Calibration of a steady-state seepage model from transient recovery of field data

The South Florida Water Management District (District) constructed three experimental test cells (20 x 50 feet = 6.1 x 15.2 m) to determine the seepage influence of a major agricultural canal adjacent to the cells. These three cells represented the first stage of a series of experimental cells to determine the efficiency of algal-bacterial communities (periphyton) to successfully remove phosphorus (P) from the water. This treatment technology is known as Periphyton-based Stormwater Treatment Area (PSTA) and is being considered under the Supplemental Technology Program (Phase II) to comply with the Everglades Forever Act (EFA) of 1994. However, the efficiency of a PSTA in removing nutrients has not been documented yet. To determine the efficiency of PSTA technology to remove phosphorus (P) from the water, it is necessary to quantify all the components of the water budget. The purpose of the cells was to evaluate the potential seepage due to the operational range of stages into and out of the test cells. A steady-state, two-dimensional seepage (FastSEEP/SEEP2D) model was applied and successfully calibrated for that purpose. The model was calibrated using mean seepage flows obtained after drawdowns in each test cell, by pumping water out. These seepage flows as well as the boundary conditions (stages) were estimated by discretizing the recovery data. The recovery data represents the seepage from the agricultural canal into each of the test cells after the drawdowns, since the stages in the canal during the experiment did not change significantly and were high. Each of the three test cells was excavated to different bottom elevations to determine the effect of substrate differences on seepage rates. Values of hydraulic conductivity for five different layers in the models were based on field observations, lithologic descriptions, and slug tests. Modeled seepage produced reasonable results for the three test cells compared with estimated seepage from the recovery data, achieving a satisfactory calibration. The errors of the simulated seepage ranged from -6.1 to 16.8 percent (under and over estimation, respectively).