The approach for this study included selection, enhancement, and application of a numerical model capable of simulating flow and solute transport within the SICS area and into Florida Bay. Data were obtained from several sources to apply, calibrate, and test the model. Additionally, results from ongoing or recently completed process studies were used to develop the model.
Ruhl, H. A., Hansler, M. E.
Rybicki, Nancy B.; Reel, Justin T.; Ruhl, Henry A.; Stewart, David W.; Jones, John W.
Wolfert, Melinda A.; Bales, Jerad D.; Goodwin, Carl R.
Swain, Eric
This paper was presented at the Second Federal Interagency Hydrologic Modeling Conference, Las Vegas, Nevada, July 28 to August 1, 2002
DeWitt, N.
Gritton, E. C.
Cyran, E.; Caruso, V.; Shupe, G.; Glover, R.; Henkle, C.
accessed as of 8/23/2010
Topography: Topographic data included land-surface elevations, embayment bathymetry, Buttonwood Embankment elevations, and the widths and bottom elevations of tidal creeks. The USGS used a helicopter-mounted global positioning system (GPS) unit and weighted line to measure land-surface elevations (Desmond and others, 2000) in the Everglades wetlands. Land-surface elevations were measured on a grid with about 400-m spacing (Henkle, 1996).
The bathymetries of Joe Bay and Florida Bay were measured by the USGS (Hansen and DeWitt, 1998), using a boat with a depth finder and a GPS unit. More than 30,000 individual depth measurements along boat track lines were made in Joe Bay. The bathymetries of other subembayments and Florida Bay were determined from nautical charts developed from National Oceanic and Atmospheric Administration (NOAA) data. The USGS Florida Bay data coincided well with the nautical charts, while providing a much more refined representation of the embayment bathymetry than was previously available. The bathymetries of West Lake and associated lakes were estimated from depth measurements made by USGS personnel in September 1999. All bathymetric and topographic elevations were referenced to the North American Vertical Datum of 1988 (NAVD 88).
The location of the Buttonwood Embankment was derived from 1:24,000 scale USGS topographic maps. Based on observations by USGS personnel, the bottom elevations of tidal creeks at the location where the creeks cut through the embankment are all at an elevation of about 1.5 m below NAVD 88. The creek widths were also determined.
Channels connecting the lakes in the southwestern part of the study area (West Lake area) were estimated to be about 24 m wide.
Vegetation: Aerial- and land-based vegetation surveys (Carter and others, 1999; Jones, 1999) were conducted, and results were compiled into a geographic information system (GIS) database (Stewart, 1997). The aerial surveys were conducted to determine regional vegetation types by using spectral reflectivity and visual onsite observations for ground truthing. These data were compared with data previously collected for the area, such as the 68-class 1993-94 Landsat vegetation map and a 20-class Landsat thematic mapper image (February 2000). The vegetation in the study area was categorized into eight classes at a horizontal resolution of 30 m
Water level, currents, and discharge: Water-level, current, and discharge data were collected at 23 sites in the study area and 1 site (P33) north of the study area. Data used for this study were collected by the SFWMD, USGS, and NPS through existing monitoring networks and research projects.
Gage datums at the continuous water-level recording stations were established relative to the National Geodetic Vertical Datum of 1929 (NGVD 29). The model was developed, however, using topographic data referenced to NAVD 88. Therefore, gage datum adjustments were made at all water-level stations. The CORPSCON datum adjustment routine (North Carolina Geographic Information Coordinating Council, 1999) was used to adjust datums from NGVD 29 to NAVD 88 at all stations except R127, TSH, P37, E146, and CP. At these stations, reference marks relative to NAVD 88 were established during land-surface elevation surveys (Gordon Shupe, U.S. Geological Survey, written commun., 1999). Subsequent surveys determined the differences between the NAVD 88 reference marks and previously established NGVD 29 datums.
Flow velocity, depth, and water-quality constituents (water temperature, dissolved-oxygen concentration, specific conductance, and pH) were measured within the study area at 137 locations, including along three approximately east-west transects in the wetlands (Schaffranek and others, 1999). Acoustic Doppler velocimeters were used to make point velocity measurements at various depths in the water column, and depth-averaged velocities were calculated at each measurement location for use in model evaluation. The data were collected between July 1997 and July 1998. Data from the first two measurement sessions were used in this study.
Salinity: Salinity measurement stations are located at the mouths of coastal creeks. Specific conductance sensors were deployed at multiple vertical positions in the water column to detect any vertical stratification.
Rainfall, wind and solar radiation: Initial applications of the SICS model (version 1.1) used rainfall data collected at the Old Ingraham evapotranspiration site and the Joe Bay weather station. Data for the model domain were interpolated from these two sites. This approach proved to be unsatisfactory, considering the large spatial variations in southern Florida rainfall. In order to improve the spatial representation of rainfall, 15-minute data were collected at a network of 14 rainfall stations, and a kriging algorithm was used to determine rainfall amounts for each cell.
Wind speed and direction were recorded at 15- minute intervals at the Old Ingraham Highway site and at the P33 site, which is located about 29 km north of the Old Ingraham Highway site.
Solar radiation data also were collected by a pyranometer at the Old Ingraham Highway and P33 sites. A comparison of solar radiation between the two sites showed a mean absolute difference (MAD) in hourly solar radiation values of 48.5 W/m2 in 1997, which is 14 percent of the daily mean solar radiation at the Old Ingraham Highway site. Because the Old Ingraham Highway site borders the study area, the pyranometer data from this site are used to represent solar radiation over the entire model domain.
Model Construction, Calibration, Testing, and Application: The modeling approach consisted of: (1) computational grid development; (2) assembly of data for model boundary conditions and model testing; (3) selection of initial values of model parameters based on results of field and laboratory process studies; (4) model calibration using data from the period September 1-30, 1997; (5) performance testing in which simulation results were compared with data collected from August 1996 to July 1997; (6) sensitivity analysis in which the effects on simulation results of small changes in model parameters and boundary data were evaluated; and (7) model application. Sensitivity analysis included evaluation of the effects of changes in the flow coefficient for coastal creeks, wind-friction coefficient, evapotranspiration rate, wetlands frictional resistance, boundary water levels, tidal function, boundary discharge, salinity, and land elevation on simulation results. The model was applied to quantify the effects of wind and of varying discharges at the Taylor Slough boundary on flows in Taylor Slough and to Florida Bay.
Model domain: The model domain is irregularly shaped and contains 9,738 computational cells. The computational cells are 305 m square, so the total area of the model domain is 905.8 km2. The maximum north-south extent of the domain is 29.90 km (98 computational cells), and the maximum east-west extent is 45.14 km (148 computational cells). The 305-m square grid cells provide good resolution of the study area, and do not require unreasonable computer resources to perform simulations. The Florida Bay open boundary of the model was positioned a sufficient distance from the freshwater-saltwater mixing zone to minimize the problems associated with gradient type boundary conditions.
Land-surface elevations were measured at about 400-m spacings (Desmond and others, 2000). A linear distance-weighted four-point interpolation of the 400-m (1,312 ft) spaced data was used to assign land-surface elevations to each computational cell of the 305-m SICS grid.
The spatial characteristics of the Buttonwood Embankment vary at a scale that is much smaller than the computational cell size. Rather than attempting to assign relatively high land-surface elevations to computational cells along the embankment, which would result in poor resolution of the embankment, flow barriers were used to represent the embankment. In cells where the embankment is diagonal to the cell, the embankment is represented by a series of north-south and east-west flow barriers within the cell. The location of the embankment was derived from field observations aided by USGS 1:24,000 scale topographic maps. All creeks flowing through the embankment are defined as cuts in the flow barriers. Elsewhere, creeks are defined as solitary flow barriers with the sill elevations at the creek bottom.
Boundary conditions: Boundary conditions are supplied at the water surface and at lateral boundaries.
Lateral Boundaries: Lateral boundaries are defined as open (having free exchange of water and salt across the boundary) or closed (having no flow across the boundary). Open boundaries can be described by a time series of discharge or water level, with discharge boundary conditions generally providing more realistic simulation results.
Model selection The two-dimensional, vertically integrated, unsteady flow and transport model SWIFT2D (Leendertse, 1987) was applied to the study area. The model was first developed for applications in Jamaica Bay, N.Y. (Leendertse and Gritton, 1971). Since that time, the model has undergone numerous revisions and updates, including enhancements described in this report for the SICS area application.
The SWIFT2D model was originally designed to simulate flow and transport in vertically well-mixed estuaries, coastal embayments, lakes, rivers, and inland waterways. SWIFT2D also includes many features such as time-stepping options, advective term discretization options, transport of passive tracers, coupled salinity transport, flooding and drying, the ability to include hydraulic structures, two alternative bottom friction terms including a form based on the subgrid scale energy level, a parametric expression for turbulence effects, various formulations for horizontal dispersion, and reactions and local inputs for transport.
Calibration: The model was calibrated by adjusting computational control parameters over reasonable ranges to produce the best match between computed and measured values of velocity flow and salinity for the calibration period. The calibrated model was subsequently applied to a different time period to verify or assure that the determined parameter values apply for conditions other than those used for calibration.
The calibration period was chosen to coincide with a time of intensive measurement of flow velocities in the wetlands. The 1-month simulation period was September 1-30, 1997. Results from September 22 to 25 were used for comparison with field velocity data, which corresponds to one of four field measurement efforts conducted from July 1997 to July 1998. A 17- day warm-up period (August 15-31, 1997) was used to allow effects of errors in the initial conditions to dissipate from the model.
Testing: The verification simulation run encompasses the period from August 1996 to July 1997. This time period also was used by the U.S. Army Corps of Engineers in the development of boundary conditions for inland and Florida Bay models. Including a 16-day warm-up period, the model run starts on July 15, 1996, and finishes July 31, 1997.
The application of the SICS model allowed both the estimation of flows at locations where field measurements are not made, as well as at measured sites for hypothetical scenarios.
A sensitivity analysis was performed on the major input parameters of the model.
For a more complete description of the model development, calibration, and testing see Two-Dimensional Hydrodynamic Simulation of Surface-Water Flow and Transport to Florida Bay through the Southern Inland and Coastal Systems (SICS) at <http://sofia.usgs.gov/publications/wri/03-4287/index.html>.
U.S. Department of the Interior, U.S. Geological Survey
Comments and suggestions? Contact: Heather
Henkel - Webmaster
Generated by mp version 2.8.18 on Mon Aug 23 12:34:38 2010