An extensive canal and water management system exists in south Florida to prevent flooding, replenish groundwater, and impede saltwater intrusion. The unconfined Biscayne aquifer, which underlies southeast Florida and provides water for millions of residents, interacts with the canal system. The Biscayne aquifer is composed of a highly transmissive karst limestone; therefore, canal stage and flow may be affected by production well pumping, especially in locations where production wells and canals are in proximity.

The U.S. Geological Survey developed a local-scale, transient, numerical groundwater flow model of a well field in Miami-Dade County to (1) quantify relations between well-field pumping and C-2 Canal (herein referred to as the Snapper Creek Canal) leakage, (2) determine primary controls on canal leakage variability, and (3) summarize results that could simplify characterization of canal/well-field interactions in other locations. In addition to the groundwater model development, stable isotope data from water-quality samples were used to characterize the relations between production well pumping and canal leakage. The results from the groundwater model and the isotope data were used to refine the conceptual flow model of the study area.

The groundwater flow model MODFLOW-NWT was used for simulating groundwater flow and quantifying interactions between pumping from the well field and Snapper Creek Canal leakage. Input data for the groundwater model included precipitation, evapotranspiration, pumping, canal stage, and regional groundwater elevation. The inverse modeling tool UCODE and groundwater data from June 2010 to July 2011 were used to calibrate the model. Parameter sensitivity analyses were performed with UCODE. Model sensitivities to geologic heterogeneity, non-laminar flow, and changes in the regional flow boundary were evaluated. The groundwater model generally fits the calibration criteria well within estimated error ranges for groundwater elevations and canal leakage values. The mean average error for heads simulated with the model was 0.19 meter, and head residuals were generally randomly distributed.

The model simulated groundwater flow under ambient conditions without production well pumping to establish background leakage. Groundwater flow also was simulated with production well pumping to estimate induced leakage from the Snapper Creek Canal that occurs in response to pumping.

Canal leakage was quantified as a percentage of total canal leakage. The percentage of leakage during pumping increased non-linearly with pumping rate, indicating a decreasing sensitivity of canal leakage to pumping at relatively large pumping magnitudes. The results for Snapper Creek Canal may serve as an upper limit for well-field interaction with surface-water features in Miami-Dade County, given the proximity (about 50 meters) of the pumping wells in this study to the Snapper Creek Canal.

The isotopic compositions of hydrogen (H) and oxygen (O) in groundwater samples were used to distinguish sources for groundwater within the study area and to assess the extent of natural mixing and pumping-induced mixing with water in the Snapper Creek Canal. Water-level data and water-quality samples were collected from monitoring well clusters, production wells, and the Snapper Creek Canal during discrete sampling events under ambient and pumping conditions. Trends in the isotope data generally follow the regional west-to-east hydraulic gradient across the study area. Data collected within the monitoring-well clusters in closest proximity to the canal indicate that groundwater/surface-water interactions are greatest within the shallow flow zone of the aquifer, especially during pumping conditions. The isotopic composition of samples collected within the study area indicates that the shallow, highly transmissive preferential flow zone receives substantial recharge from the canal. The isotope data from the production wells which are open to the deeper flow zone within the aquifer, indicate only traces of mixing with a ²H- and ¹⁸O-enriched source, suggesting little canal admixture with waters of the deeper flow zone.

Results from the groundwater model and the stable isotope data analysis indicate the importance of considering geologic heterogeneity when investigating the relations between pumping and canal leakage, not only at this site, but also at other sites with similar heterogeneous geology. The model results were consistently sensitive to the hydrogeologic framework and changes in hydraulic conductivities. The model and the isotope data indicate that the majority of the groundwater/surface-water interactions occurred within the shallow flow zone. A relatively lower-permeability geologic layer occurring between the shallowest and deep preferential flow zones lessens the interactions between the production wells and the canal.