Last updated: 22 August 2014

Co-Sponsored by CaRA, GCOOS, GOMA, and SECOORA
14-16 October 2009, St. Petersburg, FL

The report can be downloaded as a PDF.

A workshop on ecosystem modeling in the Caribbean Sea, Gulf of Mexico, and Southeast Atlantic was held at the Florida Integrated Science Center of the U.S. Geological Survey in St. Petersburg, FL, on 14-16 October 2009. The workshop was sponsored by the Caribbean Regional Association (CaRA), Gulf of Mexico Coastal Ocean Observing System Regional Association (GCOOS-RA), Gulf of Mexico Alliance (GOMA), and Southeast Coastal Ocean Observing Regional Association (SECOORA). It was attended by a variety of modelers and users of results in ecosystem modeling whose names and affiliations are given in Appendix A. The agenda is given in Appendix B.

1.0 Executive Summary and Results

The overall goal of the Ecosystem Modeling Workshop was to bring together modelers to begin a dialogue on how to advance the state of ecosystem models of the Caribbean, Gulf of Mexico, and southeast Atlantic and the use of the model results by coastal managers, educators, and others. The objectives of the workshop were:

  • To discuss issues and problems that ecosystem models are expected to address;
  • To review ecosystem modeling activities underway in the Gulf of Mexico, Caribbean Sea, and southeastern U.S. coastal waters, including estuaries and bays;
  • To identify existing weaknesses, theoretical constraints, and needed advances in ecosystem modeling;
  • To identify and prioritize marine observations and products needed for ecosystem modeling; and
  • To advance a unified, coordinated program of ecosystem modeling for the region.

There were five deliverables targeted to be developed for this workshop. These were:

  • A list of scientific and environmental issues that ecosystem models are expected to address (Section 1.1),
  • An initial list of ecosystem modelers active in this region with brief descriptions of their interest areas (Section 1.2),
  • A prioritized list of observations and products needed by ecosystem modelers (Section 1.3),
  • Suggested pilot projects to advance the development of ecosystem modeling (Section 1.4), and
  • An approach to a coordinated program of ecosystem modeling for the CaRA, GCOOS, and SECOORA regions (Section 1.5).

The workshop was focused around six discussion themes:

Theme 1: Review and discussion of history and types of ecosystem models
Theme 2: Issues/problems that ecosystem models are expected to address
Theme 3: Space/time challenges of coupling physical and biological models
Theme 4: Identify existing weaknesses, theoretical constraints, and needed advances in ecosystem modeling
Theme 5: What observations or products are needed for marine ecosystem modeling?
Theme 6: How do we advance a unified, coordinated program for ecosystems modeling of the CaRA, GCOOS and SECOORA regions?

The discussions were wide-ranging, which is not unexpected given the myriad, complex systems for which ecosystem models are being developed. Here are given the five deliverables, which together provide the key conclusions of the workshop (Sections 1.1 through 1.5). The workshop process is described (Section 2). Brief summaries of the presentations and discussion reports are given, together with links to the presentations (Section 3).

1.1 Scientific and Environmental Issues that Ecosystem Models are Expected to Address

Ecosystem models of all types are tools that are used to help people understand how complex systems work. They enable people to identify what is not understood about a system and may illuminate ecosystem drivers that need more study. Models can be used to assist with the design of optimal observing systems. Many issues on which ecosystem models might be expected to inform were identified by the workshop attendees.

  • HABs (prediction, detection, tracking, forecasting)
  • Fisheries (stock assessments)
  • Protected species (habitat impacts)
  • Nutrients (transport, fate, effects)
  • Hypoxia (impacts, forecasts)
  • Coral health
  • Aquaculture impacts (pollution, more red tides, environmental effects)
  • Genetic issues – mixing of farmed with wild populations
  • Trophic degradation – assessing the status of and changes in trophic systems
  • Anthropogenic use of freshwater (flows & quality) on coastal ecosystems
  • Watershed management; land use
  • Ecosystem impacts (e.g., habitat change) from coastal inundation events (e.g., hurricanes, tsunamis)
  • Human health (e.g., pathogens, secondary effects (shellfish))
  • Connectivity (rate of exchange of individuals between separated populations and spatial pattern of populations)
  • Invasive species (shifting populations)
  • Marine Protected Areas
  • Coastal wetland habitat loss; wetland remediation
  • Ecosystem impact of platforms, structures, and artificial reefs (spills + habitat addition) — spatial impacts of marine “land use”
  • Defense and homeland security (e.g., effects of ecosystems on sound speed and optic characteristics of water)
  • Climate change impacts to ecosystems (these issues are well-suited to model application due to the long time horizons) including issues of
    • Effects of sea level change on ecosystems
    • Changes in the carbon cycle and the impacts of ocean acidification on ecosystems, coral communities
    • Role of vertically migrating species in carbon budget (basin-wide)
    • Multiple stressors
    • Impacts of climate change on connectivity
    • Drivers of shifting species distributions and associated ecosystem structure and function
    • Changes in HABs toxicity, species selection, bloom expansion ranges
    • Increasing severity of coral reef bleaching events

1.2 Ecosystem Modelers Active in the Caribbean Sea, Gulf of Mexico, and Southeast Atlantic Regions with Brief Descriptions of Their Interest Areas

To facilitate the exchange of information among all ecosystem modelers in the region, a list of ecosystem modelers active in the region encompassing the Gulf of Mexico, Caribbean Sea, and southeast Atlantic coastal oceanic regions, will be maintained on the GCOOS-RA web site (http://gcoos.org). For the initial list, each workshop invitee was asked to prepare a one-page information sheet giving contact and biographical information as well as current interests and projects. Others are welcome to provide their information using the template given below for provision of information. There are six parts to the information sought on the one-pager:

1) Name, affiliation, contact information
2) Picture [right-justified from (1)]
3) One-paragraph biographical sketch
4) Research interests (one paragraph)
5) Image showing area of research
6) Titles of current research projects

An example bio.

1.3 Observations and Products Needed by Ecosystem Modelers

The attendees generated a list of observations and products needed by ecosystem modelers. However, they were unable to prioritize the list because the priority depended entirely on the specific use planned for the ecosystem model and the spatial and temporal scales of the problem being addressed. For example, a HAB model for use by a coastal manager will need the circulation and the N:P:Si ratio coming from estuaries, but dissolved oxygen data would not be needed, whereas dissolved oxygen data are needed for hypoxia or fisheries model studies. Below are listed the observations and products identified. The list is not prioritized.

Observations Needed

  1. Temperature: 3-D fields (in situ and satellites)
  2. Salinity: 3-D fields (in situ and satellites soon)
  3. Currents: (in situ and satellites) – for quantitative comparisons with output
  4. Winds (in situ and satellites) – model forcing
  5. Surface heat flux – model forcing
  6. Nutrients (in situ)
  7. Light: multiple spectra (in situ and satellites)
  8. Airborne nutrient sources (in situ and satellites)
  9. River inflow, freshwater input
  10. Precipitation
  11. Dissolved oxygen (for hypoxia or fisheries studies, not HAB studies)
  12. Phytoplankton (biomass and distribution; daily)
  13. Zooplankton and fish eggs/larvae (biomass and distribution; daily; same frequency as phytoplankton)
  14. Fish
    1. foraging species distribution and abundance
    2. large predators (diet to good spatial/temporal resolution; who eats whom, how much, where)
    3. habitat information (spawning/nursery areas; connectivity of young with adult habitats)
    4. recruitment
    5. changes in catchability (availability, vulnerability constant in stable habitat)
    6. improved/more surveys (need improved infrastructure; need fishery independent surveying)
    7. multi-year, monthly/seasonal; put at a similar frequency to physical sampling (sampling as the same scales of importance)
  15. Genetic data

Products Needed

  1. Definitions of habitat and other terms
  2. Operational forecasts of societal-relevant state variables (besides hurricanes)
  3. Maps

Major gaps associated with data for ecosystem models

  1. There is no centralized data source, especially for nutrients and biological data. It was recommended that a coordinated data assembly and storage resource, including historial data, be established.
  2. There is no sustained, ongoing collection of ecosystem data sets. So ecosystem modeling of long-term or climatic responses is hindered. It was recommended that sustained data collection be developed.
  3. The critically important oceanographic data from satellites (e.g., ocean color, infra-red, scatterometry, microwave-salinity/aerosols, altimetry, gravity) are subject to potential interruption or elimination. Even if a mission is not eliminated, interruptions threaten the comparability of data, and thus the length of the time series, if there is insufficient overlap between missions so that data can be calibrated together.
  4. Additionally, satellite data products with specialized, regionalized algorithms are needed.

1.4 Suggested Pilot Projects to Advance the Development of Ecosystem Modeling

On the final day of the workshop, six potential pilot projects were identified for consideration by CaRA, GCOOS-RA, GOMA, and SECOORA. A seventh pilot project had been discussed under Theme 4 on day 2. These seven suggested pilots, together with the volunteers to prepare project scoping papers (white papers), are given below.

Theme 1: Review and discussion of history and types of ecosystem models

Pilot Project 1. Ecological data archival (white paper) – Steve Wolfe, Scott Cross, Paul Montagna, Matt Howard.
Ecological data archival must be improved. There is substantial data loss every year as ecological researchers retire or die. Need to capture the data that exist, as these give evidence of past conditions. Points to the need for an archive for ecological data sets – data rescue (data centers need information on who and where – pointers to the legacy data sets). A problem is eco-data are more a collection of stuff than “data” – NODC needs more resources and there needs to be a carrot for the investigator to archive data. Private companies doing contract work and hand in a report but not data – sense of proprietariness needs changing.

Theme 2: Issues/problems that ecosystem models are expected to address

Pilot Project 2. Demonstration of a decision-making tool for siting offshore aquaculture sites – Bill Arnold, Laurent Cherubin, Cindy Heil, John Quinlan, Mitch Roffer, Chris Simoniello, and Karen Burns (observer).
Consider the effects of offshore aquaculture on ecosystems and vice versa. Use IOOS ecosystem modeling in support of marine spatial planning for protection and preservation of ecosystems. Also helps to ID info that’s missing to do aquaculture planning well. Build the infrastructure.

Pilot Project 3. Add an NPZ ecosystem modeling component to the GOMA Nutrient PIT pilot studies – Paul Montagna, Steve Wolfe, Bob Weisburg, Ann Jochens.
Multidisciplinary.

Theme 3: Space/time challenges of coupling physical and biological model

Pilot Project 4. Pilot project for NCEP model assimilation of IOOS RA coastal winds – Rob Hetland, Bob Weisburg, Laurent Cherubin, Steve Morey, Jason Lenes, John Quinlan, Catherine Edwards.
Demonstration of the added value of incorporating IOOS coastal data sets into NCEP forecasting models. Data are available, but not being assimilated. Comparisons show that NCEP forecasted winds in places where COMPS data are available are usually a mismatch. This is a particular problem because sea breeze is an important ocean current forcing factor, especially in summertime in the Gulf of Mexico when winds are small. Show the utility of improved winds forecasts on ocean currents. Ecosystem impacts come in with, for example, the HABs human health component of sea breeze that can carry the aerosol component of HAB toxins or with the influences of sea breeze on the diel behaviors of biological species. (Wind team – Bob, Rob, Steve, Catherine); (Biology team-Jason, Steve, John, Catherine)

Theme 4: Identify existing weaknesses, theoretical constraints, and needed advances in ecosystem modeling

Pilot Project 5. Pilot project on the calibration and verification of complex physical-biological simulation models of coastal and estuarine ecosystems – Y. Peter Sheng and Robert Ulanowicz, Xinjian Chen.
During the discussion of Theme 4 on Day 2, the idea of conducting a project to examine complex physical-biological simulation models – a testbed – was suggested. Peter Sheng and Robert Ulanowicz have provided a summary of the suggestion (Appendix C).

Theme 5: What observations or products are needed for marine ecosystem modeling?

Pilot Project 6. Pilot project to couple Gulf estuaries to the larger-scale regional models – Rob Hetland, Peter Sheng, Bob Weisburg, Jerry Wiggert, Behzad Mahmoudi, Steve Morey.
Gulf estuaries are different in their connections to the coastal ocean than those of the far-north estuaries that have been studied. The project would tie in with the GOMA nutrient pilot studies. Could be made into a comprehensive demonstration project that, if shown to be valuable, could be scaled up with data from multiple data providers, particularly as nutrient sensors become more data intensive. Matagorda Bay, Apalachicola Bay, and Charlotte Harbor were suggested as estuary areas to target modeling efforts. Project also will help ID why different estuaries are different and information gaps to manage each more effectively.

Pilot Project 7. Pilot demonstration using fish catch data and 3-D ocean circulation – Mitch Roffer, Behzod Mahmoudi, John Quilan, Bob Weisburg, and Karen Burns (observer).
Pilot to show the interrelationship between the fish catch data from Vessel Monitoring Systems and the 3-D ocean circulation. Potential to resolve issues associated with the spatial scales (1° averages) in the fishery data. Potential of VMS data to be used to develop indices that help stock assessment.

Theme 6: How do we advance a unified, coordinated program for ecosystems modeling of the CaRA, GCOOS and SECOORA regions?

No pilot projects were identified for this theme except for those identified above.

1.5 An approach to a coordinated program of ecosystem modeling for the CaRA, GCOOS, and SECOORA regions

Because of the wide variety of ecosystem models, as well as a lack of time, a step-by-step approach to a coordinated program of ecosystem modeling was not delineated by the group. However, a number of issues were identified for which stronger solutions could be found by coordinated efforts of the four organizations sponsoring the workshop.

A. Ecosystem models are a tool that can link the Caribbean Sea, Gulf of Mexico, and southeast Atlantic regions to solve cross-regional issues. The ecosystems of these three regions are not independent. They are connected by the Yucatan Current, Loop Current, Florida Current, and Gulf Stream. Projects in these regions are usually funded as separate entities. So one task is to find a way to shift this paradigm of independent systems and move into one of dependent systems. A first step is to provide a set of products that will demonstrate the value of ecosystem modeling to managers and the public. These products exist, but need to be more widely disseminated.

B. Ecosystem models are not solely research tools that improve understanding of system processes. They have the capability to inform and thereby improve the decision-making process of agencies. Yet some coastal managers are not receptive to models. The reasons are unclear. The issue needs to be investigated, the problems identified, and solutions found. This would be a possible coordinated project for the four sponsoring organizations. What are the reasons for lack of use of ecosystem model results by managers? What improved capabilities are needed for models to be able to solve agency problems? How can the reliability of the model results be demonstrated to the various types of managers?

C. Ecosystem models have the potential to integrate information from the watershed to the estuary/bay to the coastal ocean to the deep waters. This capability may have uses for managers, such as looking at impacts of nutrient which come from sources in the watershed and then are transported throughout the coastal ecosystem. Is there a way to bring ecosystem modeling activities into one or more GOMA water quality/nutrient pilots as way of bridging the gap between monitoring agencies and modelers? Building the capacity of a coordinated observation – modeling system, where data are assimilated into fully coupled, nested models will require a high level effort to initiate. The successes of the Task Force on corals and GODAE might be useful examples to examine.

D. The sponsoring organizations should identify one issue that needs to be addressed and that can best be done with the integration that ecosystem models can provide. The example of may be applicable to getting an ecosystem modeling initiative launched.

E. David Green, NOAA NWS, suggested the organizations should develop the lists of observing needs using a format similar to the NOAA comprehensive lists of need. These specify for each issue the location, spatial and temporal resolution, parameter set, and the needed measurement accuracy. An example would be the PPBES process associated with the NOAA project on corals.

2.0 Workshop Process Summary

The workshop was called to order at 0830 on 14 October 2009 by Ann Jochens, GCOOS Regional Coordinator, on behalf of workshop organizing committee chair Worth Nowlin, who was unable to attend. She welcomed everyone and thanked the USGS for providing the facilities. After participants briefly identified themselves and their interests, Jochens reviewed the workshop objectives, agenda, and deliverables.

The workshop was focused around six discussion themes. Each theme was introduced by a speaker, and each discussion had a moderator and reporter. The final day was a plenary review and synthesis of key points from each discussion session. The role of the moderators was to stimulate discussion and keep it moving forward and focused. The role of the reporters was to record the key points uncovered in the discussion and to present these in plenary on the morning of the last day.

Participants were reminded to provide a one-page information sheet giving their contact and biographical information as well as current interests and projects. This information will be posted to the workshop web site as reference for possible future collaborations.

On 14 October, the first three themes were presented and discussed. These were:

Theme 1: Review and discussion of history and types of ecosystem models
Theme 2: Issues/problems that ecosystem models are expected to address
Theme 3: Space/time challenges of coupling physical and biological models

On 15 October, the last three themes were presented and discussed. These were:

Theme 4: Identify existing weaknesses, theoretical constraints, and needed advances in ecosystem modeling
Theme 5: What observations or products are needed for marine ecosystem modeling?
Theme 6: How do we advance a unified, coordinated program for ecosystems modeling of the CaRA, GCOOS and SECOORA regions?

A reception for workshop attendees was held at The Pier Aquarium the evening of the 15th.

On 16 October, summaries of the Theme discussions were presented by the reporter and discussed by the group with an effort to finalize the key conclusions. The workshop ended at approximately noon. Below are given summaries of the speakers’ talks and the reports from the reporters.

3.0 Presentations and Report Out by Theme

Theme 1: Review and discussion of history and types of ecosystem models

Speaker: Clair B. Paris-Limouzy (Division of Applied Marine Physics, Rostenstiel School of Marine & Atmospheric Sciences (RSMAS), University of Miami)
Moderator: Bill Arnold (Marine Fisheries Biology Department, Fish and Wildlife Research Institute (FWRI), Florida Fish and Wildlife Conservation Commission (FWC))
Reporter: Ann Jochens (GCOOS-RA, Department of Oceanography, Texas A&M University (TAMU))

Dr. Paris-Limouzy provided the overview talk on ecosystem models. She began with a definition of what an ecosystem model is: “A simplified representation of complex ecosystems aimed at characterizing their major dynamics and predicting their behavior.” She described the various types of models–verbal, schematic, physical, mathematical, and numerical models–and summarized how models are developed, including holistic, predictive, inferential, mechanistic, and hypothesis-generating approaches. In the 1920s, models for population growth and prey/predator growth rates were developed. The next major step was to model a 3-component ecosystem model, and the NPZ (Nutrient, Phytoplankton, Zooplankton) models were born. After reviewing the history of ecosystem models, Paris-Limouzy then described the current status of ecosystem models, including 3-D, 7-compartment NP (nutrients & phytoplankton) models, spatially-explicit Lagrangian models, and static ecosystem models. She then addressed the issue of where the modelers want to be in the future. Leading the list of areas that need additional development are: working hand-in-hand with the observationalists collecting field data or data from the laboratory, assimilating data into models, and multi-scale modeling.

Theme 1 Report Out

Key Conclusions

  1. Scales depend on the purpose of the model; once purpose is known then can determine what will need:
    • Ecosystem issues (e.g., connectivity): Caribbean – Gulf of Mexico – Southeast Atlantic coasts
    • Regional HABs, hypoxia, fisheries issues
    • Local estuary/bay nutrient impact issues

    ACTION: Determine the key questions over the next 4-5 years that will be critical, then build models around these – coastal managers interactions with modelers (what issue HAS to be addressed NOW so it rises to high enough level to fund)

  2. Need new instruments that measure the biota to improve the quantity of good quality biological data for the ecosystem models
  3. Improve modeling to be able to link at boundaries; Need biophysical models that include nesting (e.g., to get exchanges between estuary & open ocean)
  4. Data issues
    1. Need sustained, long-term (decades) data sets covering the full suite of parameters needed for the ecosystem model to address the key question being modeled (e.g., physical + nutrients + phytoplankton + zooplankton + eggs/fish larvae + fish + benthos)
    2. Ecological data archival must be improved: Data loss every year (retirement) – capturing data that exist (evidence of past conditions) – need for an archive for ecological data sets – data rescue (data centers need information on who and where – pointers to the legacy data sets) – problem is eco-data are more a collection of stuff than “data” – NODC needs more resources and there needs to be a carrot for the investigator to archive data; private companies doing contract work & hand in a report but not data – sense of ownership
    3. Standardization of data if appropriate; much ecological data is not standard, but must have good metadata (good data management training) – community organize the requirements so data center can pursue – possible pilot to find a way to change the culture
  5. Improve communications:
    • Cross-talk between physical and biological partners
    • Explain to decision-makers (agency heads, legislatures) the role of data and models
    • Explain the return on investment of both data and models
  6. Need improved quantitative skill assessment on where we realistically are with EBMs

Suggested pilot projects to advance the development of ecosystem modeling

Ecological data archival (white paper) – Steve Wolfe, Scott Cross, Paul Montagna

Next Steps
What can we run now
What infrastructure is needed
What needs to be done to get there
Demos might get higher level interest


Theme 2: Issues and problems that ecosystem models are expected to address

Speaker: Brad Penta (Ocean Sciences Branch, U.S. Naval Research Laboratory (NRL))
Moderator: David Green (Marine and Coastal Weather Services Branch, Office of Climate, Water, and Weather Services, National Weather Service, National Oceanic and Atmospheric Administration (NOAA))
Reporter: Jerry Wiggert (Department of Marine Science, University of Southern Mississippi)

Dr. Penta provided information on the ecosystem modeling activities of the NRL. The Navy is involved in a wide variety of activities related to modeling. It collects data, does QA/QC, performs correlations, and brings the information into the modeling system. It has global, regional, small-scale, coastal, tide, wave, and surf models. The Navy does bio-physical modeling, sediment modeling particularly related to the effects of sediment on optics, and models to predict acoustic sonobuoy trajectories. All are aimed at use by or for the war fighter. Penta described the NCOM and HYCOM models. Satellite data are assimilated, and historical T-S relationships are used to simulate vertical profiles. The operational global ocean forecasting model was developed over ten years (1998-2008). Nesting is used to go from the global model to smaller scales.

The ecosystem models are embedded in and coupled to regional and smaller, nested models. The ecosystem models nested in the NCOM physical model are very simple. However, the Navy is working on an ecosystem model related to hypoxia. It is a box model with capabilities for looking at remineralization in the sediments (dissolved oxygen, dissolved inorganic carbon, nitrogen, phosphorus). It is based on the work of the late Dr. Peter Eldridge. The model includes zooplankton, 6 phytoplankton species, nutrients, organic matter remineralization, and sediment. The physical model output (e.g., currents) is combined with chlorophyll. At present, the Navy is working to improve the skill of the model.

Mr. Wolfe provided the perspective of a user of model output in the areas of water quality regulatory and toxicology issues. The goal is for the models to help coastal managers make better decisions. Managers have a hierachical understanding of ecosystem models, i.e., a hydrodynamical model with chemistry equates to a water quality model; add in biology to get an ecosystem model. Because they are looking at ecological risks, coastal managers need applied models, not theoretical models. They judge how good a model is by how useful it is to address problems and improve decision-making, not necessarily how accurate the model is. For this reason, there can be a tendency for some managers to be too willing to depend on model results.

Ecosystem models are needed by coastal managers to explain why the biology is the way it is and what can be done to resolve issues of changing variability. The Gulf of Mexico Alliance has identified its ecosystem model needs. Scott Cross has the model data base. The GOMA needs where models might be beneficial are:

  1. Improved understanding of nutrient effect on coastal biological communities – motivated by requirement that each state has to promulgate nutrient criteria for the fate, transport, and effect on biological communities. Ecosystem models can elucidate which nutrients and levels are harmful.
  2. Determine the pathways by which mercury gets into ecosystems. Ecosystem modeling is a key way to examine this issue.
  3. Improved forecasts for HABs are needed all around the Gulf for several species of which many are close to coast.
  4. A framework for the build out of the coastal monitoring system that GOMA is developing needs to include both coverage of what modelers need and the use of the models to help determine sensor placement.

As moderator of the discussion, David Green began with a short presentation on the issue that society depends on ecosystem modeling to predict threats and minimize risk. He explained that NOAA is working to make modeling a strategic activity. Identification by this group of the top 3-5 areas where ecological models can contribute could assist NOAA in its ecosystem modeling strategy.

Theme 2 Report Out

Issues & Problems that ecosystem models are expected to address

Identify Primary Problems of Interest

  • Divergence is likely between workshop attendees, resource managers, modelers, federal agencies
  • Can common ground or a rallying point be identified that effectively promotes coordinated effort/organization of the research community?

The NOAA Epiphany:
Ecosystem Models Are Important!

  • Past support of modeling activities and promotion of desired capabilities has been ad hoc
  • Need to develop a roadmap for how ecosystem models fit within a coordinated programmatic effort
  • How to promote use of models as a strategic tool?
  • Ecosystem models could help guide prioritization of research needs and sampling design, but this is not done in a coherent manner

Needs & Ramifications of Ecosystem Modeling

  • Predicting hypoxia is insufficient – also need to know:
    • Informed interpretation of ecological & resource response to model-derived guidance on hypoxia distributions
      • Habitat reduction, benthic organism death, reduced growth rate/catch of shrimp
    • Effects on human health, trade
  • Quantify model skill; Apply consistent metrics
  • Broader implications
    • Jobs, national security and defense (SLD, track submarines)
    • Promote collaboration w/ international partners –> foster better diplomatic relations (e.g., Cuba)

Appropriate Utilization

  • Is it currently practical to employ ecosystem models in a predictive capacity?
  • Scenario exploration a more useful/productive pursuit?

Issues on which ecosystem models might be expected to inform

  • HABs
  • Fisheries (stock assessments)
  • Protected species (habitat impacts)
  • Nutrients
  • Hypoxia
  • Coral health
  • Aquaculture impacts (pollution, more red tides)
  • Genetic issues – mixing of farmed with wild populations
  • Trophic degradation – status of trophic systems
  • Anthropogenic use of freshwater (flows & quality) on coastal ecosystem
  • Watershed management; land use
  • Ecosystem impacts (e.g., habitat change) from coastal inundation events (e.g., hurricanes, tsunamis)
  • Human health (e.g., pathogens, secondary effects (shellfish))
  • Connectivity (rate of exchange of individuals between separated populations and spatial pattern of populations)
  • Invasive species (shifting populations)
  • Marine Protected Areas
  • Coastal wetland habitat loss; wetland remediation
  • Ecosystem impact of platforms, structures, and artificial reefs (spills + habitat addition) — spatial impacts of marine “land use”

Climate Change Impacts
(Well-suited to model application)

  • Effect of sea level change on ecosystems
  • Impacts of ocean acidification on ecosystems, coral communities
  • Role of vertically migrating species in carbon budget (basin-wide)
  • Multiple stressors
  • Impact on Connectivity
  • Driver of shifting species distributions and associated ecosystem structure and function
  • Changes in HABs toxicity, species selection, bloom expansion ranges
  • Increasing severity of coral reef bleaching events

Final Thoughts

  • How to reduce to a singular recommendation?
  • Target subset of these interests/needs that can most effectively address various interrelated issues (i.e., maximize bang for buck)
  • Could focus on chemical measurements provide this unification?
  • What are mandated needs already slated for acquisition?
  • Identify what is already being addressed that could be leveraged, and could help in identifying what further observations are crucial needs
  • Gap between scientific interests and public interest/awareness of ecosystem model utility
  • How to translate to educating managers, users, stakeholders, politicians & public on the potentials of ecosystem modeling?

Theme 3: Space and time challenges of coupling physical and biological models

Speaker: Paul Montagna (College of Sciencee & Technology, Texas A&M University-Corpus Christi (TAMU-CC))
Rob Hetland (Department of Oceanography, TAMU)
Moderator: Matt Howard (GCOOS-RA, Department of Oceanography, TAMU)
Reporter: Laurent Cherubin (Division of Meteorology and Physical Oceanography, RSMAS, University of Miami)

Dr. Rob Hetland began the discussion of space-time challenges to coupling physical and biological models to build ecosystem models with the view of a physicist. He identified HABs and Fisheries as two important ecosystem issues in the Gulf. HABs generally "go with the flow" but fish can swim so currents cannot be used to track them. Oceanographic space and time modeling issues are a resource allocation problem. Just as there are trade offs in coverage at various space and time scales for making observations, so there are with models. There are high computational penalties to resolve the small scales. The goal is to optimize efficiency.

It is important to know the distinctions between theoretical models, which are causal and based on first principles, and statistical models, which are not causal and are not models, but expressions of a relationship (e.g., regression correlation analysis). The type of study will determine which should be used and what challenges will be faced. Regression is a baseline for the deterministic model, but the model itself needs to be better than the regression. The relationships between space and time scales are elucidated by the theoretical models. If space-time scales are reduced, then the problem of subgrid processes is encountered the subgrid processes problem; i.e., the impacts of processes that are at time or space scales smaller than model resolution will be missed. Subgrid scale parameterizations, which are statistical representations, are used to represent how the small scale processes will influence the large scale processes.

Dr. Paul Montagna presented the view of a biologist. What makes ecological modeling difficult are data. Physical oceanographers can collect lots of data, e.g., from a moored buoy, but biological oceanographers do not have this capability yet. Biology comes down to old-fashioned field hand-sampling. For the biologist’s temporal resolution, one measurement per month would be rich; imagine what this would do to the physicist’s model! Spatial resolution in biology is highly variable (e.g., no matter how much is sampled, don’t get a good estimate of population). One space-time issue is to determine at what scales to model; this is highly dependent on the process being studied. Another issue is how to model at the scales for which there are insufficient data. Yet another issue is how to calibrate and validate the model when the data may not clearly measure what is happening (e.g., what do fluorescence data inshore really tell us; when everything in brown waters – organic matter, humic matter, bacteria, phytoplankton – is fluorescencing, how can fluorescence be taken as a true measure of chlorophyll?). Additionally, in biology it is very difficult to develop tested parameterizations, in part due to the paucity of data.

Theme 3 Report Out

Topics: sub-scale biology, swimmers modeling, behavior, agent-based modeling

Physical and biological models have compatible time scales in theory

Though each type of data used to feed coupled models have different time scales characteristics

Why? The definition of synopticity is not equivalent between fields because of behavior, biological interaction, chemical processes. In-situ data don’t match with model grids

Remedies:

  • parameterization of sub-scale processes (biological and physical; temporal and spatial)
  • Inclusion of behavior with limitation by the type of organisms (plankton, larvae, fish)
  • Statistical correlations

Requirements:

  • Capacity of the models to fill in the gaps (nesting)
  • Model mechanistic should compensate for the gaps in the observation fields
  • Cope with the patchiness of the biological fields

Limitations: empirical parameters (organism mortality), empirical correlations, biology driven physics, in-situ sampling methods

Based on the issues we try to address, we have to find the scales that are relevant to the processes at stake and that will be sufficient to yield a realistic outcome. Therefore sampling should be done at scales that encompass the model grid size (>x2).

  • ABM: sub-scale movement added to the behavior stochastically
  • Physical models have adequate time scale to resolve biological models. However, information at short time scale is not always available. But the physics could be average to match those scales.
  • Same modeling limitations for physical and biological models
  • But problem of sampling synoptically the biology (e.g., plankton patches). Depends also on the target species and what you know about you want to simulate.
  • Limitation: empirical parameters (fish mortality)
  • NPZD model:
    • Validation: counting phytoplankton and zooplankton with ADCPs. Enables closure of the models. Provide information on the geometry of the patches that will drive the encounter rates.
    • Limitation by grid size in both directions but also by the question we want answer. Data points don’t necessarily fall in one of the model’s layers
  • Limitation of the physics by the biology: mixing by animal propulsion in the ML. Stabilization of bottom sediment by benthic organisms. Feedback of ecosystem dynamics on SST (plankton in the ML -> HYCOM)
  • Feedback of satellite measurements to ocean numerical (Chla) to test model physics
  • Lack of processes representation: tide, model data sampling
  • Pilot project: blue fin tuna survivorship based on environmental parameters: conditions for recruitment success

Theme 4: Identify existing weaknesses, theoretical constraints, and needed advances in ecosystem modeling

Speaker: Robert E. Ulanowicz (Chesapeake Biological Laboratory, University of Maryland; Arthur R. Marshall Laboratory, Department of Botany and Zoology, University of Florida)
Moderator: Peter Sheng (Division of Coastal & Oceanographic Engineering, University of Florida)
Reporter: Brian Keller (Office of National Marine Sanctuaries, NOAA)

Dr. Robert E. Ulanowicz began with an overview of the state and usefulness of ecosystem models of different types, which perform with varying levels of success in cases with a single biological process or species. He then explored the issue that “Given the very limited successes of multi-species simulations, we are justified in asking whether the entire paradigm of “nature as a clockwork” is appropriate or even counter-productive in the long run?” He recommended that “In order to understand living systems emphasis should shift away from fixed laws and towards the description of process” and then explained the development of ecological network analysis to analyze systems that are too complex to readily simulate.

Theme 4 Report Out

Speaker: Robert E. Ulanowicz, University of Florida

  • Ascendancy: how effectively resources are processed by system – tends to increase in absence of perturbations
  • Ascendancy only half of story – robust systems balance constraint and flexibility, i.e., have intermediate degree of order

Moderator: Peter Sheng, University of Florida

Goal of Ecosystem Modeling – Develop models that can be used to synthesize data and solve problems:

  • Hypoxia
  • HABs
  • Fishery
  • Climate Change Impact
  • Extreme Event Impact

Weakness of existing ecosystem models (fisheries discussion)

  • Inadequate coupling among various scales, e.g., basin- and estuary-scale
  • Basin-scale models are not coupled to estuarine and watershed models
  • Inadequate model verification and skill assessment
  • Modeled processes are not well understood
  • Model coefficients are site-specific with large uncertainty

Weakness of existing ecosystem models (sediment transport discussion)

  • Sediment significantly affects nutrient transport and ecological processes
  • Need high spatial and temporal resolution to capture the energetic events – this would help improve the skill of the sediment transport model and water quality model

Weakness of existing ecosystem models (hypoxia discussion)

  • Inadequate data and understanding of certain important processes, e.g., sediment oxygen demand
  • Need to have modelers & data collectors develop a study where they work together on full ecosystem understanding and model

Theoretical constraints

  • Mechanistic models may not work well because biology is much more complex than physics
  • Can enhance mechanistic models by applying network analysis
  • Users are managers, planners, public – although we won’t have a perfect model, we need to provide prediction, characterization, etc., of the issue being addressed

Needed advances

  • Need to identify a set of overarching science questions (physical and biological) to answer when developing ecosystem models to address any of the five specific problems (hypoxia, HAB, fishery, climate change impact, extreme event impact)
  • Develop a testbed to assess skill and conduct sensitivity tests of ecosystem models
  • Resolve the important temporal and spatial scales of processes
  • Need to have high-resolution fishery-independent data

Theme 5: What observations or products are needed for marine ecosystem modeling?

Speaker: John Walsh (College of Marine Sciences, University of South Florida)
Moderator: Behzad Mahmoudi (FWRI, FWC)
Reporter: Mitch Roffer (Roffer’s Ocean Fishing Forecasting Service, Inc.

Dr. John Walsh titled his talk "To muddle or model is the question: Reconciliation of biological and physical approaches to ecosystems analysis, confronted with observations from human-impacted coastal habitats, subject to overfishing and eutrophication: A prospectus for Penultimate systems analyses of RED tides of Karenia brevis over the next decade, as a precursor to their operational forecasts from Brownsville to Cape Hatteras." He used the experiences of Florida’s joint USF/FWRI Center for Prediction of Red Tides (CPR), which has the goal of developing an operational model to forecast Harmful Algal Blooms in the eastern Gulf of Mexico, to illustrate his points about ecosystem modeling and observations and products needed. Examples drawn from the CPR to elucidated the need of ecosystem modeling for observations and products of many different types, biological and physical. For examples, products needed are operational forecasts of societal-relevant state variables, not just hurricanes, and observations needed include zooplankton time series for model closure of lower trophic levels. Beyond just improving the balance of physical and biological sampling in funded projects, improving the fidelity of the coupled biophysical models themselves also is needed. A major stumbling block to proceeding with systems analysis on the continental shelf is that we cannot predict the movement and behavior of fish and zooplankton. But the greatest need to move ecosystem modeling forward is for true multidisciplinary environmental programs with physics and biology funded in equal measure to raise the next generation of resource managers and systems analysts, versed in quantitative tools for prudent harvests of shelf biota.

Theme 5 Report Out

Behzad Mahmoudi – Florida Wildlife Research Institute
Mitchell A. Roffer – ROFFS™ Roffer’s Ocean Fishing Forecasting Service, Inc.

THEME 5:WHAT OBSERVATIONS OR PRODUCTS ARE NEEDED FOR MARINE ECOSYSTEM MODEL

  • “TO MUDDLE OR MODEL IS THE QUESTION”
    • We did both
  • ECOSYSTEM MODELS TO HELP DECISION MAKERS
  • MODELS – NUTRIENTS – BIOLOGY

MODELS

  • HABs, Hypoxia, Fishery => We are One
  • We need requirements?
    • 3D fields – Maps – Data
      • What spatial and temporal resolution?
      • No specifics but vary to science question
        • Nyquest Frequency – Rosby Radius of Deformation
  • Local model nested to larger model
  • HYCOM – global and Gulf of Mexico regional
  • Nested models
  • WFS ROMS, NGOM, WFS FVCOM, WFS HYCOM, NCOM, others …
  • French Gulf model, Industry Gulf model, Danish Hydraulics Model™, Nowcasting™ Model
  • Models form other areas
  • Relevancy to the coastal ocean
  • Publically available
  • Operations (24-7) vs Operating Models
    • A model is available that has known quality of data, metadata, a foundation, a model that will be the focus for improvements, not necessarily running 24/7, an implied community of users
  • Estuarine models
    • Models exist and are operating for specific area and applications (e.g., St. Johns Estuary, Naples)
      • GAP: overall lack of estuarine models in the Gulf of Mexico
      • GAP: Coupling estuary and ocean models
      • GAP: Water quality models
    • GAP in validating – quantifying uncertainty – precision – accuracy

NUTRIENTS

  • Nitrogen, phosphorus, and silicon
    • N:P:Si ratio
    • GAP: no coordinated data base available (N, P, Si) – data are at local, state, fed, researchers
    • GAP: no sustained systematic, coordinated sampling and storage of data; units, methods highly variable)

BIOLOGICAL DATA

  • Zooplankton and fish eggs/larvae
    • biomass & distribution; sampled daily; same frequency as phytoplankton
  • Fish
    • foraging species distribution and abundance
    • predator species distribution and abundance
    • GAP: fish diet: who eats who
  • Habitat information:
    • structure and function of benthic & pelagic habitat
    • It is also temperature, chlorophyll variable – pelagic habitat
  • Spawning/nursery areas;
    • connectivity of young with adult habitats;
  • Recruitment forecasts (critical need by managers)
  • Changes in catchability (availability, vulnerability, catchability- q): (critical need by managers)
    • Develop indices for population assessment
  • Improved surveys – more surveys (need improved infrastructure;
  • GAP: need fishery independent surveying;
    • Spawning sites and other critical areas (e.g. where foraging is critical for red tide)
  • GAP: biological data sampled at similar/same resolution to physical sampling
    • (sampling at same scales of importance)
  • GAP: fishery dependent data at high resolution
    • Exact location or that which is needed to understand relationships between the fish and the environment
  • GAP: Genetic data
    • for subpopulation analysis
    • Forensic oceanography
  • GAP: Environmental indices
    • To be used in stock assessments (recruitment, catchability)

DATA NEEDS

  • Temperature: 3-D fields (in situ and satellites*)
  • Salinity: 3-D fields (in situ and satellites soon)
  • Currents – (in situ and satellites)
    • for quantitative comparisons with output
  • Winds (in situ and satellites)
  • Surface heat flux
  • Nutrients (in situ)
  • Light: multiple spectra (in situ and satellites)
  • Airborne nutrient sources (in situ and satellites)
  • River inflow, freshwater input
  • Precipitation
  • Dissolved oxygen (hypoxia/fish, not HAB)
  • *Oceanographic data from satellites is not as guaranteed but should be deemed critical. Data products with specialized, i.e. regionalized algorithms are needed

Other

  • List of data with resolutions
    • NOAA has comprehensive lists of need; Location, spatial/temporal resolution, parameter set, measurement accuracy (David Green) – CORAL  PPBES process – NOAA website
  • Climate change
  • Definition of Habitat

SOS


Theme 6: How do we advance a unified, coordinated program for ecosystems modeling of the CaRA, GCOOS, and SECOORA regions?

Speaker: Bob Weisberg(College of Marine Science, University of South Florida)
Moderator: Ann Jochens (GCOOS-RA, TAMU)
Reporter: Chris Simoniello (GCOOS-RA, Department of Coastal Studies, University of Southern Mississippi)

Dr. Bob Weisberg stressed that paradigm shifts are needed by both management and scientists. Scientists must come to realize that ecology is not just biology. Managers must learn that ecosystem based management is not just fish. Further the concept of independent "Large Marine Ecosystems" must be shifted to realize the dependencies of any system on adjacent and far field systems. For example, the Caribbean Sea, Gulf of Mexico, and southeast Atlantic coastal systems are all part of the same ecosystem. This calls for a coordinated approach to the estimation of all relevant coastal ocean state variables – whether physical, biological, chemical, or other – with science-based placement decisions. It also calls for a cooperative, coordinated, non-competitive approach by CaRA, GCOOS-RA, and SECOORA toward the development of the IOOS regional associations. Among his recommendations were to distribute resources in a manner that is complete enough to be useful and to provide a backbone on which to build in the future as well as to disseminate the information on the existing products from all sources that will capture the public’s attention.

Theme 6 Report Out

Ecosystem Modeling Workshop
Co-Sponsored by CaRA, GCOOS, GOMA, and SECOORA
October 14-16, 2009
St. Petersburg, FL

  • KEY QUESTIONS
    How do we advance a unified, coordinated program for ecosystems modeling of the CaRA, GCOOS and SECOORA regions?
  • LME is an empty set of words-e.g., Caribbean, GoM, SE-not independent-connected by Loop Current, FL Current, Gulf Stream-currently funded as separate entities.
    Requires a paradigm shift-how do we achieve this?

How do we push for science based placement to get at the understanding of systems?

WHAT ARE PRIORITY ACTIONS and DATA SETS

Provide a set of products that will capture the publics’ attention-we have them, but the public is unaware of the value

Need to highlight success stories…why was 2006 and 2007 years so different for red tides…generated new knowledge. How can we leverage this to educate managers?

  • Demonstrate there is more data and readily available as a result of us coming together.
  • TB program, why were they not receptive to models? IF they’re not, how are we set up for everyone else? Identify the problem.
  • Model(s) needs to solve agency problems. Help improve decisions as much as understanding process.
  • Ability to demonstrate model is reliable.
  • Need to meet with enough people at a high enough level and have these candid discussions. Determine if there is serious intent.

WHAT ARE PRIORITY PRODUCTS?

  • How many models integrate from FW (watershed) to estuary to coastal ocean to offshore? Zero. Make it obvious to agencies what we can contribute.
  • Huge gap in understanding about how the data can be used.
  • Is there a way to bring modeling activities into one or more GOMA water quality/nutrient pilots as way of bridging the gap between monitoring agencies and modelers?
  • GOMA will fund the research, but it’s not coordinated around state-individual grants to conduct work locally.
  • Need to communicate in simple terms what we are discussing here to others; topic not sexy enough-need to really sell.

SLIDE 5:

  • Use Task Force on corals as a model to get people at higher levels engaged-create a central data base-use that initiative to launch this system.
  • Simplify-think about one goal-something big, obvious, involves everyone and NEEDS TO BE DONE. Caveat-demonstrates that the problem cannot be solved without integration of data and collaboration.

 


 

Appendix A: Ecosystem Modeling Workshop Attendees

Name Affiliation 14 15 16
 
Bill Arnold FL Fish and Wildlife Commission
Karen Burns Gulf of Mexico Fishery Management Council
Dave Chagaris FWC  
Xinjian Chen Resource Projects Dept., SW Florida Water Management District  
Laurent Cherubin* University of Miami, RSMAS/MPO
Scott L. Cross NOAA/NESDIS/NCDDC at Center for Coastal Environmental Health and Biomolecular Research
Catherine R. Edwards Florida State University  
Nicholas A. Farmer NOAA National Marine Fisheries Service SE Regional Office    
Graciela García-Moliner Caribbean Fishery Management Council  
David Green NWS Lead for Ecosystem Forecasting and Coastal Hazards
George Halliwell RSMAS, Division of Meteorology and Physical Oceanography
Cindy Heil* FL FWRI/FWC
Robert Hetland* Oceanography, TAMU  
Eric J. Hochberg National Coral Reef Institute, NSU Oceanographic Center    
Matthew K. Howard TAMU
Ann Jochens TAMU
Brian D. Keller NOAA Office of National Marine Sanctuaries
Villy Kourafalou RSMAS, Division of Meteorology and Physical Oceanography    
Kristen Laursen NOAA    
Jason M. Lenes University of South Florida, College of Marine Science
Behzad Mahmoudi Florida Fish and Wildlife Research Institute
Paul Montagna* TAMUCC
Steve Morey FSU  
Woody Nero NOAA Fisheries  
Claire B. Paris-Limouzy RSMAS, Division of Applied Marine Physics    
Brad Penta Naval Research Laboratory, Ocean Sciences Branch  
John Quinlin NOAA, Stock assessment group at SEFSC in Miami
Mitchell Roffer Roffers Ocean Fishing Forecasting Service  
Peter Sheng University of Florida  
Chris Simoniello GCOOS Education and Outreach Coordinator
Robert E. Ulanowicz Chesapeake Biol. Lab, University of Maryland, U. of Florida    
John Walsh* University of South Florida  
Bob Weisberg* USF College of Marine Science    
Jerry Wiggert Univ. Southern Mississippi
Steven Wolfe Florida Department of Environmental Protection
Roger Zimmerman NOAA Fisheries Galveston

*Workshop Organizing Committee Member; members who were not able to attend were Worth Nowlin, Naseer Idrisi, Ruoying He, and Gil Rowe.

 


 

Appendix B

Ecosystem Modeling Workshop Co-Sponsored by CaRA, GCOOS, GOMA, and SECOORA
October 14-16, 2009, St. Petersburg, FL

Provisional Agenda 6

14 October 2009
8:00 COFFEE and Juice
8:30 Opening
Welcome
Introductions
Objectives and Deliverables
Adoption of agenda
9:30 Speaker to introduce Theme 1: Review and discussion of history and types of ecosystem models–Claire B. Paris-Limouzy (RSMAS) and Nasseer Idrisi (University of the Virgin Islands)
10:15 COFFEE
10:30 Plenary discussion of Theme 1.
Moderator–Bill Arnold
Reporter–Ann Jochens
12:00 LUNCH – Lunch provided at the meeting
13:00 Speaker to introduce Theme 2: Issues/problems that ecosystem models are expected to address–"Some Navy uses of environmental models", Pat Hogan and Brad Penta (U.S. Naval Research Laboratory); "Perspective of a GOMA State Manager"
13:45 Plenary discussion of Theme 2
Moderator–David Green
Reporter–Jerry Wiggert
15:00 REFRESHMENT BREAK
15:15 Speaker to introduce Theme 3: Space/time challenges of coupling physical and biological models–Paul Montagna (TAMU Corpus Christi) and Rob Hetland (TAMU)
16:00 Plenary discussion of Theme 3
Moderator–Matt Howard
Reporter–Laurent Cherubin
17:15 Review schedule for Day 2
17:30 ADJOURN FOR DAY
 
October 15, 2009
 
8:00 Coffee and juice
8:30 Resume Workshop
Introduction of Speaker
8:45 Speaker to introduce Theme 4: Identify existing weaknesses, theoretical constraints, and needed advances in ecosystem modeling (Robert E. Ulanowicz, University of Florida)
9:30 Plenary discussion of Theme 4
Moderator–Peter Sheng
Reporter–Brian Keller
10:00 COFFEE
10:15 Continue plenary discussion of Theme 4
11:00 Speaker to introduce Theme 5: What observations or products are needed for marine ecosystem modeling?
(John Walsh, University of South Florida will speak on "To muddle or model is the question: Reconciliation of biological and physical approaches to ecosystems analysis, confronted with observations from human-impacted coastal habitats, subject to overfishing and eutrophication"
12:00 LUNCH – Lunch will be provided at the meeting
13:00 Plenary discussion of Theme 5
Moderator–Chris Simoniello
Reporter–Matthew Howard
14:15 Speaker to introduce Theme 6: How do we advance a unified, coordinated program for ecosystems modeling of the CaRA, GCOOS and SECOORA regions? (Bob Weisberg, University of South Florida)
15:00 REFRESHMENT BREAK
15:15 Plenary discussion of Theme 6
Moderator–Ann Jochens
Reporter–
17:00 ADJOURN FOR DAY
19:00 RECEPTION at the Pier Aquarium
 
October 16, 2009
 
8:00 Coffee and Juice
8:30 Resume Workshop
Review schedule for Day 3
8:45 Present, discuss, and finalize Key conclusions from Theme discussions
Moderator–
Reporter–Chris Simoniello
10:00 COFFEE
10:15 Continue considerations of key conclusions
11:15 Closing remarks
11:30 ADJOURN WORKSHOP

 


 

Appendix C: The Calibration and Verification of Complex Physical-Biological Simulation Models of Coastal and Estuarine Ecosystems

Y. Peter Sheng and Robert Ulanowicz
University of Florida
Gainesville, FL 32611-6580

With advances in computer technology, it has become possible to create very large and complex 3-dimensional models of the physics, chemistry and biology of coastal and estuarine ecosystems [1]. Due to the existence of multiple temporal and spatial scales and manifold processes, the calibration and verification of these complex physical-chemical-biological models require huge amounts of data and are very tedious. Typically hundreds or thousands of model simulations are required to arrive at a “baseline” simulation. We are proposing a pilot project to demonstrate the use of ecological network analysis (ENA) as a robust and systematic tool to facilitate the calibration and verification of complex simulation models. We propose to demonstrate the method on an existing coupled physical-chemical-biological model of a Gulf region estuary (e.g., Charlotte Harbor) [2].

During the calibration and verification processes, it isn’t always easy to trace the origins of any deviant behavior. Changes in any single parameter are likely to affect matters elsewhere far removed from the process associated with the parameter. Some systematic way of approaching the task of calibration/verification is needed. Although inverse methods can be used to determine the key model parameters [3], a more promising and robust approach might be to use ENA which was developed to meet the challenge of analyzing systems that were too complex to readily simulate [4]. This toolbox consists of a suite of mathematical methods from linear algebra and information theory that allow the user to trace all direct and indirect connections, assess trophic status, delineate the pattern of cycling, identify the kinetic bottlenecks [5] and evaluate overall system status.

It is a straightforward exercise to modify the simulation code to provide a “snapshot” of all the transfers and stocks occurring at any instant (or a “movie” sequence of such networks.) Then, whenever a particular stock or flow in the model behaves poorly, one can analyze the corresponding snapshot to explore the afferent and efferent environments that may be influencing such pathology. Furthermore, one can examine nearby “bottlenecks” (limiting processes) to examine what is controlling the phenomenon in question, etc.

ENA even has the potential to guide the inverse problem in analogy to how the maximization of entropy process (MEP) has been employed for the same purpose. Real, sustainable ecosystems, for example, appear to maintain a balance between order-building and entropic tendencies (both of which can be quantified using information theory.) It is now possible to calculate how each process and stock should be changed to move the system toward that balance [6], thereby providing clues to how parameters should be modified.

  1. Sheng, Y.P. and Kim, T., 2009. Skill Assessment of an Integrated Modeling System for Estuarine and Coastal Ecosystems, J. Mar. Sys. 76:212-243.
  2. Kim, T. and Sheng, Y.P., 2008. Simulation of Hypoxia in Charlotte Harbor, FL. Poster Paper, Ocean Sciences Meeting, Orlando, Florida.
  3. Kim, T. and Sheng, Y.P., 2009. Estimation of Water Quality Model Parameters, Korean J. of Sci. and Tech., In Press.
  4. Ulanowicz, R.E. 2004. Quantitative methods for ecological network analysis. Computational Biology and Chemistry 28:321-339.
  5. Ulanowicz, R.E. and D. Baird.  1999. Nutrient controls on ecosystem dynamics:  The Chesapeake mesohaline community.  J.  Mar. Sys. 19:159-172.
  6. Ulanowicz, R.E., Goerner, S.J., Lietaer, B. and Gomez, R. 2009. Quantifying sustainability: Resilience, efficiency and the return of information theory. Ecological Complexity 6:27-36.