) understanding of interaction between the two hydrologic systems within the basin is limited, but indicates considerable variability in space and time. The modeling analysis is designed to provide improved understanding of spatial and temporal characteristics of interaction under different hydrologic conditions, such as the drought conditions experienced from 2011 through 2013.

Due to paucity of data and limited current understanding of interaction between the two hydrologic systems, model results will be inherently uncertain, making uncertainty analysis of paramount importance in the model development and use. As such, linear-based uncertainty analyses, also known as first-order, second-moment analysis, are being used to design the parameterization and objective function prior to the computationally-expensive history-matching process. For this process, we use pyEMU, an open-source python module for linear-based computer model uncertainty analysis. Results show that the prior uncertainty in model inputs yield largely uncertain local-scale interaction estimates. However, by appropriately designing the objective function and parameterization to be aligned with the focus of the modeling analysis, the posterior uncertainty of many local-scale interaction estimates can be reduced.