Grantee Research Project Results

Final Report: Development of an Integrated End-To-End Marine Contaminant Management System

EPA Grant Number: R825197
Title: Development of an Integrated End-To-End Marine Contaminant Management System
Investigators: Luther, Mark E. , Friel, Christopher A , Schmidt, Nancy J. , Van Vleet, Edward S. , Vincent, Mark S. , Galperin, Boris
Institution: University of South Florida , Florida Department of Environmental Protection
EPA Project Officer: Aja, Hayley
Project Period: October 1, 1996 through September 30, 1999 (Extended to September 30, 2000)
Project Amount: $588,777
RFA: High Performance Computing (1996) RFA Text |  Recipients Lists
Research Category: Human Health , Aquatic Ecosystems , Environmental Statistics

Objective:

The purpose of this project is to develop an integrated end-to-end marine contaminant management system by building the software interfaces to link existing components in Tampa Bay with those under development. This system incorporates real-time data acquisition, hydrodynamic modeling, spill trajectory modeling, a geographical information system (GIS) resources-at-risk database, and data dissemination into a user-friendly problem solving environment designed to assist local, state, and federal decision makers to quickly respond to, manage, and remediate accidental releases of petroleum products or other hazardous materials into Tampa Bay. Although we focus on oil spills in Tampa Bay for our testbed problem, the system we have developed is equally applicable to any contaminant in the marine environment and can be ported to other estuarine systems. The individual components of the system are highly computationally intensive. Linking them together in an integrated end-to-end system constitutes a task in high performance computing.

The Tampa Bay Physical Oceanographic Real-Time System (PORTS), together with streamflow and rainfall measurements from the U.S. Geological Survey and the National Weather Service, supply all the data necessary to provide boundary conditions and independent verification for a high-resolution 3-D hydrodynamic circulation model of Tampa Bay. Output fields of current from the hydrodynamic model drive a Lagrangian trajectory model to simulate the movement of contaminant plumes. The real-time observations, the output from the circulation model, and the plume trajectories are incorporated as layers in a GIS that also includes layers of resources-at-risk in Tampa Bay. Maps produced from the GIS can be disseminated to resource managers in electronic form on notebook computers in an easily understood format. The circulation and trajectory models can be run in hindcast, nowcast, or forecast modes to provide guidance for contingency planning as well as for optimal utilization of resources in an actual spill.

As estuarine residual circulation is controlled in large part by freshwater inflow, we have worked to quantify the relationship between natural climate variability, like the El Ni?o-Southern Oscillation (ENSO) and the North Atlantic Oscillation (NAO), and freshwater input to Tampa Bay and resulting residual salinity distribution (Schmidt et al., 2001; Schmidt and Luther, 2001a, b). The effects of this variability in freshwater inflow on the occurrence of human pathogens in Tampa Bay and Charlotte Harbor were investigated as a spin-off of this research (Lipp, et al., 2001). In addition, we have used the modeling system developed herein, under separate funding, to evaluate the cumulative and individual effects of planned freshwater diversions and discharge from a seawater desalination facility on Tampa Bay salinities and residual circulation (Vincent et al., 2001).

This project is a partnership between the University of South Florida USF) College of Marine Science and the Florida Department of Environmental Protection (now Florida Fish and Wildlife Conservation Commission) Florida Marine Research Institute and leverages the resources of the Tampa Bay National Estuary Program, the National Oceanic and Atmospheric Administration (NOAA) National Ocean Service (NOS), the U.S. Coast Guard, and the NOAA National Weather Service.

Summary/Accomplishments (Outputs/Outcomes):

Model Description

The USF College of Marine Science has developed a three-dimensional time-dependent model of the hydrodynamics of circulation in Tampa Bay (Galperin et al., 1992a, b; Vincent et al., 1997, 2000), based upon an advanced version of the Princeton Ocean Model (Blumberg and Mellor, 1987). The governing equations consist of conservation of mass and momentum and conservation equations for thermal energy and salt. Equations also are solved for the turbulence kinetic energy and turbulence macroscale. Salient features include a curvilinear orthogonal grid in the horizontal plane, and a bed and free surface following sigma coordinate system in the vertical axis. Turbulence closure is provided by an embedded Mellor-Yamada 2.5 level closure submodel (Mellor and Yamada, 1982) as modified by Galperin. Time splitting allows for the fast external or barotropic waves to be solved for explicitly, and the slower internal baroclinic waves implicitly. Specified forcing boundary conditions include the free surface elevation and temperature/salinity profiles at the open water boundary; the flow rate, temperature, salinity and level of inflows or outflows; surface heat flux; and surface stresses due to wind, precipitation, and evaporation. Among the important parameters computed are free surface height, magnitude and direction of current velocity fields, and temperature and salinity fields. Various model versions have been deployed and tested at numerous sites such as the Hudson-Raritan Estuary, Chesapeake Bay, Delaware Bay, Apalachicola Bay, Florida Bay, the lower Mississippi River and adjacent continental shelf, and the New York Bight. The present version of the Tampa Bay model uses a 70-by-100 horizontal curvilinear grid (Figure 1) with 11 sigma levels in the vertical (Figure 2). Boundary conditions for the Tampa Bay model are provided by the PORTS data stream. The model is run on a SGI mini-supercomputer and is automatically updated every 6 minutes to provide a "nowcast" of present conditions in the bay. Every 4 hours, a 24-hour forecast is performed using winds from the National Weather Service ETA model and water levels at the mouth of the bay extrapolated from present observations. Model nowcast and forecast fields are presented in graphical format. They can be viewed on the Ocean Modeling and Prediction Laboratory (OMPL) Web Site (http://ompl.marine.usf.edu/TBmodel ) and can be provided via ftp.

Spill Trajectory Model

The hydrodynamic model output velocity fields drive a spill trajectory model to predict the movement of hazardous material spills in Tampa Bay. Spills are treated in two ways: (1) as a passive Eulerian tracer modeled by an advection-diffusion equation on the same grid as the hydrodynamic model, where the advective velocities and eddy diffusivity are taken from the hydrodynamic model; and (2) as a cloud of a large number of Lagrangian particles, each modeled by a first order Markov process using instantaneous velocities from the hydrodynamic model and a Markovian dispersion coefficient calibrated using observed drifter tracks. To date, both the passive tracer and surface drifter model show promise in the ability to track both surface trapped species (external Lagrangian drifter) and plumes that are neutrally buoyant (internal passive tracer). The particle tracking model compares well against observed drifter tracks (Figure 3). Information on location and time of a spill or marine accident can be entered into a Web-based application (TRACKER), which will return a 24-hour forecast of the trajectory of the spill (Figure 4; http://ompl.marine.usf.edu/~burwell/ ; this form is password protected due to the computationally-intensive nature of the application. Contact burwell@marine.usf.edu for access to this application). The information on oil distribution from either spill model is ingested into the state's Florida Marine Research Institute's Marine Spill Assessment and Response System (FMSAS), a GIS-based spill mitigation tool. The predicted distribution of oil from the spill model forms a layer in the MSAS database and can be used as a template to cut through the resources-at-risk data layers to arrive at an inventory of resources exposed.

Residence Time

Estuarine residence time is estimated by seeding each model grid cell with large numbers of particles or with a passive tracer, as described above for the spill trajectory module (Burwell et al., 2000). The e-folding time for either particle or tracer concentration is computed in each grid cell under a variety of boundary conditions observed in the bay (Figure 5). The resulting residence times vary widely in space and time. Residence time is most sensitive to variations in wind forcing and to variations in freshwater input. Residence time is a critical control of water quality and pollutant concentration.

Wireless Data Delivery

Information from the real-time observations or model output and on the predicted distribution of oil can be delivered in real-time to harbor pilots, shipping agents, resource managers, or others in the field using wireless Internet technology. We are using two different wireless technologies in Tampa Bay. The first is Cellular Digital Packet Data (CDPD) or Wireless IP that uses the commercial cellular network to connect to the Internet from a notebook computer wherever the local cellular provider supports this service (usually the wireless cellular provider; see http://www.attws.com/business/gov/explore/wireless_ip/network/ for more information). The second is via a dedicated 900 Mhz spread spectrum radio modem link using Point-to-Point Protocol (PPP) between the USF St. Petersburg campus and a remote notebook computer. This radio link has a radius of approximately 30 miles, depending on antenna height. The advantage to the spread-spectrum radio link over the CDPD link is that the cellular frequencies often become crowded and throughput is hindered. The spread-spectrum radio link has much faster throughput even under optimal conditions for CDPD. The disadvantage to the spread-spectrum radio is its range limitation, although multiple radios can be used as repeaters to easily set up a local network. Because they work in the 900 Mhz frequency band, no special licensing is required. Using either wireless communications link, the remote computer can access the predicted spill trajectory or other model or observational products using standard Web browsers.

In addition, the wireless delivery technology can be implemented through our collaboration with Ross Engineering, the developer of the Tampa Bay Vessel Information and Positioning System (see http://www.rossdsc.com/ais.htm ). Ross provides a wireless wide area network in the Tampa Bay region and has the capability to transmit real-time data from Tampa Bay PORTS to the pilot carry-on units in use in Tampa Bay (Husick, 1999).

Quality Assurance

All data collected are subjected to quality control/quality assurance standards of the NOAA National Ocean Service. The Tampa Bay PORTS data are continuously monitored 24 hours/7 days per week by NOS personnel. The hindcast, nowcast, and forecast models have been validated extensively against all available data sets using the standard NOAA skill statistics of Hess and Bosley (1992). Skill scores for the model nowcasts and hindcasts exceed 0.96 for water levels and 0.90 for velocities. The mean absolute error between modeled and observed salinities is less than one part per thousand. The operational nowcast/forecast model protocol contains numerous error-trapping contingencies and redundant data pathways to ensure reliability. Skill assessment for the forecast product is ongoing.

Figure 1. 70 x 100 horizontal curvilinear orthogonal grid with Tampa Bay PORTS observing sites.

Figure 2. Typical cross section of 11 sigma layers in the vertical dimension.

Figure 3. Test of Lagrangian trajectory model. Black lines are observed track of two surface drifters. Colored lines are tracks of approximately 1000 particles in the model.

Figure 4. Example of output from the TRACKER application for a hypothetical spill to the northwest of the Sunshine Skyway Bridge. Color represents time since the initial spill, with purple to blue being less than 2 hours and reds being greater than 18 hours.

Figure 5. Residence time based on the particle trajectory (Lagrangian) method for boundary conditions observed during the fall and winter of 1997-98.

References:

Blumberg A, Mellor GL. A description of a three-dimensional coastal ocean circulation model. In: Heaps NS, ed. Three-Dimensional Coastal Ocean Models. American Geophysical Union, Washington, DC, 1987, pp. 1-16.

Burwell D, Vincent M, Luther M, Galperin B. Modeling residence times: Eulerian vs. Lagrangian. In: Spaulding ML, Butler HL, eds. Estuarine and Coastal Modeling, ASCE, Reston, VA, 2000, pp. 995-1009.

Galperin B, Blumberg A, Weisberg R. A time-dependent three-dimensional model of circulation in Tampa Bay. In: Treat S, Clark P, eds. Proceedings of the Tampa Bay Area Scientific Information Symposium 2, Tampa, FL, February 27-March 1, 1991, 1991a;77-97.

Galperin B, Blumberg A, Weisberg R. The importance of density-driven circulation in well-mixed estuaries: the Tampa Bay experience. In: Proceedings of the 2nd International Conference on Estuarine and Coastal Modeling, Tampa, FL, November 13-15, 1991, 1992b;332-343.

Hess K, Bosley K. Techniques for validation of a model for Tampa Bay. In: Spaulding M, Blumberg A, eds. Proceedings of the 2nd International Conference on Estuarine and Coastal Modeling, ASCE, Reston, VA, 1992;83-94.

Husick C. Tampa Bay setting the pace with Automatic Identification System. Professional Mariner 1999;38:37-79.

Lipp EK, Schmidt N, Luther ME, Rose JB. Determining the effects of El Nino - Southern Oscillation events on coastal water quality. Estuaries 2001;24(4):491-497.

Schmidt N, Lipp EK, Rose JB, Luther ME. ENSO influences on seasonal rainfall and river discharge in Florida. Journal of Climate 2001;14(4):615-628.

Schmidt N, Luther ME. Modulation of ENSO impacts in Florida by the NAO. Journal of Climate 2001.

Schmidt N, Luther ME. ENSO impacts on salinity in Tampa Bay, Florida. Estuaries 2002;25(5):976-984.

Vincent M, Burwell D, Luther M, Galperin B. Real-time data acquisition and modeling in Tampa Bay. In: Spaulding ML, Blumberg AF, eds. Estuarine and Coastal Modeling, ASCE, Reston, VA, 1998, pp. 427-440.

Vincent M, Burwell D, Luther M. The Tampa Bay Nowcast-Forecast System. In: Spaulding ML, Butler HL, eds. Estuarine and Coastal Modeling, ASCE, Reston, VA, 2000, pp. 765-780.

Vincent M, Luther M, Burwell D, Galperin B. Cumulative effects of a proposed desalination facility and fresh water diversions on residual salinity and circulation in Tampa Bay, Florida. Estuaries 2001.

Journal Articles:

No journal articles submitted with this report: View all 13 publications for this project

Supplemental Keywords:

RFA, Scientific Discipline, Waste, Ecosystem Protection/Environmental Exposure & Risk, Hydrology, Remediation, chemical mixtures, computing technology, Hazardous Waste, Ecological Risk Assessment, Hazardous, data sharing, environmental decision making, HPCC, real time data acquisition, computer science, geographical information systems, hydrological transport model, oil spills, remediation of accidental releases, data analysis, GIS, information technology, petroleum spill trajectory modeling, contaminant management, hydrodynamics, marine contaminant management system

Relevant Websites:

http://ompl.marine.usf.edu/PORTS/
http://ompl.marine.usf.edu/TBmodel/
http://comps.marine.usf.edu/

Progress and Final Reports:

Original Abstract

1997

1998

1999