ESTIMATING GROUNDWATER AVAILABILITY AND LAND-SURFACE SUBSIDENCE IN THE COASTAL LOWLANDS AQUIFER SYSTEM USING A MODFLOW 6 MODEL AND UNCERTAINTY ANALYSIS (Invited Presentation)

The Coastal Lowlands Aquifer System (CLAS) is a principal aquifer of the United States that underlies the coastal regions of Texas, Louisiana, Mississippi, Alabama, and the panhandle region of Florida. In 2016, the U.S. Geological Survey began a 5-year study focused on understanding groundwater availability, water budgets, land-surface subsidence, and the value of the existing monitoring-well network in the CLAS. Models of the CLAS have been developed in the past on both regional and local scales. Aquifer property and water-use information from these models are being used either directly or to form the prior information for a new MODFLOW 6 model. The new model is based on the existing understanding of the hydrostratigraphy of the Chicot, Evangeline, Jasper, and Catahoula units across the study area. This understanding is based on hydrostratigraphic framework analysis by Intera in Texas, USGS datasets in Louisiana, and other datasets that have been published since the previous Regional Aquifer System Analysis (RASA) investigations of the 1990s. A new MODFLOW 6 model is in development within an uncertainty analysis framework to improve understanding of the groundwater system and make predictions of quantities of interest (QoIs). Regional QoIs include forecasted water levels, land-surface subsidence, base flow in streams, and water-budget components. QoIs will be estimated using two types of uncertainty quantification (UQ): First-Order Second-Moment (FOSM) analysis (also known as linear, or Bayes, analysis) and non-linear estimation using an iterative ensemble smoother algorithm in PEST++ (or “pestpp-ies”). A beginning posterior ensemble of model parameters and predictions of QoIs will be developed using FOSM analysis as part of the USGS PEST++ and pyEMU suite of software. The UQ is being used to guide the next steps in model development by quantifying changes in predictive uncertainty resulting from a specific model update. The FOSM posterior ensemble will be used in tandem with the ensemble from pestpp-ies to guide model development, further parameterization changes, and observation processing. QoIs will be evaluated for baseline conditions as well as different combinations of anthropogenic and climate scenarios.