Hai V. Pham, Department of Civil and Environmental Engineering, Louisiana State University, Baton Rouge, LA 70803: hpham28@lsu.edu; Frank T.-C. Tsai, Department of Civil and Environmental Engineering, Louisiana State University, Baton Rouge, LA 70803: ftsai@lsu.edu
This study develops a high-resolution three-dimensional groundwater model to simulate saltwater encroachment due to the massive pumping from aquifers under the Baton Rouge area. The model covers ten aquifers that extend from the “400-foot” sand to the “2800-foot” sand. We also include in the model 161 pumping wells, the connector well and the scavenger wells for the “1,500-foot” sand, the Baton Rouge (BR) fault, and the Denham Springs-Scotlandville (DSS) fault.
The groundwater model is developed based on MODFLOW-2005. We first adopted indicator kriging to construct fine vertical discretization geological architecture from 583 wireline well logs. Second, we introduced an upscaling procedure to convert the geological architecture into a MODFLOW grid of 93 rows, 137 columns, and 162 layers. Each grid cell size is 200 m by 200 m with an average thickness of 7.01 m. The simulation period was from January 2010 to December 2014 using 61 monthly stress periods. We used 713 groundwater head records from 43 USGS observation wells to estimate the initial condition, the boundary conditions and model parameters. Model parameters to be estimated include hydraulic conductivity, specific storage, and fault permeability. Final, we implemented a parallel version of the Covariance Matrix Adaptation-Evolution Strategy (CMA-ES) in LSU supercomputers to calibrate the model and estimate model parameters. The objection function to be minimized is the root mean square error of the simulated to the observed groundwater heads.
The parallel CMA-ES overcomes the computational difficulty and significantly reduces computational time. High resolution MODFLOW grid matched well complex geological architecture and resulted in good model calibration. The study found that the BR fault and the DSS fault are low-permeability faults that restrict horizontal groundwater flow. Sand displacements at the Baton Rouge fault mainly control the encroachment of saltwater.
Key words: MODFLOW, model calibration, CMA-ES, high performance computing, optimization