Progress Review of Florida Bay Models:
Report of the Model Evaluation Group
May 11, 1998
Model Evaluation Group
Christopher F. D’Elia, co-chair
CBL, Center for Environmental Science
University System of Maryland
U.S. Geological Survey
Kenneth H. Dunton
University of Texas
Port Aransas, Texas
Department of Civil Engineering
University of Virginia
Steven C. McCutcheon, co-chair
Hydrologic and Environmental Engineering
Submitted to the
Program Management Committee
Florida Bay Research Program
Summary and Recommendations
The Florida Bay Research Program Management Committee (PMC) invited the Model Evaluation Group (MEG) to participate in a progress review of the Florida Bay modeling effort. The MEG’s major recommendations to the PMC are given below.
1. Create a Web page as a central depository for all field and laboratory data to:
2. Appoint a PMC program coordinator to:
3. Determine whether the CE-QUAL-ICM model will adequately address sea grass issues, commissioning additional process and data collection investigations if necessary, or formulating more elaborate complementary models. Although the MEG believes that there is a reasonable expectation that the proposed simulation of two classes of sea grass may address most management issues, this vital issue must be reconsidered annually until restoration goals are achieved.
4. Address whether the FATHOM model is necessary. If the CE-QUAL-ICM model does not meet the needs of biologists, then the water quality modeling study should be stopped and re-planned. If FATHOM is necessary, then validation and establishment of credibility is essential to support management decisions.
5. Conduct a formal review to verify that early concerns about the coverage and intensity of water quality data have been addressed:
6. Make key improvements in the analytical and data-management components of the monitoring program:
7. Monitor closely the linking of a finite-element model and a box model to ensure that the extraordinary challenges of doing so are adequately met.
8. Improve internal management communications in the Florida Bay project. The members of the PMC and their agencies need to set explicit priorities, coordinate effectively, and as appropriate, communicate clearly goals, objectives, results and needs of the different components. This challenge to improve program integration must be addressed effectively in the near term if the modeling effort is to meet expectations in a cost-effective and timely fashion. Accordingly, the MEG recommends that a routine coordination conference call, meeting or electronic update be developed to improve communications between modelers and other components of the program. In addition, recommendations made by previous review panels with regard to organizational improvements for program management should be pursued without further delay.
9. Employ the MEG to address and support the efforts in Florida Bay by hearing reports, in concert with the PMC:
Preceding the 1998 Florida Bay Science Conference, May 12-14, 1998, the Program Management Committee (PMC) of the Florida Bay Research Program convened a progress review of modeling efforts subsequent to the October, 1996 Workshop on the design and specifications for the Florida Bay Water Quality Model (D’Elia, et al., 1996). The focus of the meeting was to review progress to date and changes in the scope of the Water Quality Model (WQM) necessitated by reduced available funding. The Modeling Evaluation Group (MEG) heard a series of brief presentations on May 11 to summarize progress to date, needs, changes in scope, and future plans of the modeling effort. The agenda for this meeting is given in Appendix I.
Suggested Discussion Topics
The PMC recommended that the MEG evaluate the presentations ad modeling effort according to a series of discussion topics:
The RMA model is (nearly) completed whereas the WQ model and the biological models are only beginning. Exactly how will the RMA and WQ models be used by WES? Is there an intention by other agencies or scientists to use the models further?
The work plan has gone from a four-year $2M program to a two-year ($0.75M cost). Will it deliver what the PMC requires?
What are the apparent gaps in agency data collection plans? Are the key parameters being collected in the right way in the right places. What mechanisms are needed for incorporating data from non COE projects into the water quality and other models?
What are the key steps in assuring that model and data integration succeed. To what degree are the models interdependent? Will any models be directly coupled? What is required of the funding agencies, and modelers to achieve integration?
How can the PMC use MEG services next?
Against the background of the above discussion topics are a series of general recommendations made in previous workshops (Appendix II), which are considered explicitly here as is relevant. The organization of the present report with that of the discussion topics that the PMC has suggested.
I. Implication of Model Development Scenarios
The RMA model is (nearly) completed whereas the WQ model and the biological models are only beginning. Exactly how will the RMA and WA models be used by WES? Is there an intention by other agencies or scientists to use the models further? The flow model (RMA2 [2-D version of RMA10]) developed for the Florida Bay was presented at the May 11, 1998 meeting. While the model results of water surface and tidal currents match the measured data quite well, we would like to see more comprehensive comparisons between calculated salinity and data. As WES has noted before, the vertical averaged 2-D flow model like RMA2 de-couples the tidal currents with salinity. Essentially, the flow field is not affected by the salinity in this case by neglecting the baroclinic effect. Thus, it is important to compare the model-calculated salinity with the data, as stated by the hydrodynamic modeling staff of WES at the meeting. This exercise should be performed with the fine, unstructured grid for the model. Since the hydrodynamic model will be used to assess the impact of freshwater flow changes on the salinity levels in the Bay, it is extremely important for the model to match the salinity data during the model calibration and validation effort.
The developers of the controversial RMA model may have missed a vital opportunity to investigate the effects freshwater inflow changes from the restudy areas. This was one of two reasons to justify the use of this model. The Jacksonville District and the PMC should investigate why this golden opportunity to provide vital information to a high visibility project was missed. It also seems that project is one or more years overdue. The original schedule should be compared with progress to redirect this project as necessary.
The water quality model is in a very early stage of development. It is not clear to us now whether and how other agencies and scientists would use the water quality model.
II. Implications of Reduced Scale of Water Quality Model
The work plan has gone from a four-year $2M program to a two-year $0.75M cost. Will it deliver what the PMC requires? While we are uncertain what the PMC specifically expects from the water quality model, it is clear that the water quality model must predict and quantify the impact of freshwater flow changes on the sea grass in the Florida Bay. With the funding cutback and due to being in the early stage of water quality model development, it is difficult to assess what the water quality model will deliver and what the model is capable of predicting.
The only presentation about the water quality model was on the linkage between the flow and water quality models. The idea of using a coarser grid for the water quality model, that is congruent with the hydrodynamic grid, is acceptable. Due to the unstructured grid used in the RMA2 model, the coarse grid for the water quality model can be configured easily. It is consistent with the original call for the integration of two models without spatial averaging. While preliminary results from a small portion of the Florida Bay on the model integration was presented at the meeting, it is crucial to demonstrate that this scheme works for the entire study area. Linkage involving temporal or spatial averaging where wetting and drying occurs is notoriously difficult (Elder, 1994; Lung and Huang, 1990) using finite difference models. Linking finite element and finite volume solutions (or box models based on finite difference solutions) is expected to be a remarkable advance that may divert attention from simulations that address vital management goals and objectives.
In addition, mass transport in the water quality model must be verified against the salinity data. Further, salinity results from the water quality model (coarse grid) must match the salinity results from the hydrodynamic model (fine grid) to a reasonable degree defined a priori, consistent with management objectives.
III. Critical Data Needs
Apparent gaps in agency data collection plans. In general, there is now a substantial monitoring and research program underway in Florida Bay that provides strong and essential service to the modeling efforts. The water quality data may be adequate to support the modeling work, but data on sea grass is questionable. Moreover, the monitoring program is heavily oriented to the collection of "state variable measurements" (salinity, nutrients, chlorophyll, etc.) and not directed at making "rate measurements" of processes (primary productivity, benthic nutrient fluxes, etc.). However, the research program does include studies that measure rates of some key processes that should be integrated into the modeling. As is often the case in comparable situations elsewhere, the principal investigators of these research studies are, by the nature of the funding process, not in as systematically close contact with those leading the modeling effort. Accordingly, considerable effort must be made to incorporate the results of this process-oriented research, as is appropriate (see below). The MEG recommends that the program identify and systematize the incorporation of essential rate measurement data required by the model into the monitoring database. An overall assessment of the coverage and intensity of the data to support the water quality modeling is necessary, however, to follow up questions originally raised by Mark Dortch in formulating the modeling plane.
Appropriateness of collection of key parameters. Most of the parameters measured in the monitoring program seem appropriate for support of the modeling effort. There are, however, some exceptions, and the MEG recommends strongly that adjustments be made to the monitoring effort, as is appropriate:
Mechanisms to improve incorporation of data from non-COE projects into models. See next section below.
IV. Essential Integration Tasks
Key steps in assuring that model and data integration succeed. Modeling cannot proceed effectively without access to the diverse array of data that is being collected by researchers. Various principal investigators have remarked that model development is virtually at a standstill because some researchers are unwilling to release data (they are perhaps concerned about premature publication of the data in modeling papers). Nonetheless, it is absolutely crucial that a central depository for data or metadata be established on the Web. Similar depositories exist for the Chesapeake Bay (www.chesapeakebay.net/bayprogram/data/model/data.htm) and for the Joint Global Ocean Flux Study (www1.whoi.edu/jgdms_info.html). These provide good examples of how modeling and data web pages might be constructed for Florida Bay. Furthermore, the WES modelers must communicate with the Florida Bay scientists closely to take full advantage of the existing data and information available to support the modeling work.
A non-publication agreement among users would perhaps facilitate the timely release of field and laboratory data for modeling efforts. The listing of data on and a data policy on a special Web page would also facilitate communication between researchers and modelers, encourage collaboration, and result in more rapid identification of key data gaps that are critical in the development of accurate models. The success or failure of the entire Florida Bay Modeling Program is dependent on access to data that is specifically collected for various models, but the data must be made available.
During the workshop to formulate water quality modeling goals (D’Elia et al. 1996), Mark Dortch expressed concern about coverage and intensity of water quality state variables to calibrate the water quality model. Furthermore, useful rate investigations for some processes are underway, but it is not clear that all PI’s understand the PMC objectives and modeling goals. The PMC should require or undertake two assessments:
Next, a critical assessment of the RMA10 flow model calibration is necessary to be sure that the model will adequately support critical water quality management objectives beyond the original Jacksonville District goals to simulate salinity and flow trough the Keys. For example, does the calibrated RMA10 model adequately simulate exchange between basins to simulate hypersalinity, nutrient fluxes, turbidity and light attenuation?
Interdependence of models. The PMC is now faced with the complex task of integrating the FATHOM model, the RMA2 simulations, the WES model for water quality and sea grass, and the higher trophic level model or analysis framework. The PMC is correctly posing the question of whether a more focussed sea grass model may be necessary. Further, the USGS and other PMC members have already questioned whether the WES and water quality and sea grass model will address and answer the needs of biologists. They have decided that a simpler box model is required as the original water quality modeling plan specified (Dortch et al., 1997), but perhaps in disunity, selected an unvalidated, empirical approach that may or may not be useful.
The FATHOM model is based on simple linear reservoir theory that ignores the explicit effects of baroclinic circulation, wind mixing and requires extensive calibration without achieving adequate validity for forecasting or projection (FATHOM proposal to PMC, 1995). Already the investigators seemed to have rebuffed an offer from WES to provide the full water quality model grid (WES, personal communication, S.C. McCutcheon, February 26, 1998) so that a simpler composite FATHOM grid would overlap exactly for ease of transferring information from one scale to another. First, this bodes ill for overall coordination if managers with the biological lead prejudge the water quality model as inadequate. Second, the PMC should decide if the FATHOM model is actually necessary and develop an adequate peer review and validation program so the effort can support management decisions.
Finally, the upper trophic level modeling effort seems to have developed solid goals and objectives. It is now useful to define any support required from the water quality model as well as from the RMA circulation models, the sediment resuspension investigations, and the sea grass studies.
Direct coupling of models. The only direct coupling planned is to use RMA10 to drive the CE-QUAL-ICM. As noted above, this is a challenge that is only feasible because some of the "best in the business" are involved. The indirect linkages of other models to transfer information seem adequate at this time.
Requirements of funding agencies and modelers to achieve integration. Virtually every review or advisory committee empanelled comments on the managerial difficulties inherent in the committee structure of governance, i.e., the Program Management Committee (PMC), for the Florida Bay Program. This committee functions at minimal overhead cost to the program, but each of the members of the committee has primary duties in his or her own agency that take priority. Thus, no one on the PMC has been able to devote full attention to the complex and demanding coordinating activities necessary to keep PI’s in the modeling, monitoring and research components on track (and some slippage of deadlines has occurred). The MEG is encouraged that the there are now plans to hire an individual to perform this coordinating function. Nonetheless, the members of the PMC and their agencies need to be especially vigilant in setting explicit priorities, coordinating effectively, and as appropriate, communicating clearly goals, objectives, results and needs of the different components.
With specific regard to the modeling effort, it is crucial that communication between WES/COE and other program components be enhanced. Modeling is generally regarded as most effective when there is a "recursive" or "iterative" development process in place, i.e., when regular communications between modelers and others provides for continuous improvement of models (McCutcheon, 1989). The MEG emphasizes that the challenge to improve program integration must be addressed effectively in the near term if the modeling effort is to meet expectations in a cost-effective and timely fashion. Unfortunately, there is yet inadequate regular communication and coordination among key components of this effort. From statements made by presenters at the workshop and in casual conversation with MEG members after the meeting, the MEG realizes that additional mechanisms for communication must be developed. Accordingly, we suggest that the following be considered:
The MEG recognizes that the specific suggestions above may not be the ones that best suit the Florida Bay Program, its researchers and WES/COE. However, the MEG feels strongly that the interval of time between the Annual Science Meetings is too great to provide the interaction needed between managers, modelers and researchers. These individuals may have better ideas to improve communications and coordination, and accordingly, should initiate discussions specifically aimed at developing and implementing their ideas.
V. Concluding Remarks
Although there are several areas where better coordination is possible, the PMC continues to do an effective job within the constraints involved. Leadership and vision are evident in many projects. Hydrographic data collection now seems well coordinated. Tom Lee’s able leadership is to be complimented. The sediment resuspension investigations by Prager and Halley continue to be effective and provide vital bottom roughness and wind mixing information for the circulation and salinity modeling effort. Several NOAA projects provide not only primary information for decision making but also give important ocean boundary conditions and meteorological forcing. Overall, the coordination is effective. The effort remains a notable project.
Armstrong, N.E., DiToro, D.M., Hansen, D., Jenter, H, King, I.P., McCutcheon, S., Raney, D.C., and Signell, R. 1996. Report of the Florida Bay Model Review Panel on Florida Bay Modeling Workshop, April 17-18, 1996.
Boesch, D.F., Armstrong, N.E., Cloern, J.E., Deegan, L.E., Perkins, R.D., and Williams, S.L. Report of the Florida Bay Science Review Panel on Florida Bay Science Conference: A Report by Principal Investigators, October 17 & 18, 1995.
Boesch, D.F., Armstrong, N.E., D’Elia, C.F., Maynard, N.G., Paerl, H.W., and Williams, S.L. 1993. Deterioration of the Florida Bay Ecosystem: An Evaluation of the Scientific Evidence. Report to the Interagency Working Group on Florida Bay.
Boesch, D.F., Caffrey, J.M., Cloern, J.E., D’Elia, C.F., DiToro, D.M., and Walker, W.W. Jr. 1996. Florida Bay Nutrients: Perspectives on the July 1-2, 1996 Workshop.
D’Elia, C.F., E. Callender, W-S. Lung, and S.C. McCutcheon. 1996. Workshop on the Design and Specifications for the Flroida Bay Water Quality Model: Report of the Model Evaluation Group, Key Largo, Florida, 22-24 October, 1996.
D'Elia, C.F., J.G. Sanders, and D.G. Capone. 1989. Analytical chemistry for environmental sciences: A question of confidence. Env. Sci. Technol. 23:768-774.
Elder, K. 1994. MS. Thesis, Department of Civil Engineering, Clemson University, South Carolina.
Knap, A.H., Michaels, A.F., Dow, R.L., Johnson, R.J., Gundersen, K., Sorensen, J.C., Close, A.R., Howse, F.A., Hammer, M., Bates, N., Doyle, A., Waterhouse, T. 1993. BATS Methods Manual, Version 3. U.S. JGOFS Planning Office, Woods Hole, MA.
Lung, W.S. and C.C. Hwang. 1990. Integrating hydrodynamic and water quality models for the Patuxent Estuary. Estuarine and Coastal Modeling. Vol.1, pp.420-429.
Martin, J.L., Nirchardson,S.C. McCutcheon, and J. Paul. 1988. Water Res. Bull.
McCutcheon, S.C. 1989. Water Quality Modeling, Vol. I: Transport and Surface Exchange. CRC Press, Boca Raton, Florida.
Williams, S.L., K.H. Dunton, K.L. Heck, Jr., and W.M. Kemp. 1998. Report on the Florida Bay Seagrass Model Workshop, Key Largo, Florida, 12-14 January, 1998.
Presentations and Agenda
Florida Bay PMC-Model Evaluation
May 11, 1998
Knight Center, University of Miami
Purpose of Meeting, PMC Response to Recommendations of 1997 MEG Report;. — Tom Armentano, NPS/ENP; Co-Chair, PMC
Overview of PMC Modeling Program and Concepts for Integration — Mike Robblee, USGS/BAD; PMC Member
WES Lab Modeling: Bay Circulation and FEMWATER-ACOE/WES
Status report on related physical modeling and available physical oceanographic data — Tom Lee, UM/RSMAS
Status Report on Sediment Data Base — Ellen Prager, USGS/GD
Sea Grass Modeling Needs-Recommendations of Panel Reviewed Workshop — Mike Robblee and Mike Durako, UNC-Wilmington
Status Report on Hydrological Modeling in Lower Everglades Area — Eric Swain, USGS/WAD
Status Report on WES Lab Water Quality Modeling Plan — Mark Dortch, COE/WES
Status Report on Mass Balance Modeling Effort — Bill Nuttle, FIU
Status Report on Data Available for Water Quality Model Development and Testing —Joe Boyer, FIU and David Rudnick, SFWMD
Status Report on Water Column Food Chain Modeling — George Jackson, Texas A&M University
Upper Trophic Level Modeling Needs. Recommendations of Panel Reviewed Workshop — Nancy Thompson, NOAA/NMFS; PMC Member
Open Discussion on Issues. Plans and Problems — led by Model Evaluation Group
Appendix II. General Recommendations of Previous Workshops
(Compiled by Program Management Committee )