ABSTRACT: A digital model of the flow system in the highly permeable surficial Biscayne aquifer of southern Dade County, Florida, was constructed for the purposes of better understanding processes that influence the flow system and of supporting the construction of a subregional model of the transport of brackish water from a flowing artesian well. Problems that needed resolution in this endeavor included the development of methods to represent the influence of flowing surface water in seasonally inundated wetlands and the influence of a network of controlled canals developed in stages during the simulation time period (water years 1945-89). An additional problem was the general lack of natural aquifer boundaries near the boundaries of the study area.
The model construction was based on a conceptual description of the Biscayne aquifer developed from the results of previous U.S. Geological Survey investigations. Modifications were made to an existing three-dimensional finite-difference simulator of ground-water flow to enable an upper layer of the grid to represent seasonally occurring overland sheetflow in a series of transient simulations of water levels from 1945 to 1989. A rewetting procedure was developed for the simulator that permitted resaturation of cells in this layer when the wet season recurred. An “equivalent hydraulic conductivity” coefficient was assigned to the overland flow layer that was analogous, subject to various approximations, to the use of the Manning equation. The surficial semiconfining peat and marl layers, levees, canals, and control structures were also represented as part of the model grid with the appropriate choices of hydraulic coefficient values.
For most of the Biscayne aquifer grid cells, the value assigned to hydraulic conductivity for model calibration was 30,000 feet per day and the value assigned to porosity was 20 percent. Boundary conditions were specified near data sites having long-term records of surface-water stages or water-table altitudes, and modifications to the simulator permitted the specification of time-varying pressures at boundary grid cells. Rainfall data from a station in Homestead generally were used as an areally uniform rainfall specification throughout the modeled region. Maximum evapotranspiration rates ranged seasonally from a minimum of 0.08 inch per day in January to a maximum of 0.21 inch per day between June and October. Shallow-root and deep-root zone depths for the evapotranspiration calculation were 3 and 20 feet in the coastal ridge and were 0.10 and 5 feet in the glades regions where peat and marl covers occurred.
Results of sensitivity analyses indicated that the simulations of stages and water levels were relatively unresponsive to 50 percent changes in aquifer hydraulic conductivity, porosity, and the equivalent hydraulic conductivity of overland flow. However, 20 percent changes in rainfall and maximum evapotranspiration rates produced significantly different water levels, as did interchange of coastal ridge and glades deep-root zone (extinction) depths.
Water levels were simulated very well at most measurement sites. Sensitivity analyses illustrated the significant influence of the uncontrolled agricultural drainage canals on pre-1968 regional water levels and the further influence of Black Creek Canal in draining a region of high water after 1961. Other analyses indicated that the flood-control system of 1968-82 lowered peak water levels in the affected region by as much as 1.5 feet in the wet summers of 1968, 1969, and 1981, and that Levee 67 Extended channeled flows from the S-12 spillway structures and raised overland flow stages in Shark River Slough. Hypothetical scenarios of well-field pumping in the vicinity of Levee 31N indicated that the pumping induced a significant amount of recharge from the adjacent borrow canal, the degree of which depended on the distance between the canal and the well field. The computed ratio of evapotranspiration to rainfall recharge ranged from 88 to 94 percent during water years 1945-82. The ratio increased to about 97.9 percent during water years 1983-89, possibly because of changing water-management practices and deficient rainfall.