INTRODUCTION

Recognition of the importance of wetlands to wildlife and to water quality has accelerated research on the role of hydrology as a driving force for wetland processes (Good et al. 1978, Ivanov 1981, Carter 1986, Hemond and Benoit 1988). Hydrologic fluxes in lakes and wetlands often include surface inflow and outflow, precipitation, evapotranspiration, and exchange between surface water and ground water. Of these, the interaction between surface water and ground water is generally recognized as the most difficult flux to estimate because it is often relatively small compared to other components of the water budget (Kadlec 1983, LaBaugh 1986). However, even a small flux between surface water and ground water might be very crucial to biogeochemical processes due to the typical sharp contrast between chemistry of surface water and ground water in wetlands (Howes et al. 1996). As a result, ground-water and surface-water interactions have the potential to vastly alter redox-sensitive solutes or contaminant mass balances in wetlands.

Exchange of water between ground water and surface water is difficult to quantify in wetlands because it cannot usually be measured directly. Rather, ground-water exchange fluxes must be inferred from related hydrologic and chemical measurements. In the past, wetland/ground-water exchange fluxes sometimes were assumed to be insignificant, or estimated as a residual term in wetland surface-water balances, or determined by Darcy flux calculations using relatively limited data sets. Progress in quantifying the water balance of lake systems provided useful guidance for wetland research (Crowe and Schwartz 1985, Stauffer 1985, Krabbenhoft and Anderson 1986). More recent approaches in wetlands used water or solute mass balances in the subsurface (Harvey and Odum 1990, Nuttle and Harvey 1995, Hunt et al. 1996). Mass balance calculations in the wetland subsurface have the possible advantage of improved accuracy over surface-water mass balances, but the results from a particular location cannot be easily extrapolated to other areas of the wetland. Ground-water-flow modeling offers a means to extrapolate site-specific results to larger wetland areas (Hunt et al. 1996, Guardo and Prymas 1998). A drawback is that reliable modeling often requires extensive installation of wells, which still may not be enough to characterize aquifer heterogeneity adequately (Krabbenhoft and Anderson 1986). Also, hydraulic conductivity estimates in peat are sometimes unreliable (Ingram et al. 1974, Hemond and Goldman 1985, Nuttle and Harvey 1995). Common to all methods is the difficulty of quantifying the uncertainties in ground-water discharge and recharge estimates.

Figure 1. (a) Palm Beach County and vicinity showing location of ENR relative to Everglades Agricultural Area (EAA) and Water Conservation Area (WCAs). (b) Location map illustrating hydrologic and chemical monitoring sites in Everglades Nutrient Removal (ENR) project [click on images above for larger version]

The present study was undertaken in the Everglades Nutrient Removal area, a 1544-hectare wetland constructed in 1994 to test the removal efficiency for nutrients in agricultural runoff. The ENR is located in south Florida, USA (26° 38'N, 80° 25'W) between the eastern edge of the Everglades Agricultural Area (EAA) and the western edge of Everglades Water Conservation Area 1 (WCA-1) (Figure 1a). The ENR area is primarily covered by peat soils with very small topographic relief and an average ground elevation of 3 m NGVD. The peat layer is underlain by layers of permeable sand and limestone that comprise the surficial aquifer. The ENR project area is mainly covered with either monospecific stands of cattails, mixtures of many macrophytes including cattails, or open water with submerged vegetation.

Like other wetlands constructed for water treatment, ENR is often subjected to higher surface-water flows and more frequent fluctuations in water level compared with natural wetlands. In addition, the supply waters for treatment wetlands can vary substantially, representing vastly different source waters with different chemistry. Because of these unstable hydrologic and chemical conditions, time-varying estimates of ground-water fluxes were essential to understand the ground-water interactions in the ENR.

Ground-water interactions at the ENR were examined in several previous investigations (Hutcheon Engineers 1996, Guardo and Prymas 1998). Those authors estimated wetland/ground-water exchange using water-level data, hydraulic conductivity estimates, and ground-water-flow modeling. Guardo (1999) and Moustafa (1999) made use of those results in their water balance and nutrient balance studies at ENR. However, neither of previous ground-water investigations simultaneously estimated discharge and recharge, and no uncertainties were estimated. As a result, there is still some question whether previous estimates of ground-water interactions are reliable or not. Due to the above considerations, reliable estimates of ground-water discharge and recharge are still needed at ENR.

In this study, we estimated time-varying ground-water discharge and recharge in the ENR using a coupled water and solute (chloride) mass balance approach. We estimated the error in our analysis, and we examined individual components in mass balance to identify the key contributors to uncertainty. Also, we compared our results with other approaches used at ENR, including our past results using seepage meters as well as ground-water-flow modeling results from past studies. Through comparison, we identified the most reliable estimates of ground-water interactions and applied that knowledge to develop an initial understanding of how the design and operation of the ENR increases interactions between ground water and surface water.