Assessment of Regional Management Strategies for Controlling Seawater Intrusion

Published: 10/26/2017

Recent groundwater development led to a decline in groundwater storage and consequently problems with seawater intrusion in coastal areas. The subject article assesses strategies involved in controlling this problem within the coastal basins. The two seawater intrusion control methods examined by the authors are increased injection into barrier wells and in lieu delivery of surface water. The two strategies were assessed using simulation-optimization models conducted by the authors and other researchers offering quantitative insight into the cost and benefits of multiple future scenarios and allowing for better management of the precious groundwater resources. The study area includes the Central Basin, West Coast Basin, the Hollywood, and the Santa Monica Basin. The article focuses mainly on the West Coast Basin for the optimization analysis.

A uniform difference model grid was used to develop the initial groundwater simulation model. This initial model was used to approximate conditions in water year 1971 and took into account the separate layers for the four main aquifer systems described above. Hydraulic properties presented by the California Department of Water Resources were used to better understand the flow patterns and conductance within the aquifers. Time-varying specified heads were used to represent boundary conditions between the Central Basin and the Orange County groundwater basin and the implications of this representation were thoroughly tested as part of the sensitivity analysis conducted later on. The initial model assumed steady-state and used sensible geographic boundaries for the modeled area. Other assumptions and limitations include grouping multiple aquifers into four model layers, the uncertainty of estimating urban areal discharge, and the representation of several of the boundary conditions. These are all fair assumptions considering the complexity of the study area, and the authors make a point out of mentioning them.

The simulation of future scenarios focused on the 3rd model layer, the upper San Pedro aquifer system, as it accounts for 80% of the pumping and 60% of the injections that occur. This focus means that the simulation may not be as accurately calibrated to model other aquifer systems. The model simulated water levels for the year 2000 were used as initial conditions for future scenarios. All future scenarios demonstrated lowering water levels and that seawater intrusion will continue through 2025. The first simulation of the future scenario assumed no change in pumping rates. The second simulation of future scenarios assumes an increase of the current pumping rates by 25% over 5 years. The simulation thus assumes a linear increase in pumping rates and can not account for any irregularities in the pumping rates but is more conservative than the first scenario. The simulations also assume a constant flow boundary along the Central Basin - Orange County boundary. However, this assumption was examined and determined not to differ significantly from historical fluxes in the specified head along this boundary.

The goal of running the optimization analysis on this model is to find the most cost-effective method to raise water levels. The optimization of the model was done using iterative solutions and stopping criterion of less than 0.5% change in the objective function.

The optimization model yields individual rates for each injection well and in lieu of cells which allows for the consideration of varying parameters. These parameters include temporally varying costs and water level constraints. Reduced cost is found by analyzing the change in the objective function resulting from a small change in the value of the parameters.

Given the base-case assumptions, the final optimization simulation suggests that resources would be best used towards purchase of in lieu water. The limitations and assumptions made in obtaining this result are as follows: the simulation exclusively considered hydraulics and not density-dependent transport of water, the detailed hydrostratigraphy of the aquifer systems was was not fully represented, the results are presented in terms of annual average injection, the optimization assumes temporally constant and linear coefficients, and the does not assess on an individual well-by-well basis.