GSA Annual Meeting in Phoenix, Arizona, USA - 2019

Paper No. 250-11
Presentation Time: 11:00 AM

GROUNDWATER MODELS AS PREDICTIVE TOOLS: USING MODEL UNCERTAINTY TO GUIDE SUSTAINABLE MANAGEMENT STRATEGIES IN BASINS IMPACTED BY COUPLED STRESSES


QUINLAN, Peter1, TARTAKOVSKY, Daniel2, UM, Kimoon2, YU, Yikyung3 and JONES, Trevor1, (1)Dudek, Encinitas, CA 92024, (2)Energy Resources Engineering, Stanford, Stanford, CA 94305, (3)University of California, San Diego, La Jolla, CA 92093

Large-scale numerical models provide the ability to quantitatively predict the response of a groundwater basin to anthropogenic and natural stresses that threaten groundwater availability. Classic deterministic modeling approaches provide unique solutions to the regional competition between groundwater replenishment and depletion. However, because the exact properties of the subsurface cannot be known over the entire model extent, predictions of future groundwater availability using these tools are uncertain and non-unique. Understanding the principal components that drive model uncertainty is critical to caveat estimates of key metrics, such as the basin’s sustainable yield, that guide effective groundwater management strategies. In this study, we present results from a global sensitivity analysis of a large-scale groundwater model used to quantify the sustainable yield in a basin affected by ongoing flux of seawater into the regional aquifer systems. To perform this sensitivity analysis, we ran a set of Stratified Monte Carlo simulations and used the ANalysis Of VAriance (ANOVA) method to map uncertainty in key aquifer properties to variance in the model performance (e.g. RMSE) and model-predicted seawater flux. We then propagated the uncertainty estimated from the Monte Carlo simulations into a simple surrogate model that related groundwater production and seawater flux under future scenarios to generate confidence intervals for sustainable yield in the principal aquifer systems. Results from this sensitivity analysis showed that the sustainable yield of the basin varied by nearly 25%. Importantly, over 20% of the uncertainty in the sustainable yield was driven by hydraulic properties in artificial recharge zones located far from the ocean boundary. This key finding, which was elucidated through the application of the ANOVA methodology, helps focus management strategies aimed at minimizing the threat of groundwater resource degradation as a result ongoing seawater intrusion into the basin.