publications > scientific investigations report > surface-water and ground-water interactions > comparison of results from different methods of estimating recharge and discharge

Each of the new techniques for quantifying recharge and discharge in the Everglades that has been discussed in this report has advantages and disadvantages. The first new technique was Darcy-flux calculations. Darcy-flux calculations demonstrated the changing direction of recharge and discharge over timescales ranging from days to weeks, to seasons, and beyond. Another result was that time-averaged values of recharge and discharge in the interior wetlands were much larger than could be explained by a simple hydrogeologic model for WCA-2A (that only could account for recharge and discharge associated with ground-water flow beneath the levees located at the boundaries of WCA-2A). The main disadvantage of the Darcy-flux approach was the significant uncertainty associated with estimating vertical hydraulic gradients and vertical hydraulic conductivities in peat. An additional disadvantage was the high cost of installation of wells and installation and maintenance of surface-water level recorders at multiple sites. Although the installation of research wells may not be justified strictly for hydraulic measurements, it should be noted that the same wells serve the dual purpose of providing ground-water samples for a broad suite of water-quality parameters, including constituents that are potentially useful as environmental solute tracers (e.g. radium and tritium). Because of the uncertainties in the measurements, the resulting estimates of recharge and discharge were considered to be relatively uncertain until independent confirmation was available.

A second level of evaluating methods is a comparison between results of different methods. Seepage meters performed well in some parts of the Everglades (Sonenshein, 2001; Choi and Harvey, 2000), but as explained in Harvey and others (2000), the relative inaccessibility of research sites in WCA-2A, in addition to differences in the hydraulic properties of the peat compared with other parts of the Everglades (Harvey and others, 2004), meant that seepage meters were not an option for the present investigation in WCA-2A. The next available method was modeling vertical transport of short-lived radium isotopes in peat (Krest and Harvey, 2003). Radium modeling in peat averages recharge and discharge fluxes over timescales similar to the decay constants of the isotopes (days to weeks). A disadvantage of this technique was the large amount of aqueous and solid-phase analyses for radium isotopes that were necessary to achieve one-time estimates of recharge or discharge at just a few sites. There is ongoing work, however, to simplify these analyses. Nevertheless, a direct comparison of the resulting water fluxes with Darcy-flux calculations was possible using the radium technique, and was discussed in detail in a previous section of this report. A final comparison involved results based on modeling of tritium transport and decay in the limestone/sand aquifer beneath the Everglades (detailed in a previous section of the present report). This approach provided estimates of recharge and discharge that are averaged over decades due to the longer timescale of tritium decay. At a minimum the technique requires several wells, with short screens emplaced at different levels of the aquifer to characterize the vertical distribution of tritium. Another disadvantage of tritium modeling is that uncertainties of modeled ages of ground water increase as the amount of time increases since the bomb pulse of the late 1950s and 1960s. However, the ability to quantify and interpret ³H/³He has improved the reliability of the overall method.

The basis of comparison between methods is the magnitude of the vertical exchange flux (q_E, defined earlier in this report), a parameter closely related to recharge and discharge in a system where both recharge and discharge occur as a flux across the same interface (wetland ground surface). In a wetland system such as the Everglades, the vertical exchange flux is generally a good estimate of both recharge and discharge because these quantities are linked both spatially or temporally. In other words, the exchange flux represents water that recharges the subsurface and then later discharges back to surface water. When recharge and discharge are averaged over relatively long spatial and temporal timescales in a very flat setting such as the Everglades, the difference between recharge and discharge (the "net" vertical flux) is usually a small quantity involving only that part of recharge or discharge which results from hydraulic communication with areas outside the WCA-2A basin. In WCA-2A there is evidence for a small net vertical recharge of water (Harvey and others, 2002), and that flux is considerably smaller than "total" recharge and discharge fluxes in WCA-2A. The estimate of net recharge that is included in table 2 is assumed to account for the small portion of the total recharge in WCA-2A that flows beneath levees out of the WCA-2A domain. Because net recharge is so small, the vertical exchange fluxes in table 10 can be interpreted as temporally and spatially averaged estimates of both recharge and discharge in WCA-2A (table 10).

Temporally and spatially averaged vertical exchange fluxes for the central part of WCA-2A were 0.2 cm per day for the Darcy-flux approach, 0.9 cm per day for the radium modeling approach in peat pore water, and 0.01 cm per day for the tritium modeling approach in ground water. A related study introduced a salt tracer (KBr) by injecting a tracer solution into surface water with arrival of the tracer measured at shallow depths in peat ranging between 1.5 and 15 cm deep (Harvey and others, 2005). That investigation determined an exchange flux of 3 cm per day. All of those estimates can be contrasted with results of the SFWMM, for which the time-averaged estimates of vertical exchange ranged between 0.03 and 0.1 cm per day depending on the particular simulation (table 10). Considered together, the estimates of exchange flux reported above range over two orders of magnitude. The remainder of this section discusses possible reasons for such a broad range of estimates.

Central to a comparison of results from different measurements of exchange flux is recognizing that each technique of vertical exchange has a unique timescale of averaging, depending on frequency and location (that is, depth) of measurements, and whether the measurements are made in peat pore water or in underlying ground water. Figure 23 summarizes the results discussed above. The respective averaging depths of the tritium modeling, Darcy-flux calculations, radium modeling, and Br tracer experiments are 5.25, 1, 0.66, and 0.03 m, respectively. These depths are illustrated as the flux planes shown in figure 23. Part of the difference in estimates of exchange flux is also attributable to factors such as the decay rate of the tracers used (days to tens of days for the short-lived radium isotopes and 12.4 years for tritium). As a result, each technique is sensitive to a different temporal scale of variation in water exchange. For example, the technique based on modeling ²²⁴Ra in peat pore water produced one of the highest estimates of exchange flux. That estimate appears to have been primarily sensitive to rapid reversals in vertical flow that drive short-term (daily to weekly) reversals in the direction of vertical exchange. Those short-term reversals are only effective in causing water exchanges between surface water and peat pore water. On the other hand, because of the longer half-life of tritium and because the sampling was done in the sand and limestone aquifer and not in peat pore water, results from modeling tritium transport are sensitive only to seasonal and longer-term components of exchange, which ignores the relatively high-velocity but short term components of exchange that are only effective in exchanging surface water with peat pore water. As a result, the estimate of vertical exchange produced by tritium modeling was two orders of magnitude lower than the estimate from ²²⁴Ra modeling method in peat pore water.

Both the Darcy-flux calculations and SFWMM results are based on hydraulic computations using daily averaged measurements of water level, which makes the methods potentially sensitive to weekly to monthly timescale fluctuations associated with relatively frequent reversals in the direction of vertical flow between recharge and discharge (Harvey and others, 2004). Both the Darcy-flux and SFWMM measurement approaches depend on hydraulic head, which should respond very quickly to changing surface-water levels and energy potentials in ground water that drive flow into or out of the sediment. The relevant criteria involve calculations of the rate of pressure propagation through peat, and these calculations suggest that the equilibration time for hydraulic head in ground water following a change in surface-water head is on the order of minutes (Harvey and others, 2004). Spatial and temporal averaging of the SFWMM reduces the sensitivity of that modeling approach, because SFWMM results are typically presented as monthly or annual averages of daily estimates. When run on a decadal timescale (1979-1990) the SFWMM produced a smaller time-averaged exchange flux (0.03 cm per day) compared to when the model was run over a shorter (5-year, 1991-1995) time period (0.1 cm per day) (table 10). In addition, SFWMM calculations are made on a 2- x 2-mi grid, which smoothes hydraulic head values by averaging peaks and troughs in hydraulic heads associated with the movement through the wetlands of water pulses released by water-control structures. SFWMM results are therefore less sensitive to weekly to monthly timescale interactions between surface water and peat pore water and more sensitive to longer-timescale patterns of interaction with ground water of the sand and limestone aquifer. This dependence of the estimated exchange flux on spatial and temporal averaging scales is not new. Instead, it is consistent with results from an investigation in Wisconsin using the USGS ground-water flow model MODFLOW (Stoertz and Bradbury, 1989). Stoertz and Bradbury (1989) found that greater spatial and temporal averaging of hydraulic heads decreased the estimates of recharge and discharge. The order-of-magnitude differences in the exchange fluxes determined by using independent methods in WCA-2A do not imply that any particular estimate is "wrong." Instead, each estimate provides information about the magnitude of recharge and discharge that is relevant to a particular timescale of fluctuating directions of recharge and discharge, and depth of water exchange in the subsurface. For example, the weekly to monthly timescale fluctuations in recharge and discharge that are detected by bromide tracer, radium modeling, and Darcy-flux calculations are mainly informative about exchange of surface water with peat pore water. Instead, the effectively longer-term averaging accomplished by modeling tritium in ground water, or by spatially and temporally avering hydraulic cacluations in the SFWMM, produces techniques that are only sensitive to exchange between surface water and deeper ground water in the aquifer underlying the peat and wetland.

Therefore, there is no single measure of recharge and discharge that can be applied for all research purposes in the central Everglades. However, some estimates are more appropriate than others to meet a particular objective. Methods that are sensitive to the short-term fluctuations that mainly cause exchange between surface water and pore water in the peat are most appropriate for models focused on water quality and ecology. For example, the higher values of recharge and discharge (table 10) are more appropriate for use with the Everglades Landscape Model (ELM) (Fitz and others, in press; also see South Florida Water Management District, 2002) or with Dynamic Model for phosphorus in STAs (DMSTA) (Kadlec and Walker, 1999). In those water quality simulation the effects of vertical water exchange between surface water and peat pore water (and associated solute transport and biogeochemical reactions) are highly relevant to water quality. In contrast, water-balance investigations that are averaged over large spatial scales or longer timescales (for example, simulations with SFWMM) probably can make better use of longer-term estimates (which are lower numbers) determined by ground-water measurements and modeling of tritium as an independent check on recharge and discharge fluxes used in the models. In those investigations, the net exchange flux is more relevant than the larger exchange fluxes that mainly affect water quality. An exception may be in regional simulations built upon the parameters of the SFWMM that also consider water quality. One example is modeling the fate of freshwater in the Everglades which is slowly increasing in its contents of total dissolved solids. Part of the problem is with the presence and operation of canals, which pierce the relatively low conductivity peat substrate to gain a direct hydraulic connection with the underlying aquifer which contains high concentrations of relict sea salt in its lower two thirds. Abrupt water level changes in the wetland basins of the Everglades and in canals that are the result of water management operations can increase the vertical hydraulic gradients that cause upward mixing of those salts. In most regional simulations of hydrology in the Everglades, the influence of vertical mixing on upward movement of salt into shallow ground water and surface water has not been considered.