SETTING SALINITY TARGETS FOR THE SOUTHERN COASTAL SYSTEM

Transcription

1 SETTING SALINITY TARGETS FOR THE SOUTHERN COASTAL SYSTEM An Informational Paper to Assist in the Decision-making Process Developed by SCS Regional Coordinators for the February 1 and 2, 2011 meeting FIU Biscayne Bay Campus 1. Introduction One of the unique and important products that emerged from the 2009 Systems Status Report (SSR) was the evaluation of CERP s Southern Coastal System s (SCS) salinity performance measure (PM) using both empirical monitoring data to assess the current condition, and model output to compare the predicted restored condition to the PM target. This work successfully linked for the first time the assessment and evaluation components using a single RECOVER performance measure, and it was regarded as ground-breaking work by RECOVER managers. The use of one set of performance measures that can handle both the evaluation and assessment components yields output in a common physical or ecological currency which is helpful to managers for planning restoration and adaptively managing the system. While the work described above represented an important linkage of evaluation and assessment components, it also revealed flaws in the SCS salinity performance measure including: Metrics are too complex and confusing; Targets are sometimes conflicting; and Target properties are redundant. These findings strongly suggest the need for revising the PM, which made its revision a high priority task in the RECOVER Federal Fiscal Year 2011 (FY11) Work Plan. Additionally, Principal Investigators who conduct ecological monitoring funded by the RECOVER Monitoring and Assessment Plan (MAP) in the SCS have repeatedly expressed the need for restoration salinity targets in order to appropriately revise the ecological PMs. Revision to the ecological PMs is anticipated to allow consistent evaluation of the ecological PMs with the salinity PM. Accordingly, this makes possible the revision of the performance measures for submerged aquatic vegetation, fish, and pink shrimp, which are also priority tasks in the RECOVER FY11 work plan. To accomplish the above tasks, the SCS Regional Coordinators (RCs) are convening a meeting in February 2011 to determine the tools that will be used to set salinity targets for the SCS. Setting of these targets is imperative for revising the salinity PM and the SCS ecological PMs, as noted above. The purpose of this paper is to describe the viable options for predictive salinity tools and targets (i.e., availability, time to development, and cost for development), as well as to provide the RCs recommendations. The primary objective for setting these targets and revising the PMs is to enable the developers of the next SSR to include evaluations of all PMs in terms of MAP monitoring data and predictive scenarios (especially the restored condition) to the extent possible, thus making the SSR a more useful and effective product for understanding and management of the system. 1

2 It is preferable to have the salinity targets in the SCS sub-regions defined by a specific tool (i.e., model) rather than a target based on best professional judgment (BPJ), e.g. the current RECOVER Florida Bay PM target is a BPJ target based on the Florida Bay Florida Keys Feasibility Study (FBFKFS). The reasons for this preference are the following. Targets that are defined by models provide temporal resolution not possible from BPJ estimates. Model targets can provide hourly or daily time steps; whereas, BPJ estimates are generally limited to annual or seasonal resolution. Modeled targets provide a better spatial aspect. The goal is to hindcast ecological conditions based on modeled hydrologic conditions, and BPJ estimates do not lend themselves to this type of detailed spatial and temporal analysis. If BPJ estimates must be used, they should reflect the latest and best available science, especially the paleoecological data now available. The current RECOVER Florida Bay PM specifies a BPJ estimate based on outdated information and requires reformulation using FATHOM basins, which are currently recognized as a better way to sub-divide Florida Bay than the FBFKFS zonation (Nuttle et al. 2000, Boyer et al. 2007, Madden et al. 2009). 2. Temporary (i.e. interim) Target Because there may be controversies surrounding the establishment of permanent salinity targets, the SCS RCs believe that a temporary or interim target may be more palatable to some agencies and facilitate reaching consensus during the decision-making process. We have suggested an expiration date for the temporary targets as the publication date of the 2012 (September 2012). At that time, the targets and the tools or methods which produced the targets, will be re-evaluated to determine if they are still appropriate or require modification. In the interest of avoiding confusion with CERP s Interim Goals and Targets (IGITs), these targets will be referred to as temporary targets in this paper. 3. Greater Everglades Regional Model The sub-regional models that the SCS team chooses to use for Biscayne Bay, Florida Bay, and the Southwest Coast (see attached SCS regions) may be driven by a Greater Everglades regional model (see attached model flow diagram). At this time there exists several pre-drainage Greater Everglades regional model scenarios, but there is no definitive guidance directing the SCS team on which of those models should be applied to the regional evaluations. The model options are currently being reviewed and evaluated by a RECOVER sub-team to determine the preferred model and pre-drainage target to be used for CERP. The SCS team will defer to this sub-team s recommendations, if and when the decision is made. If it is apparent no definitive guidance will be forthcoming, the SCS team will perform their own evaluation and recommendations for the SCS only. The RC s realize the recommendations on the Greater Everglades regional model may not occur in time to produce the 2012 SSR. If no decision is forthcoming by the RECOVER sub-team and insufficient time remains for the SCS to perform their own evaluation prior to the drafting of the 2012 SSR, the SCS will use the NSM as indicated in the current 2

3 Interim Goals and Interim Targets (IG/IT) document: Indicator Salinity Patterns in Florida and Biscayne Bays, How is the interim goal for this indicator predicted?. 4. Model Evaluation Marshall and Nuttle (2011) used a set of evaluation parameters to evaluate predictive models available for south Florida. The evaluation methodology used herein is based on a collaborative effort by south Florida modelers coordinated through the Florida Bay Science Program (PMC, 2000). The following set of evaluation factors for models (see Marshall and Nuttle 2011 for definitions of terms) were used in this study: Portability, Validity, Fidelity, Focus, Ease of use, Temporal Coverage, and Spatial Coverage. These evaluation factors should be considered by this sub-team in determining the best tools for each sub-region. Each of the models in the southern Biscayne Bay and Florida Bay sub-region were assigned a score for achieving the desired result of each evaluation factor. The scale of scoring is as follows: 1 = poor 2 = fair 3 = average 4 = very good 5 = excellent. The model scores were then compared in several ways in order to understand the strengths and weaknesses of each model. The available ratings for some the models are presented in Tables 8.3 and 10.3 in the applicable subregional sections below. Goodness of fit statistics (RMSE = root mean squared error, MAE = mean absolute error, R 2 = coefficient of determination, NSE = Nash-Sutcliffe model efficiency) should also be taken into account where applicable and available. These statistics indicate how well the models performed in the calibration/validation process and provide insight to the level of uncertainty of the models. Marshall and Nuttle (2011) provides detailed definitions and explanations for each the five statistics. The available statistics for some the models are presented in Tables 8.2 and 10.2 in the applicable subregional sections below. Essentially, the larger the value of the error statistic indicates the poorer the goodness of fit and the greater the uncertainty in the model. 3

4 5. Wetland Transition Area SFWMM v TIME After the Restudy in 1999, the Florida Bay Program Management Committee (PMC) described limitations in the South Florida Water Management Model (SFWMM = 2X2) used to evaluate CERP project alternatives, particularly with regard to how alternatives affect the wetland transition area. The PMC recommended pursuing an alternate model being developed at the time by the U.S. Geological Survey (USGS). The present result of this recommendation is the USGS Flow and Transport in a Linked Overland/Aquifer Density Dependent System (FTLOADDS) model. The FTLOADDS model is broken into two model domains: Biscayne and Southern Everglades Coastal Transport (BISECT), which incorporates the Tides and Inflows in Mangrove Ecosystems (TIME) domain, and the Ten-Thousand Islands (TTI) domain (for model domains see Marshall and Nuttle 2011). BISECT and TTI offer the potential for better tools in areas where the SFWMM produces less reliable data. However, an Everglades National Park (ENP) review of the TIME model domain resulted in unresolved issues which will be addressed in a pending contract between the USACE and USGS and require approximately one year to complete. The resulting updated TIME model will require a RECOVER review to determine its viability as a useful tool for the SCS. At this time, because of the above issues, the TIME model will not be available for the developers of the SCS PMs and the 2012 SSR. 6. Ten Thousand Islands (TTI)/Picayune Strand The following options are available for use in TTI/Picayune Strand for salinity models and targets: 1. FTLOADDS/TTI- Multi-Linear Regression (MLR) 2. Best Professional Judgment (BPJ) This area was not conceived within the original area of the Southern Coastal System; however, it is the location of a CERP project, the Picayune Strand Restoration Project. In order to adaptively manage the project, there is a need for the ability to model the system expectations in that area. The FTLOADDS/TTI model was developed specifically for looking at project alternatives during the project planning phase. The FTLOADDS/TTI model uses boundary conditions from the SFWMM. It is the recommendation of the RCs that the FTLOADDS/TTI model be finalized by the USGS and delivered to the Interagency Modeling Center (IMC) for review, acceptance, and implementation. That model would be used for inputs to MLR models that are currently being developed to simulate salinity at specific salinity locations in the estuary around the mouth of Faka Union Canal. Updates to the MLR salinity models are scheduled to be completed December

5 Table 6.1. Summary table showing the target options for the Ten Thousand Island/Picayune Strand area. Included in the table are significant practical pros and cons for using these models, as well as the estimated funding and time required to get the models ready for use. The RC choice is also shown. Model Advantages Disadvantages Presently available FTLOADDS /TTI-MLR BPJ Based on desired biological conditions. Based on salinity requirements of species of interest. 7. Southwest Coast The FTLODDS/TTI model has not been reviewed and accepted by the IMC. Must rely on USGS for FTLOADDS/TTI model runs. Need to have USGS commit to delivering to the IMC with all documentation. Less spatially specific, covers broad regions, no time series. Greatly reduced temporal and spatial resolution. Biological conditions are assumed. BPJ conditions have not been developed for this area. The following salinity models and targets are options for use in Southwest Coast: 1. NSM- MLR 2. NSM- MLR Paleo adjusted 3. Paleo 4. BPJ Funding/Time Constraints none Undetermined amount of time and funding; would most likely require more time that is allowed for this exercise. RCs Choice Predictive salinity models in the Southwest Coast are limited. MLR salinity models currently exist for areas of the Southwest Coast that rely upon four stage gages in Big Cypress Preserve. The MLR models are for salinity gage station locations at Lostman s River (LO) and Willy Willy (WW) on Lostmans River and Cannon Bay (CA) and Watson Place (WP) on Chatham River. The SFWMM does a poor job of simulating stage in the Big Cypress Preserve portion of the domain. Until this deficiency is corrected, these four models are not useful for either evaluation or assessment purposes. The RCs have submitted a request to IMC to start producing and review the quality of the stage values for this area. 5

6 Table 7.1. Summary table showing salinity models and target options for the Southwest Coast. Included in the table are significant practical pros and cons for using these models, as well as the estimated funding and time required to get the models ready for use. The RC choice is also shown. Target Advantages Disadvantages Funding/Time Constraints RCs Choice NSM-MLR MLR models are available SFWMM does a poor job of none NSM- MLR-Paleo adjusted Paleo Best Professional Judgment (BPJ) 8. Florida Bay Inform decision on what target is. Models are available Biologically based on actual data Data is available Based on desired biological conditions. Based on salinity requirements of species of interest. Some targets avail for the Whitewater Bay area. simulating stage in this area. SFWMM does a poor job of simulating stage in this area. Compounds assumptions of paleo data. No time series. Lose temporal variability. Reduced temporal and spatial coverage. Fewer site information available. Less spatially specific, covers broad regions, no time series. Greatly reduced temporal and spatial resolution. Biological conditions are assumed. Not available for most of the region none none Undetermined amount of time and funding. The following salinity models and targets are options for use in Florida Bay: 5. NSM-MLR (driven by SFWMM and NSM, RSM and NSRSM, or TIME) 6. NSM-MLR-Paleo adjusted 7. NSM-Flux Accounting and Tidal Hydrology at the Ocean Margin (FATHOM) 8. BPJ The current salinity PM for Florida Bay includes several analyses based upon the Florida Bay/Florida Keys Feasibility Study recommendations using best professional judgment salinity ranges and zones. When this PM was used in the 2009 SSR, possible flaws were discovered in the salinity ranges recommended for some zones. As a result, the proposal here in is to use targets developed from model output so that the targets are consistent across the landscape. It also is proposed to use an alternative zonation system developed for the FATHOM model that delineate the basins within Florida Bay. These basins are further grouped by interpretations by others (Nuttle et al, 2000; Boyer et al, 2007; Madden et al., 2009) into six zones based upon water quality/salinity characteristics. Currently, the SFWMM model is used to model stage within the model domain and MLR models are available to model salinity at the Marine Monitoring Network (MMN) stations using stage, wind, and, sea level. IMC currently can post-process values from the SFWMM and the NSM to produce daily salinity values at the MMN stations using the MLR models. An additional salinity target for Florida Bay MMN stations based upon MLR output adjusted using paleo-salinity information is being developed (Marshall and Wingard collaboration) and should be available in March

7 The FATHOM model is currently being reviewed by IMC and has the potential to provide monthly salinity values averaged over the individual model basins (based upon Florida Bay basins). The FATHOM model uses freshwater flow inputs from specific cells of the SFWMM. All of the box models being reviewed have the disadvantage of only producing monthly salinity output. Ecologists working in the region that are monitoring and modeling various ecological parameters indicate that monthly temporal resolution is not adequate and, at a minimum, daily values are needed. During the Florida Bay/Florida Keys Feasibility Study, attempts were made to develop the Environmental Fluid Dynamics Code (EFDC) hydrodynamic model with a three-dimensional grid in Florida Bay. That model may be available for use in Florida Bay; however, an evaluation of the applicability and affordability of further developing this model is warranted before any final decision is made by the SCS for application of this model for target setting purposes in Florida Bay. Table 8.1. Summary table showing the model and target options for Florida Bay. Included in the table are significant practical pros and cons for using these models, as well as the estimated funding and time required to get the models ready for use. The RC choice is also shown. Target Advantages Disadvantages NSM-MLR Available now, daily output. Site- specific output unless Ability to adapt to FATHOM aggregated to FATHOM basins basins. NSM-MLR- Paleo adjusted NSM- FATHOM BPJ targets Would provide a more scientifically-defensible target than unadjusted MLRs daily output. Ability to adapt to FATHOM basins. Available now. Has been reviewed by IMC with minor comments, no show stoppers. good spatial resolution; can provide NSM target. Can be used to confirm the paleo- MLR. Will be housed at the IMC ~ Based on desired biological conditions. Based on salinity requirements of species of interest. Greater uncertainty than MLRs, paleo only avail for 6 MMN sites. Perpetuates uncertainties of paleo data. Model adds uncertainty. Monthly output at the FATHOM sub-basin level based on flow and stage boundary conditions from the SFWMM. Less spatially specific, covers broad regions, no time series. Greatly reduced temporal and spatial resolution. Biological conditions are assumed. Currently avail for FKFBFS Zones: need to be updated using paleo data and converted to FATHOM zones. Funding/Time Constraints none Available in March 2011, no new funding required None. Undetermined amount of time and funding; Would most likely require more time that is allowed for this exercise. RCs Choice Checked against FATHOM 7

8 Table 8.2. Model uncertainty statistics for Florida Bay models, where available (RMSE = root mean squared error, MAE = mean absolute error, R 2 = coefficient of determination, NSE = Nash-Sutcliffe model efficiency). The larger the value of the error statistic indicates the poorer the goodness of fit and the greater the uncertainty in the model. Model Name Mean Error RMSE MAE R 2 NSE MLR FATHOM TIME salinity creek mouths *The values provided in the tables were obtained from various model calibration/validation reports. As such, caution should be taken when making model to model comparisons because each model has a unique geographic domain with unique hydrologic characteristics, and each model uses a unique calibration/validation period of record. The bottom line is that is that these statistics should be considered as examples of goodness of fit, and some examples are best case values. However, they are provided here because we believe that some information is better than no information regarding the models performance. Table 8.3 shows how each Florida Bay model scored using the scheme described in Section 4, Model Evaluation. The models were scored by three coastal hydrologists with a wide range of experience using or interpreting most of the south Florida models evaluated herein. The widest ranges of scores were for portability, fidelity, and temporal coverage, reflecting the constraints of model sophistication. The narrowest range of scores was associated with validity - these models have withstood the test of use and have been developed, reviewed, and updated to a point of application-ready acceptance. The most useful information comes from looking at the strengths and weaknesses of the models as expressed in the scores. For this evaluation a model is defined as having a strength if it scores excellent (5) and a weakness if it scores poor or fair (1 or 2). Model Table 8.3. Summary of scoring for Florida Bay predictive salinity models using the model evaluation factors described by Marshall and Nuttle (2010). Score is from 1 = lowest to 5 = highest. Parameters simulated Portability Validity Fidelity Ease of Use Temporal Coverage Spatial Coverage Sum MLR salinity Four Box Florida Bay FATHOM Florida Bay EFDC Florida Bay salinity salinity salinity

9 9. Card Sound, Barnes Sound, and Manatee Bay The following salinity models and targets are options for use in this area: 1. NSM-Biscayne Bay (BB) Box Model 2. NSM-BB Box FATHOM interpolation 3. NSM-MLR 4. NSM-FATHOM (Barnes & Manatee only) 5. BPJ These basins have limited connection to both Florida Bay and Biscayne Bay; however, it is important to understand the effects of restoration on these basins because of the association with the C-111 Spreader Canal project. The Biscayne Bay Box Model extends southward into these basins and, because of the tenuous connection of these basins to Biscayne Bay, the basins can be modeled independently from the larger Biscayne Bay portion. As indicated above, this box model currently has monthly salinity in limited boxes within the basins, but it appears to perform well in this region. The MLR salinity models for Manatee Bay and Barnes Sound (MMN station MN and MD, respectively) currently use EVER1 as input. It has been determined that EVER1, which is adjacent to areas of interest, does not perform well in the SFWMM due to hydrologic isolation and other factors. The MLRs need to be reformulated based on other stage gages to provide a better relationship. This is demonstrated in Table 9.1. The NSM model estimates are lower than existing data. Table 9.1. Quartiles of estimated, modeled, and actual stage data at EVER1 (see notes for periods of record). Quartiles SDC NSM Data Q Q Q Q Q SDC = Estimated stage duration curve from the current graphical output for the Performance Measure for Manatee Bay and Barnes Sound NSM = Natural System Model bias adjusted results ( ) for EVER1 Data = available stage data ( ) for EVER1 The stage duration curve estimated in the table above is actually a representation of the stages at locations immediately west of U.S. Highway 1 from EVER1, represented by EVER2 (A&B) and EVER 8. A potential target in this area could be to have stage at EVER1 be equivalent to stage at these stage gage locations west of U.S. Highway 1. Table 9.2 provides the quartiles from the available data at these locations. The approach of using this stage gage location comparison as a target provides a viable performance measure proxy until adequate models become available. 9

10 Table 9.2. Quartiles of actual stage data at EVER1, EVER 2(A&B), and EVER8. Quartiles EVER1 EVER2A EVER2B EVER8 Q Q Q Q Q The FATHOM model also extends into Barnes Sound and Manatee Bay. Refer to discussion on the FATHOM model in the Florida Bay section above. Table 9.3. Summary table showing the salinity model and target options for Card Sound, Barnes Sound, and Manatee Bay. Included in the table are significant practical pros and cons for using these models, as well as the estimated funding and time required to get the models ready for use. The RC s have no recommendation for this area. Target Advantages Disadvantages Can extract this portion out of the entire model and run independently. Can be connected to Florida Bay and Biscayne Bay proper for transfer of salinity values in this area. NSM-BB Box model NSM-BB Box FATHOM interpolation NSM-MLR NSM- FATHOM NSM-ADH BPJ Regression estimate between FB and BB model boundaries Can be updated for use in Available now. Will be housed at the IMC ~ High spatial resolution; accurate; 2-D converted to available platform. Based on desired biological conditions. Based on salinity requirements of species of interest. Limited temporal resolution (monthly output ) in limited boxes (grids) two in nearshore Hasn t been done. Not available now. Presently forced to use EVER1. Funding required to update code to utilize a better suited stage gage. Currently a good stage comparison PM between EVER1 and EVER2 (A and B) and EVER8. Only available for Barnes Sound and Manatee Bay (no Card Sound). Monthly temporal resolution. Requires re-verification against the original version Less spatially specific, covers broad regions, no time series. Greatly reduced temporal and spatial resolution. Biological conditions are assumed. Currently avail for FKFBFS Zones: need to be updated using paleo data and converted to FATHOM zones. 10 Funding/Time Constraints 3 months and $30K to convert to daily output; $30K. $15K to upgrade to a gage from EVER1. None. Included in Biscayne Bay costing (4 months /$65K) Undetermined amount of time and funding; would most likely require more time that is allowed for this exercise. RCs Choice Checked against FATHOM

11 10. South Biscayne Bay The following salinity models and targets are available for use in Biscayne Bay: 6. NSM-Biscayne Bay Box Model 7. NSM-MLR 8. NSM-Biscayne Bay Simulation Model (BBSM) 9. NSM-ADH 10. NSM-ADH Oyster adjusted 11. TABS-MDS Paleo adjusted 12. National Park Service Western Bay Zone (WBZ) 13. BBSM-WBZ 14. Current salinity PM targets A number of flow/salinity targets have been independently developed for the south Biscayne Bay area. Many of these targets are described in a technical report created by the National Park Service [South Florida Natural Resources Center (SFNRC), 2008]. This suite of flow targets were developed using a variety of methods and target areas. Included in the suite is a target developed and recommended for use by the NPS. It proposes a much larger target area (10,000 acres) than the currently approved RECOVER target area (3,200 acres), and stretches from Shoal Point to Turkey Point. NPS refers to this area as the Western Bay Zone (WBZ) (SFNRC, 2006). To achieve desired or target salinities within the 10,000 acre target, NPS estimates that a total annual volume of 960,000 ac-ft would be required, with 37,000 ac-ft/month delivered in the dry season and 110,000 ac-ft/month deliver in the wet season. This was considered supportive of lesser estimates of flow requirements in the smaller (3,200 acre) area (22,000 ac-ft/month/66,000 ac-ft/month in the dry/wet season) to meet similar salinity requirements within that smaller area. The Biscayne Bay Box Model has been developed for all of Biscayne Bay (see Marshall and Nuttle 2011 for model description and area). Some development will be needed to use specific model input as boundary conditions. As indicated above, there are limitations in the current output. The model has poor temporal and spatial resolution for CERP purposes (i.e. monthly time steps and in a limited number of boxes). The addition of 2 inshore boxes would greatly improve spatial resolution and the modification to daily output would improve temporal resolution. This would make the model a more viable tool for CERP purposes. Several individuals have attempted to develop relationships between upland gages and salinity in the south Biscayne Bay area. There are problems caused when gages are located near existing canals. The development of MLR models can be pursued for a moderate cost in time and money (see table below). The Biscayne Bay Simulation Model (BBSM, a.k.a. Wang CAFÉ-2D Biscayne Bay Hydrodynamic model) has been developed for south Biscayne Bay. The model has been used to develop scenarios associated with the Biscayne Bay Coastal Wetlands project. Further modification would be required to couple the BBSM to NSM (or NSRSM). The BBSM could 11

12 also be used to generate spatially-distributed time-series (at model nodes) based on the aforementioned NPS WBZ target, as well as other proposed restoration scenarios. TABS-MDS, a multi-dimensional hydrodynamic numerical model, was developed by the USACE for the Biscayne Bay Coastal Wetlands project. The updated version of TABS-MDS is known as the Adjusted Hydraulics (ADH) model. The ADH model is compatible to various present-day operating systems and can be run from any desktop. There is potential for the use of the model within the south Biscayne Bay area. The TABS-MDS could also be used to generate spatially-distributed time-series (at model notes) to evaluate various restoration scenarios. This option can be pursued for a moderate cost in time and money (see table below). Assuming that a functional model for Biscayne Bay is available, a slightly different approach to define a Biscayne Bay restoration salinity regime is being proposed. The exact physical location and extent of the historic oyster beds that were once present are known. From the literature, and experience in the NE module, the salinity regime that must have existed historically where these oysters once were is also known. The proposal is to model flow Bay entering the Bay thru the 22 relic creeks such that the salinity regime over the timeframe of the model (e.g., for the Wang that would be ) at the mouths of those 22 relic creeks was such as to promote oyster survival and growth. Since increasing head such that freshwater springs might reproduce the historic boils in mid-bay is implausible given current realities, groundwater flow is set at 10% of current condition. This defines a defensible, biologically-based flow regime (average ac-ft/year) that in turns defines a Bay-wide salinity regime, and provides a spatially-dense (at model nodes) time-series for ecological exploration. 12

13 Table Summary table showing target options for Biscayne Bay. Included in the table are significant practical pros and cons for using these models, as well as the estimated funding and time required ot get the models ready for use. The RC choice is also shown. Target Advantages Disadvantages Can use SFWMM as In current configuration input (i.e., provides NSM Limited temporal resolution or ENPmod2 predrainage (monthly output ); spatial targets and can resolution very low. evaluate CERP SFWMM input known to be alternatives). Can be questionable east of L-31N. ready to use by NSM-BB Box model NSM-MLR Uses SFWMM input. Allows simulation of salinity in areas along the mainland shoreline. Better temporal resolution than the current Box model. Can be ready to use by Limited spatial extent at daily resolution (only available for a few sites nearshore South Biscayne Bay); coverage increased at monthly resolution; no documentation currently available; Sitespecific output. NSM-BBSM Good spatial resolution. Not available now. Currently provides AltO/Reuse restoration scenario as target. Adapting to use regional model boundary conditions will require unknown amount of work. NSM-ADH NSM-ADH verified to historic oyster regime TABS-MDS paleo adjusted NPS-WBZ Target BBSM -NPS- WBZ Target Current salinity PM target High spatial resolution; accurate; 2-D converted to available platform. High spatial resolution; accurate; 2-D converted to available platform. Oyster information available. Information available in an existing report Available now. Good spatial and temporal resolution. Currently available. Based on desired biological conditions. Based on salinity requirements of species of interest. Based on Meeder target. Requires re-verification against the original version Requires re-verification against the original version Incorporates a larger area than we are focusing on (C-100 to C-111). Uses older NSM run. Based on BPJ. Will not allow hindcast ecological relationships. Does not estimate a time series. Not available now. Based on BPJ. Independent to regional models. Less spatially specific, covers broad regions, no time series. Greatly reduced temporal and spatial resolution. Biological conditions are assumed. BPJ. 13 Funding/Time Constraints 2 months and $60K to convert to daily output and add 2 inshore boxes $25K to finish models and optimize existing models. considerable effort and time to produce 4 months /$65K 4 months /$65K none None Would require NPS support to produce none RCs Choice

14 Table Model uncertainty statistics available for Biscayne Bay models, where available (RMSE = root mean squared error, MAE = mean absolute error, NSE = Nash-Sutcliffe model efficiency). The larger the value of the error statistic indicates the poorer the goodness of fit and the greater the uncertainty in the model. Model Name Mean Error RMSE MAE R 2 NSE Biscayne Bay Box Model Biscayne Bay Simulation Model < 4 TABS MDS Biscayne Bay Table Summary of scoring for Biscayne Bay salinity models using the model evaluation factors described by Marshall and Nuttle (2010). Score is from 1 = lowest to 5 = highest. Parameters Ease of Temporal Spatial Sum Model simulated Portability Validity Fidelity Use Coverage Coverage Biscayne Bay Box Model salinity Biscayne Bay Simulation salinity Model TABS MDS Biscayne Bay salinity North and Central Biscayne Bay There are no CERP Projects specifically designed to provide measurable benefits in north and central Biscayne Bay. A salinity target setting exercise similar to this one will be necessary to determine the best path forward for RECOVER PM refinement in central and northern Biscayne Bay. A new sub-team will be formed shortly following the publication of this sub-team s recommendations for South Biscayne Bay, Florida Bay, TTI, and the Southwest Coast. Literature Cited: Boyer, J.N. et al Nutrient dynamics. In: Hunt, J. and Nuttle. W. (Eds.), Florida Bay Science Program: A Synthesis of Research on Florida Bay. Fish and Wildlife Research Institute Technical Report TR-11.iv. pp Madden, C.J., D.T. Rudnick, A.A. McDonald, K.M. Cunniff, and J.W. Fourqurean Ecological indicators for assessing and communicating seagrass status and trends in Florida Bay. Ecological Indicators 9S. Marshall, F.E., III and W.K. Nuttle South Florida Hydrology and Salinity Models. A document for RECOVER Greater Everglades and Southern Coastal Systems Performance Measure Targets: Hydrology and Salinity Simulations Using Estuarine and Freshwater Paleoecological Characterizations Coupled with Ecological Models for Comprehensive Everglades Restoration Plan (CERP) Restoration Scenario Evaluations. USACE Contract No. W912EP-09-R Available as a read-ahead. 14

15 Nuttle, W.K., J.W. Fourqurean, B.J. Cosby, J.C. Zieman and M.B. Robblee The influence of net freshwater supply on salinity in Florida Bay. Water Resources Research 36: PMC Florida Bay Models Coordination Meeting. Report of a workshop held in May SFNRC Technical Report (2). Estimates of Flows to Meet Salinity Targets for Western Biscayne National Park. South Florida Natural Resources Center, Everglades National Park, Homestead, FL. SRNRC Technical Series 2008:2. 28 p. SFNRC Technical Report (1). Ecological and Hydrologic Targets for Western Biscayne National Park. South Florida Natural Resources Center, Everglades National Park, Homestead, FL. SRNRC Technical Series 2006:1. 25 p. 15

16 Appendix A: reserved for minority reports 16

17 Picayune Strand Southern Coastal Systems Regions South Biscayne Bay Southwest coast Manatee Bay and Barnes Sound Florida Bay

18 General Flow of Models to Performance Measures Regional Model Stage Flow Sub regional model Salinity Ecological Performance Measures Biscayne Bay Mangrove fish Pinkshrimpand associated epifauna Seagrass Florida Bay Juvenile Seatrout Pinkshrimpand associated epifauna Seagrass