Twenty five years of operational experience using GW modeling to evaluate sustainable use of coastal aquifers for major water supplies

The sustainable development and management of groundwater resources in coastal aquifers is complex and, historically, challenging to accomplish. Groundwater models play an essential role in addressing these complexities and providing the basis for planning future sustainable development. For more than 25 years, the authors have applied three-dimensional groundwater models to manage large scale coastal aquifers.

The paper will present case studies demonstrating the application of groundwater models to evaluate conditions in complex coastal environments and to develop sustainable groundwater management strategies. These studies include Long Island, a sole source aquifer system in New York serving nearly 3 million people; aquifers in Southern California where injection barriers are used to prevent saltwater intrusion; and Savannah, Georgia in the southeastern US, where concentrated groundwater pumping has contributed to saltwater intrusion at a nearby resort island, and planning is underway to ensure a sustainable groundwater supply to both local industries and municipalities.

Rapid urbanization in coastal cities worldwide has given cause for municipalities to consider increasing groundwater output to address potential water supply shortfalls, provide lower salinity water to reverse osmosis treatment facilities, and improve the resilience and security of water supplies. The sustainable development and management

• Several aquifer layers than can provide water

• Multiple saltwater intrusion pathways

• Transient by nature

• Recharge areas far inland

• Offshore storage of fresh water

• May not yet have reached equilibrium

• Large study area, potentially very deep

• Lack of offshore data

• Multiple pumping wells

• Seasonal demand and recharge patterns

of these groundwater resources is complex (see inset) and challenging to accomplish, with instances in recent history of overuse and depletion. Groundwater modeling can be used to address these complexities and provide the basis for future sustainable development of the resource. For more than 25 years, the authors have applied three-dimensional groundwater models to manage large scale coastal aquifers. Three examples are presented in this paper as case studies.

In the case studies presented DYNSYSTEM was used. This (www.dynsystem.com) is a suite of finite element codes that simulates three-dimensional groundwater flow of uniform or variable density fluids, contaminant transport, multiphase flow, and complex biodegradation and transformations (Fitzgerald et al. 2001). The core of the system is the fully three-dimensional flow code, DYNFLOW, which evolved from the MIT-developed AQUIFEM code in the 1970s. DYNFLOW is a comprehensive finite element code, proven during hundreds of practical applications, that simulates fully three-dimensional multi-layer aquifer systems, and allows a wide range of stresses and boundary conditions to be applied.

The companion DYNTRACK code is a random-walk based, mass-transport code used to simulate the migration of dissolved contaminants. It is also fully three-dimensional, and simulates advection, dispersion, retardation, and decay. It is fully integrated with DYNFLOW, and has been applied by the authors at numerous hazardous waste sites, and for studies of development and protection of groundwater resources.

Variable density flow regimes can be simulated using DYNCFT, which is a coupled flow and transport code. It is a linkage of DYNFLOW and DYNTRACK, and incorporates the many years of practical experience with both of these codes. Like the others, it is fully three-dimensional, and can address multiple intrusion zones.

To address regional-scale saltwater intrusion, another code, DYNSWIM, may also be appropriate. This is a sharp-interface code, developed directly from DYNFLOW. The code is also fully three-dimensional and can evaluate the potential migration of sharp intrusion zones into a complex aquifer system. The migration of the intrusion can be thus vertical (such as through ‘holes’ in clay layers) as well as horizontal. DYNSWIM can also be applied in localized areas where DYNFLOW is being used for overall aquifer evaluation. In practice, the sharp-interface algorithm in DYNSWIM has been found exceptionally useful in evaluating potential intrusion into over-stressed aquifer systems, and has clearly indicated the root causes of several unexplained saltwater impacts on near-coast well fields (Maimone 2001).

In Southern California, Savannah, Georgia, and Long Island, New York (Figure 1), groundwater pumping was relatively limited prior to 1900. However, rapid development in groundwater use in the early to mid-20th Century in each of these communities produced damaging intrusion of saltwater into aquifers and significant changes in groundwater management moving forward. Case studies from each of these communities are presented below to illustrate specific challenges and opportunities in evaluating and planning for sustainable groundwater use in coastal aquifers.

Historically, intensive pumping in on Long Island, New York (Figure 2), caused significant drawdown of groundwater levels and damaging intrusion of saltwater into the groundwater system. This intrusion has been the driving force behind significant changes in public supply pumping over the past 100 years on the island. Saltwater encroachment at water supply wells in the New York City borough of Brooklyn is documented as early as the 1920s, and concentrations as high as 600 mg/L were observed near the center of the county in the 1940s (Figure 3). In 1947, all public water supply pumping in Brooklyn stopped because of widespread saltwater intrusion. With much of the supply pumping subsequently shifting to Queens (the easternmost New York City borough), groundwater elevations in Queens decreased (Figure 2), and significant saltwater impacts at wells in southern Queens occurred in the 1960s (Figure 3). Increases in pumping from Nassau County also caused saltwater intrusion to occur in the southwest part of the county, and on north coast peninsulas.

Regional groundwater flow models utilizing DYNFLOW and DYNSWIM have provided the basis for water management planning, saltwater intrusion studies, and well water quality protection studies for the past 25 years, for aquifers serving as the sole source of drinking water for over three million Long Island residents. The models cover the entire island, extend over 10 kilometers offshore, incorporate multiple aquifers, simulate fresh and saline water, and include time varying pumping from thousands of supply, remediation, dewatering, and industrial wells (Gamache et al. 2008). To evaluate the feasibility of potential future water supply projects, a 100-year (1905–2005) historical transient simulation was developed to demonstrate that the models can reasonably represent the significant drawdown and subsequent recovery of piezometric heads resulting from pumping activities (Figure 2) as well as the regional saltwater intrusion that occurred during the 20th century.

The primary water bearing units in western Long Island are the Lloyd (confined), Magothy/Jameco (semi-confined), and Upper Glacial (unconfined) aquifers. Figure 2 shows a cross section of model stratigraphy through Queens, with the present-day simulated saltwater wedges shown. Along the southern coast a complex, triple wedge interface is represented, with saltwater present onshore in the shallower Upper Glacial and Magothy/Jameco aquifers, and at an undetermined distance offshore in the confined Lloyd aquifer, where the 200-foot thick Raritan clay inhibits discharge of fresh water and slows the intrusion of saltwater into the Lloyd aquifer (Fitzgerald & Maimone 1992). Pre-development saltwater wedge positions were estimated for the Lloyd aquifer and simulated to approximate equilibrium conditions in the Upper Glacial and Magothy/Jameco aquifers.

Model simulation results were evaluated by comparing the extent of simulated saltwater intrusion with observed chloride concentrations for periods when historical saltwater encroachment occurred.

Figure 4 shows the simulated position of the saltwater interface in the Upper Glacial aquifer in 1949. The simulated saltwater wedges are shown as contoured saltwater thicknesses in each aquifer. Observed maximum groundwater chloride concentrations for this time period are also shown as color coded symbols at well locations. Saltwater encroachment in Brooklyn was observed at wells in both the Upper Glacial and Magothy/Jameco aquifers. The simulated 1949 extent of saltwater intrusion shown is consistent with the locations of observed elevated chloride concentrations from 1940 to 1950.

Figure 5 shows the simulated position of the saltwater interface in the Upper Glacial aquifer in 1969. Observed maximum groundwater chloride concentrations are also presented for that time. Chloride concentrations in Upper Glacial wells in the Woodhaven franchise area (southwestern Queens) increased from the 1950s to the mid-1970s because of increased pumping, almost entirely from the Upper Glacial aquifer (Buxton & Shernoff 1999). The simulated extent of saltwater intrusion in 1969 is consistent with observed groundwater chloride concentrations and reports of saltwater intrusion in the Woodhaven franchise area of southwestern Queens.

The groundwater flow models developed for Long Island have formed the basis for many studies of water supply alternatives, including extraction and treatment of brackish groundwater. Brackish groundwater pumping was simulated using DYNCFT to evaluate whether there was a potential for hydraulic interference with existing or proposed pumping further inland, and to evaluate the likely groundwater chloride concentrations of the pumped groundwater. Simulations indicated no interference with inland pumping centers, and no additional threat of saltwater intrusion at the pumping rates evaluated. Chloride concentrations in the pumped brackish groundwater from hypothetical wells along the southern coast of Long Island ranged from approximately 1,000–7,000 mg/l, economically treatable using present-day reverse osmosis treatment technology.

Decades of industrial pumping in the Savannah, Georgia, region in the mid to late 20th century resulted in groundwater elevations more than 100 feet below sea level that contributed to saltwater intrusion impacts to water supply wells on Hilton Head Island, South Carolina, approximately 30 km from Savannah.

The authors developed a coupled groundwater flow and saltwater transport model to study historical saltwater migration in the Savannah and Hilton Head area. The modeling effort built on an earlier modeling analysis the authors conducted for the US Army Corps of Engineers of the potential saltwater intrusion impacts from proposed harbor dredging in Savannah Harbor, Georgia. These two studies, and others that have been conducted in the region, demonstrate the benefits of having a regional model as a planning and analysis tool, to support local decisions and understanding of long-term groundwater trends.

The model used in both studies covered the entire coastal Georgia region, and extended into portions of Florida and South Carolina, with focused grid detail in the Savannah and Hilton Head areas (Figure 6). The model includes multiple layers representing the Coastal Plain Surficial and Floridan aquifer system, and an intervening Miocene confining layer that, in the focus area, varies in thickness and permeability (Figure 7). Groundwater and saltwater migration were simulated using the variable density coupled flow and transport model, DYNCFT.

The model was applied to simulate transient groundwater flow for the period from 1915 to 2007, to study the historical migration of saltwater from the surficial aquifer into the Upper Floridan aquifer at locations north of Hilton Head, where the Miocene confining layer thins and is more permeable. The model was calibrated to groundwater levels, observed response to tidal fluctuations, a local pumping test, and specific conductivity data.

The transient groundwater flow model replicates the decline in Upper Floridan water level elevations in response to increased industrial pumping, and the slight rebound in water levels once groundwater pumping decreased. This decline in groundwater elevations contributed to the saltwater intrusion observed beneath Hilton Head Island. A historical simulation of saltwater intrusion into the Upper Floridan aquifer, and migration in the Upper Floridan aquifer beneath Hilton Head Island and vicinity, was completed to demonstrate that model-simulated saltwater migration was reasonably consistent with observed conditions at monitoring wells (Figure 8). The model matches measured regional and temporal historical trends in both water level elevations and groundwater chloride concentrations.

Once calibrated, the coupled flow and transport model was used to study the relative impacts of Savannah and Hilton Head pumping on historical saltwater migration, and to evaluate the impact of potential pumping reductions and hydraulic barriers on the projected saltwater transport beneath Hilton Head. Model simulation results are being used by an interstate committee tasked with developing options for managing saltwater intrusion to ensure groundwater supply to local industries and municipalities.

Rapid development of groundwater use in southern California in the early and mid-20th century caused coastal groundwater levels to fall well below sea level, by as much as 30 m, and induced serious impacts on water quality by saltwater encroachment (Liles et al. 2001). Hydraulic injection barriers were constructed to protect the potable groundwater supply in the Los Angeles and Orange County groundwater basins. Three-dimensional groundwater modeling was used to evaluate the effectiveness and design enhancements for injection barriers at Dominguez Gap, Talbert Gap, and West Coast Basin.

For the Dominguez Gap study, the model domain incorporated both on and offshore (over 3 km from the coast) areas and utilized the ability of finite element grid meshing to finely represent the 6-km line of 30 injection wells that make up the barrier. Large pumping and injection flows were incorporated into the 10-layer flow and transport model using DYNFLOW (flow model) and DYNTRACK (transport) to quantitatively assess alternatives for improving the barrier's effectiveness for mitigating saltwater intrusion. Groundwater flow and chloride transport simulations identified weak points in the barriers that were allowing saltwater to pass inland. In this case, density effects were not significant, and the combination of DYNFLOW and DYNTRACK, run in an ‘uncoupled’ mode, has proven to be very effective and cost-efficient, in assessing aquifer conditions and management strategies.

The flow and transport models were also used to evaluate the feasibility of using reclaimed water for barrier injection at each of the three barriers. Recycled water travel times and dilution trends were evaluated for a series of injection alternatives, which ultimately guided where recycled water can be injected such that the recharged reclaimed water would remain in the subsurface for at least 1 year and constitute no more than 50% of the groundwater extracted at any domestic supply well, in accordance with local regulations. For Dominguez Gap, the recommended alternative that met these criteria utilized the western barrier wells only for recycled water and eastern wells for imported water for the first few years of operation. Thereafter, the two water types would be blended in the well header pipeline just before injection. The significant piezometric depression located in the center of the area was found to prevent sufficient inland water from recharging the coastal zone to allow a higher fraction of recycled water to be used (Thomas et al. 2001). As a result of the modeling, plans for supplemental injection wells and revised basin operations were developed.

Practical experience in New York, Savannah, and California has proven that groundwater models are vital tools for understanding the complexities of coastal aquifers. Their development and continued use over time can provide insights into sustainable system operations and help protect groundwater wells from being compromised. As coastal cities continue to expand and develop their groundwater resources, model-assisted planning will be at the forefront of the water resources solutions of the 21st century.