Breakout Session Notes Estuary and Coastal Waters Workshop on Water Quality Modeling for National-Scale Economic Benefit Assessment Washington DC, February 9-10, 2005 ------- DISCLAIMER These proceedings have been prepared by Alpha-Gamma Technologies, Inc. under Contract No. 68-W-01-055 by United States Environmental Protection Agency Office of Water. These proceedings have been funded by the United States Environmental Protection Agency. The contents of this document may not necessarily reflect the views of the Agency and no official endorsement should be inferred. ------- Breakout Session Notes, Estuary and Coastal Waters February 10, 2005 ESTUARY AND COASTAL WATERS BREAKOUT GROUP NOTES PARTICIPANTS Aikman, Frank (National Oceanic and Atmospheric Administration) Ali, Ghulum (US EPA, Office of Water) Allen, Ashley (US EPA, Office of Water) Besedin, Elena (Abt Associates, Inc.) Cerco, Carl (US Army Corps of Engineers) Dettmann, Edward (US EPA, Office of Research and Development) Gross, Tom (National Oceanic and Atmospheric Administration) Hayter, Earl (US EPA, Office of Research and Development) Helm, Erik (US EPA, Office of Water) Leeworthy, Bob (National Oceanic and Atmospheric Administration) MacWilliams, Lauren (US EPA, Office of Water) Parameswaran, Siva (Texas Tech University) Pascual, Pasky (US EPA, Office of Research and Development) Patchen, Richard (National Oceanic and Atmospheric Administration) Powers, John (US EPA, Office of Water) Sheng, Peter (University of Florida) Slawecki, Theodore (Limno Tech, Inc.) Swanson, Craig (Applied Science Associates, Inc.) NOTE-TAKER Peter von Loewe (Tetra Tech, Inc.) i ------- Breakout Session Notes, Estuary and Coastal Waters February 10, 2005 Estuary and Coastal Waters, Session 1 February 9, 1:00-3:30pm Discussion Chairs: Carl Cerco (US Army Corps of Engineers) Breakout Question Scope out and characterize different approaches to modeling water quality in estuaries and coastal waters for national scale benefit assessment (e.g., steady state, hydrodynamic, empirical, probabilistic, etc.). What characteristics make these modeling approaches more or less suitable for evaluating water quality benefits at a national scale? What are the capabilities and limitations, advantages and disadvantages, or tradeoffs among the different water quality modeling approaches? How would one go about developing each approach? How long should it take and how much should it cost to develop each alternative? Rank order or rate the modeling approaches in terms of their suitability for national-scale benefit assessment. Identify four off- the-shelf models that EPA could use (e.g., modified tidal prism model, CE-QUAL-ICM). Rank or rate these models in terms of in terms of their suitability for national-scale benefit assessment Provide a rationale for each ranking or rating. Discussion To structure the discussion, Carl Cerco asked the participants to write down three modeling approaches. Five minutes were allotted for this exercise, during which time the participants were instructed to focus on approaches to modeling or modeling methodologies, rather than selecting particular model codes. Questions were raised (Richard Patchen and Craig Swanson) prior to the start of the exercise as to what the "target" or "endpoint" should be for the modeling exercise. Carl Cerco explained that the modeling approach should address the issue of modeling on a national scale for economic benefit assessment, as described by the discussion topic for this session. After the 5 minutes had expired, Carl Cerco asked each of the participants what they had selected as their 3 modeling approaches. There was some duplication of approaches selected (3D Time- dependent, Box Model, and Data Assimilation), but 12 separate approaches were identified during the exercise, as shown in the following table. ------- Breakout Session Notes, Estuary and Coastal Waters February 10, 2005 Approach Votes Steady State 2D, 3D Hydro and Groundwater I 3D Time Dependent III Box Model/Tidal Prism Nil Ecosystem I Data Assimilation/Statistical/Regression/Neural Networking III Physical Model I Load Reduction Model (50% reduction = 50% improvement) I Sentinal Estuaries I Baysian I Community Based I 4D Integrated Model I Near-field Mixing Model I Carl Cerco suggested adding a "comparative" model approach to the list of approaches, but discussion was refocused and this approach was not added. A dialog began to clarify the definition of some of the approaches (John Powers and Richard Patchen) and whether some of the categories were redundant or inappropriate. There was some discussion regarding some of the approaches as to whether or not these were approaches or requirements. Those deemed as requirements, as opposed to approaches, were separated out into a separate list below. Requirement Good, Cheap, Easy Predictive of Swimability Predict Fish Habitat Considers Commercial Fishing Considers Aquaculture Additives Considers Fish Species Flexible Ability to Identify Uncertainty Based on the inventory of approaches and requirements, the discussion was turned (Theodore Slawecki) to focus on the evaluation of national modeling approaches. As a result, Carl Cerco asked the participants what national-scale pollutants would need to be addressed in such a modeling system. The following list was produced. i ------- Breakout Session Notes, Estuary and Coastal Waters February 10, 2005 Pollutants to be Considered Pathogens Solids Metals Nutrients Drugs Organics Temperature Invasive Species Carl Cerco then asked what the time scale/time step would have to be to consider modeling these pollutants and to obtain results useful in assessing economic benefits. The group agreed that this would be a question for an economist rather than a water quality modeler, and the question was posed to John Powers. John Powers' response was that the time interval would depend on the pollutant and criteria for that pollutant. There was some discussion regarding a daily time step (Earl Hayter) requirement for situations where the majority of pollutant loading occurs during episodic high flow periods (i.e., Hg). The discussion also considered using seasonal time steps for some pollutants (i.e., metals). In a continuation of the discussion regarding requirements of economic modeling, Bob Leeworthy described how NOAA looks at "place-based" economic models. He mentioned that annual time scales work best for these, and had also looked at daily, weekly, and monthly time scales. Richard Patchen then asked John Powers, being an economist, whether a model producing output at 30-second time steps would represent a non-attainment condition if one time step fell below the criteria for a given pollutant. John Powers responded that the economy field is still trying to figure this out, but that in that case, frequency of non-attainment should be considered. John Powers mentioned that it is unclear whether the water body must be "safe" or "attaining" for all periods, or frequently. Richard Patchen suggested that a 6-48 hour minimum time step be used to represent critical episodic events. Carl Cerco introduced another exercise for the group to investigate the preferences of the group. The following chart was drawn on the board and the participants were asked which type of modeling they would prefer with respect to modeling on a national scale. 4 ------- Breakout Session Notes, Estuary and Coastal Waters February 10, 2005 Valuable Less Valuable Easy Hard A discussion ensued regarding which types of models were included in each of the boxes. It was decided that the valuable/easy category would include statistical and box models, and the valuable/hard category would include 2D/3D hydrodynamic and groundwater and 4D integrated models. Easy/less valuable models would be load reduction models. The votes were then tallied: Valuable Less Valuable 4 CO Easy Hard In an effort to define which modeling methods are easy, hard, valuable, and less valuable, the participants voted on the particular methods and where they thought they should be put in this matrix. The following shows the results of the voting. 2 4 1 10 5 3 1 Steady-state 3-D Box/Tidal Prism 1 9 1 12 7 1 3 Statistics Load reduction 4-D integrated 10 Ecosystem 4 1 4 Mixing zone Following the tally, Tom Gross questioned the ability of statistical models to forecast. He noted that to do so, they would use extrapolation techniques that are unbound by physical or empirical data when simulating conditions outside of observations. He mentioned that these are useful when operating within conditions confined by observed values, but extrapolation may give inappropriate results. John Powers asked if physical models could extrapolate through time and space. A number of responses were yes, including from Tom Gross, Carl Cerco, and Earl Hayter. s ------- Breakout Session Notes, Estuary and Coastal Waters February 10, 2005 A discussion comparing physical and statistical models ensued. Statistical models are not confined by physics, and would be less appropriate for long-term simulations where the likelihood of extreme conditions is greater. Carl Cerco asked whether the easy/valuable models should be developed for application on a national scale in the short term, and the hard/valuable models be developed over the long term. The group thought that these types should be developed coincidentally and not staggered. End of Breakout Group Discussion I ------- Breakout Session Notes, Estuary and Coastal Waters February 10, 2005 Estuary and Coastal Waters, Session 2 February 10, 8:30am-10:45am Discussion Chairs: Gross, Tom (National Oceanic and Atmospheric Administration) Breakout Question Achieving a degree of spatial and temporal scale and resolution in water quality models that is consistent with national-scale economic benefit assessment needs has been recognized as an issue in past benefit assessments. o How are different scales and resolutions more or less appropriate for different types of benefits and different approaches to benefit assessment? o Finer resolution in model outputs is generally obtained by increasing the precision, and resolution of model inputs. However, accurate specification of the input variables in large- scale model applications has been a problem in the past because of data limitations and inconsistencies in the resolution of input variable databases. How could the choice of a modeling approach or a model address this problem? o Would EPA be doing itself a disservice by using a model intended to achieve relatively high levels of precision without sufficient knowledge about the inputs of a site being modeled to achieve that level of precision? o Refer to the alternative modeling approaches you characterized in Breakout Session 1. Rank order or rate these modeling approaches in terms of their ability to achieve both the spatial and temporal scale and resolution needed for national scale benefit assessment. Provide a rationale for each ranking or rating. Discussion To structure the discussion, Tom Gross asked the participants to write down three national-scale indices of water quality. Five minutes were allotted for this exercise. The following table was developed based on the responses of the participants. Initially, the indices were listed, but the group felt that some of the indices should be separated into two groups: Services and products of nature. 7 ------- Breakout Session Notes, Estuary and Coastal Waters February 10, 2005 Indices Services(Q s) \Properties of Nature (X,) Diversity X Biomass X Positive Recreational Use X Aquatic Health X Bacterial Contamination X X Turbidity X Sediment Alteration X Woody Debris X Sediment Transport and Potable Water X Fish Safety Consumption X DO X Algal Concentrations/ HAB X Navigable Waters X SAV X Sustainability X Aesthetics X Good Flushing X Nutrient Flux X Fishable X Swimmable X Boatable X Potable X John Powers suggested use of the following formula to determine the utility of a given index: U= U(Qs(Xi)) where U is utility, Qs is the designated use/service, and X is a property of nature. os=n x,wi 1 where W is the index. Discussion ensued (Bob Leeworthy, Tom Gross, and John Powers) regarding willingness to pay (WTP) approaches, and whether economic models exist to calculate the public's WTP. John Powers responded that there are economic models to do so. Tom Gross felt the need to group the indices into coarse service (Qs) groups for further discussion. The groups chosen were fish safety, aquatic health, recreational use, and human health. I ------- Breakout Session Notes, Estuary and Coastal Waters February 10, 2005 Tom Gross asked the group for their thoughts on which properties of nature (Xi) were applicable to maintaining each of the services, and for the time scale or time step each would require for national scale modeling purposes. The table below was developed based on group responses. Properties of Nature (X,) DO Turbidity Mixing Nutrients ;Sediment Transport Algal Blooms ; Bacterial Contamination Temperature/ Salinity ; Ecosystem Webs/Biodiversity ; Biomass Pathogens Coarse Q s Groups Coresponding X, Required Time Scale Fish Safety Bioaccumulation (toxins) Seasonal, regional Bioaccumulation (pathogens) episodic (daily, hourly) DO Patchy Aquatic Health DO Seasonal, regional Turbidity episodic (daily, hourly) Flushing, mixing Patchy Recreational Use Pathogens Seasonal, regional Algal Productivity episodic (daily, hourly) Patchy Human Health Risk Assessment Seasonal, regional pathogens, toxics episodic (daily, hourly) Patchy Richard Patchen asked if higher resolution is worth it with respect to economic benefit analyses. General discussion concluded that for a large range of time scales, it is important, depending on the particular index and property of nature. Shorter time scales would be useful in simulating discrete periods where criteria are being exceeded, while longer time scales would be useful for forecasting. It was concluded that forecasting for risk probability assessment would require full resolution. John Powers explained that high temporal resolution is valuable because a WTP value can be derived for a change in the number of days of beach closures, for example. Theodore Slawecki noted that statistical models cannot predict beach closures for a specific day because they are probabilistic in nature. Tom Gross mentioned that Hindcasting or probabilistic studies for a 10-year period, for example, have value to build a likelihood of closures per year. i ------- Breakout Session Notes, Estuary and Coastal Waters February 10, 2005 Craig Swanson suggested that a database of likelihoods could be developed for beach closures base on existing data. Richard Patchen mentioned that he could never see not needing fine time scale models with predictive capability. Carl Cerco thought that a box-type model would potentially be a good candidate for Hg in sediment, and possibly better than a 3D hydrodynamic model, as data availability is not available for that system. This is commonly a limiting factor for model resolution. Discussion ensued regarding conversion or communication of model results from one type of model to another. Someone noted that the coarsening of a 3D model to a box model is relatively easy, while the conversion of a box model to a 3D model is not possible. It was noted that computer power is cheap and to keep the technology curve in mind. Computer power is likely to get even cheaper. Peter Sheng discussed the concept of Grid computing and the Integrative modeling paradigm where computer resources are exploited to their max to enable researchers to connect and use a full suite of models for most any application. There remains little reason to not do the fully resolved model anymore. Discussion occurred regarding the need for collaboration of research labs and the development of new paradigms for modeling on a large scale. End of Breakout Group Discussion iQ ------- Breakout Session Notes, Estuary and Coastal Waters February 10, 2005 Estuary and Coastal Waters, Session 3 February 10, l:00pm-3:30pm Discussion Chairs: Cerco, Carl (US Army Corps of Engineers) Breakout Question What modeling and benefit assessment issues should EPA be thinking about as it develops a strategic development plan for water quality models that could be used in national scale benefit assessment in estuaries and coastal waters? Discussion To structure the discussion, Carl Cerco asked the participants to write down five properties of the "ideal model." Five minutes were allotted for this exercise. There was a general discussion following this request regarding the system and pollutants being modeled, and the temporal and spatial scale of the model. Carl Cerco responded that the characteristics should describe a model designed to consider economic benefits and water quality at a national scale. Following the five allotted minutes, the following list of characteristics was produced. Interactive: 5 properties of an ideal model Transparent and documented Calculates uncertainty Adaptable Spatially-explicit to about 1km2 Multiple pollutants Scalable to National/Regional Multi-dimensional Coupled to hydro, watershed Pre and post-processor Verifiable Universal Answers management questions Fast Self-calibrating Scalable in Space and Time Flexible in Formulation, Endpoints and Modules Can be run by a non-modeler Useful in Forecasting Doesn't take a lot of Data, and data is available Clearly-defined inputs Carl Cerco then asked the group about what would be needed to develop the "ideal" model. a ------- Breakout Session Notes, Estuary and Coastal Waters February 10, 2005 A large list of needs was produced and summarized into the following categories. ¦ Money ¦ People ¦ Data ¦ Objectives ¦ Buy-in ¦ Time ¦ Framework Carl Cerco asked the group to lump the remaining responses on a group by group basis. The first group to be discussed was the "Framework" group. Richard Patchen asked whether the components of the framework could be hypothetical or if they had to currently exist in order to add them to the list. Carl Cerco responded that he would add components that currently do not exist and make note that they do not. The list of "Framework" components is shown below, with a "?" designating a component that does not exist or exists but is not satisfactory. ¦ Good Theory ¦ Flydro model ¦ Atmospheric Deposition ¦ Eutrophication ¦ Sediment Transport ¦ Ecological Model? ¦ Pollutant Transport and Fate ¦ Watershed ¦ Processors? ¦ Endpoints? ¦ Economic Model ¦ Linkage? Richard Patchen asked John Powers about the types of economic models that are available today, and if they are as rigorous as the water quality models. ------- Breakout Session Notes, Estuary and Coastal Waters February 10, 2005 John Powers responded that economic models exist, but are not process based and as rigorous as the water quality models, but he thought that existing models could be adapted. ¦ Observations (Temporal, Spatial) ¦ Forcing Functions ¦ Socioeconomic Data ¦ Description of Domain ¦ Health Criteria Carl Cerco then asked the group to list the types of data required for the "ideal" model. The responses are listed below. Carl Cerco then asked the group to list the types of people or staff required for the "ideal" model. The responses are listed below. ¦ Programmers ¦ Information Technology ¦ Biologists ¦ Hydrologists ¦ Multidisciplinary Scientists ¦ Economists ¦ Managers ¦ Leaders Carl Cerco asked the group if these personnel are in supply in the workforce. The group thought that all were available, but some more than others, and that some require a higher pay scale than others. Carl Cerco began to ask the group about the pay scales for each position, but dismissed that exercise. Carl Cerco noted that programmers are a necessity in most advanced modeling projects, but are a "dime a dozen." Richard Patchen noted that although programmers are an important component, a programmer with an environmental or economic background would be preferable, since based on his ------- Breakout Session Notes, Estuary and Coastal Waters February 10, 2005 experience, many programmers are not familiar with water quality processes and don't have the ability to QA/QC model code. The group agreed that multidisciplinary scientists familiar with hydrology and ecology and who also program would be key personnel. Theodore Slawecki suggested that a leader would have to be driven and interactive in the process, rather than a macro-manager. He thought someone with Bill Gates' abilities would be appropriate for the role as leader. ¦ Political Savvy ¦ EPA, NOAA, USACE, etc.. ¦ Scientific Community ¦ Regulated Population ¦ International Carl Cerco asked the group what would be needed for buy-in. The group acknowledged that each of the buy-in components would be required except for the international component. This component would be more of a supplement for instilling buy-in for the remaining stakeholders. Carl Cerco asked the group what would be needed for time-accelerators. The following list was developed. ¦ <10 Years ¦ Immediate Prototype for Buy-in (2 Years) ¦ Training, Communication ¦ Identify Parallel Tasks ¦ Economic, Engineer Linkage ¦ Deadline ¦ What's Different ¦ Environmental Endpoint ¦ Public Well-being ¦ Multiple User ¦ Economy of Scale Tom Gross thought that a prototype for buy-in would be one of the more important components on the list. i4 ------- Breakout Session Notes, Estuary and Coastal Waters February 10, 2005 The group, in general, agreed that the economic-engineer linkage is a vital component that would need to be established early on to promote progress on the system. Carl Cerco asked the group what would be needed in terms of money to develop the "ideal" modeling system. The following list was developed. ¦ Prototype ¦ First Shot ¦ Great Job $1,000,000 $500,000 $10,000,000 End of Breakout Group Discussion a ------- |