RESULTS

Water Budget and Coupled Mass Balance

Our analysis of surface-water and ground-water interaction in the ENR began with estimation of the net ground-water flux (G_i - G_o) for the study period (1994-1998) using equation [3]. The magnitude of the net ground-water flux (G_i - G_o) varied from -27.5 to -1.1 ha-m/day, with an average over time of -12.0 ha-m/day (Figure 3a). Estimated net ground-water fluxes are almost entirely negative values, which means that the ground-water recharge generally exceeded ground-water discharge. Therefore, ENR loses water on a net basis due to interaction with a ground-water aquifer. First-order error analysis indicated that uncertainty in estimated net ground-water flux varied from ± 0.5 to 4.5 ha-m/day (Figure 3a), with an average over time of 2.4 ha-m/day.

On the other hand, our initial estimates of ground-water discharge derived from the combined water and solute mass balance (Eq. [5]) fluctuated greatly. In the computation of ground-water discharge, the difference, _Gi - _Go, sometimes approaches zero. During those periods, even a small relative error in either _Gi or _Go an cause ground-water discharge (G_i) to be computationally inflated. We attempted to reduce uncertainty by filtering out those two-week periods when a reliable estimate of ground-water discharge was obviously impossible. To accomplish this, we excluded cases for which the difference of _Gi and _Go was less than 15 mg/L (30% of observations). After filtering was completed, discharge estimates were still unrealistically large, ranging between -235.0 and 220.0 ha-m/day (Figure 3b) with an uncertainty averaging 54 ha-m/day and ranging between ± 20.0 and 215.0 ha-m/day (Figure 3b). We attempted to further improve our estimate of ground-water discharge by selecting an optimal time period with relatively lower uncertainty (period II in Figure 4a and b). During period II, ground-water discharge averaged 1.4 ha-m/day (Table 3) with uncertainty averaging 37.0 ha-m/day. Since the uncertainty was still large compared to the average estimate of ground-water discharge, we were forced to conclude that time-varying ground-water discharge could not be estimated by the combined mass balance alone. However, we hoped that our best estimate of average ground-water discharge (1.4 ha-m/day or 0.9 cm/day) would prove to be reliable, as shown by comparisons with other independent estimations of ground-water discharge.

Figure 3. (a) Net ground-water flux and (b) ground-water discharge into ENR estimated from combined water and solute mass balance. Upper and lower ranges of the estimations indicated by solid lines show the uncertainty [click on images above for larger version]

---

Figure 4. Uncertainty in (a) estimated ground-water discharge and (b) comparison of uncertainty in estimated ground-water discharge and chemical difference between ground-water recharge (Go) and discharge (Gi). The greater the difference between Cl concentrations in ground water (Gi) and surface water (Go), the lower the uncertainty in estimated ground-water discharge rate [click on images above for larger version]

Sources of Uncertainty

Figure 5. Contributions (%) of individual uncertainties of hydrologic and chemical measurements to the overall uncertainty in estimating (a) net ground-water exchange and (b) ground-water discharge into ENR [larger image]

What are the individual sources of uncertainty in estimating ground-water interactions by the combined mass balance approach? Sources of uncertainty were separately identified by inspecting the uncertainty of each component in the water and combined water and solute mass balances, respectively. Results showed that most uncertainty in estimating net ground-water flux (G_i - G_o) was due to the uncertainties in S_i, S_o, and V. However, uncertainty in estimating ground-water discharge was mainly controlled by chemical (Cl) variables in equation [5], such as C_o, C_t, _Go, and _Gi. The relative contributions of all variables to uncertainties are shown in Figure 5.

Which physical or chemical attributes of the ENR wetland were most important in controlling overall uncertainties? In order to answer this question, uncertainties in G_i - G_o and G_i were compared with key hydrologic and chemical variables. For example, the uncertainty in estimating net ground-water flux was compared with daily surface inflow rate and daily change in surface-water storage during the study period. Both of those comparisons show that uncertainty in estimated G_i - G_o increases with higher surface-water inflow rate and higher change in surface-water storage. These relationships indicate that it becomes increasingly difficult to estimate the net ground-water flux (G_i - G_o) with accuracy when the system is actively responding to change in surface-water pumping rate. On the other hand, the uncertainty in estimating ground-water discharge (G_i) was mainly controlled by the difference in chloride concentration between ground-water discharge and ground-water recharge (_Gi - _Go) (Figure 4a and b).

Ground-water discharge can more reliably be estimated during a subset of the 4-year period at ENR when the tracer concentration in ground water differed markedly from that in surface water. For example, the uncertainty in estimated ground-water discharge was relatively high during period I when _{Gi Go} while the uncertainty was lower during period II when _{Gi Go} or _{Gi Go}. As explained in previous section, our best estimate of ground-water discharge was derived by average data for period II that had the lowest uncertainty.

Table 3. Water-balance fluxes in ENR from coupled water-solute mass balance: 8/19/94-8/19/98.
	Description	Fluxes (ha-m/day)	Percent (%) of Inflow Pump
Inflow	Inflow pump (Si)	43.1	100
	Inflow from seepage canal (Ri)	9.4	21.9
	Precipitation (P)	6.2	14.5
	Shallow seepage inflow (Li)	1.6	3.8
	Ground-water discharge (Gi)	1.4	2.8
Outflow	Outflow pump (So)	42.5	98.6
	Outflow to seepage canal (Ro)	0.6	1.4
	Evapotranspiration (ET)	5.6	12.9
	Ground-water recharge (Go)	13.4	31.0
	Change in storage (V)	-0.3	0.7

< Methodology | Discussion >