6

Science to Support Decision Making

Science has always been at the core of the Comprehensive Everglades Restoration Plan (CERP). By embracing adaptive management as a key tenet of restoration implementation, the Yellow Book (USACE and SFWMD, 1999) and subsequently the Programmatic Regulations (33 CFR §385.31) recognized not only the importance of a solid scientific underpinning of the plan but a need for continual scientific and engineering information to support ongoing restoration decision making. The Committee on Independent Scientific Review of Everglades Restoration Progress has recognized advances in science, including monitoring, modeling, and synthesis, in previous reports (e.g., NASEM, 2016; NRC, 2007, 2014). In recent years (NASEM, 2016, 2018; NRC, 2014), the committee’s recommendations have focused on science to support long-term planning and setting forth realistic expectations of the future condition of the system. Yet, this is just one way that science can and should inform decision making. As the CERP and related programs come to fruition and critical pieces of the long-envisioned restoration infrastructure begin operations, new opportunities emerge for the application of existing knowledge and the development of a deeper understanding of system function and ecosystem response to water management.

Two non-CERP efforts—the Combined Operational Plan (COP; see Chapter 4) and the Lake Okeechobee System Operating Manual (see Chapter 3)—represent current large-scale examples of the potential utility of applying scientific information to inform operations. Scientific analysis in support of the COP used hydrologic and ecological models to evaluate alternatives and, to the extent possible, manage trade-offs across the system. As more information becomes available and models are improved, the operational plans can be refined through adaptive management to better meet ecosystem and other objectives. Because of the scale of the effect of the COP and the Lake Okeechobee System Operating Manual, systems-scale thinking and analysis has been essential (Box 6-1).

Presently these appear to be driven more by progress on project planning and authorization than their contributions to overall system performance. Systems analysis could be applied in the development of the next IDS by evaluating alternative project implementation schedules over a number of scenarios of key external factors (e.g., state and federal funding levels), using a set of performance metrics representative of the multiple CERP objectives. Doing so would ensure that the schedule of projects is purposeful in terms of meeting future objectives by ensuring those projects that make a difference at the system scale are prioritized.

2. Assessment of restoration outcomes and adaptive management. A project-level adaptive management process (RECOVER, 2011b) has been crafted to support decisions on when or how CERP projects or project components need to be adjusted based on monitoring data to optimize their benefits. The outcomes of some CERP projects may be limited by factors outside the control of the CERP (see Chapter 5), making within-project adjustments and adaptive management less successful than expected. Understanding project-level and systemwide responses and their interacting causal factors through data analysis and modeling is essential to support timely decisions that can improve overall restoration outcomes. If the effect of actions on the system cannot be accounted for, adaptive management is impossible.
3. Near-term operational decision making. Although major operational changes, such as the COP (see Chapter 4), involve a lengthy planning process with structured evaluations of systemwide effects, CERP agencies face many other near-term operational decisions, such as those resulting from adaptive management or those involving extreme events. As more projects come online, operational decisions will be more complex, and these decisions, including operational changes, will benefit from tools and strategies that can efficiently and actively bring science and systems-level understanding to bear on near-term decisions.
4. Science planning and investment. As the system changes due to CERP implementation and other factors, CERP agencies will need to identify and prioritize the science needs to support future management of CERP infrastructure. Until now there has been little opportunity to consider how science in support of the CERP should evolve in the future. Ongoing research on issues such as peat collapse and the effects of sea-level rise (see NASEM, 2016, 2018) demonstrate that, as the system changes, science in support of the CERP should itself adapt. Decisions on the highest priority investments would be best informed by a systems-level perspective considering which information can add value to restoration progress.

Informing these decisions requires focused and deliberate application of science, and the Everglades restoration community has built a solid foundation of

monitoring data and models. The long-awaited evolution of the program toward implementation and operation of multiple projects, however, requires a different kind of application of this toolkit of monitoring, modeling, and synthesis. In this chapter the committee explores how new strategies can improve the development of science to inform decisions and stimulate the systems perspective long envisioned. It will also examine key elements that support the use of science, once that science has been developed.

DEVELOPING THE SCIENCE AND SYNTHESIS TO INFORM DECISIONS

Scientific information and its application to restoration can take many forms. The complexity of a system like the Everglades requires detailed measurements and understanding of small-scale processes that are meshed with landscape-scale dynamics and regional factors such as meteorological and climate variability. Findings need to be captured in ways that support decision makers and foster learning about the restoration process by all involved. In this section three key scientific tools are described: monitoring, modeling, and synthesis. For each, the committee briefly summarizes the current status and then identifies strategies that can be used to further the development and application of relevant and timely science for each tool.

Monitoring to Inform Decisions

Current Monitoring and Assessment to Support Decision Making

The CERP Monitoring and Assessment Plan (see NASEM, 2018; NRC, 2004; RECOVER, 2006, 2009) guides the collection of an impressive array of data on hydrology, water quality, and key ecological components, such as vegetation, wading birds, alligators, and oysters (Figure 6-1). These data are used to determine whether changes are occurring as a result of ongoing disturbance or restoration projects and to track performance measures. Every 5 years the System Status Report (SSR; see Chapter 3) provides regional analysis of the status and trends of different ecological attributes relative to identified targets for each attribute. Monitoring data are also collected specifically for each CERP project to evaluate progress and support adaptive management (see Chapter 3). Additionally, monitoring data are used to calibrate and validate hydrologic models and to develop response relationships for use in ecological and water quality models.

The ongoing data collection effort is impressive, but the program is failing to meet its potential. The value of data sets can be limited by lack of an analytical design targeted at the information most needed by decision makers. For example,

the COP Adaptive Management component, which demonstrates the utility of monitoring data for operational decisions, is being supported with existing monitoring (see Chapter 4). Existing monitoring designs intended for assessing general status and trends may not be sufficient to answer specific questions. In the northern estuaries, an improved, intensive seagrass monitoring program was initiated in 2018 to better inform predictive models that support estuary management decisions (see Chapter 5).

For the adaptive management to be effective, it is necessary to ensure that the data collected are adequate to address the highest priority questions. NASEM (2018) concluded that improvements are needed in the design of monitoring plans and that the ways data are analyzed can limit their usefulness. Furthermore, project-level monitoring varies in effectiveness (Chapter 3; NASEM, 2018).

Enhancing the Value of Monitoring

Monitoring typically falls into one of two categories: surveillance and targeted. Surveillance monitoring focuses on the status and trends of a system, while targeted data collection focuses on scientific hypotheses or management-related questions (Nichols and Williams, 2006). Targeted monitoring in a management context is connected to decision processes, which recognize the connections between objectives, potential management actions, models of response, confidence in models, and the monitoring program. The motivation for monitoring in the CERP is to meet both purposes, but, to do that effectively and inform decisions, monitoring should be specifically designed with the decision(s) in mind. Monitoring may need to be modified over time as responses to restoration are identified, new management questions arise (or some former questions no longer need to be addressed), and issues not previously anticipated need to be tracked.

One way to substantially improve the value of monitoring for decision makers is strategic monitoring design. Decisions will be best supported when monitoring plans are designed to address key management questions (e.g., will operational changes better allow the project to meet its goals?) in light of natural variability and sampling constraints (NASEM, 2017). Models can be used to optimize the design of monitoring station placement and gauging station density, as well as the adequate temporal frequency of field observations (Baker and Culver, 2010; Mclaughlin and Graham, 1986). Such an optimization process might lead to substantial cost savings for monitoring programs and could free up resources for other purposes, such as data analysis to answer critical management questions. Given the extensive resources directed toward project planning and support for construction design, assessing current monitoring in the light of decision needs can focus resources and ensure appropriate data are being collected.

Some of the best ways to improve the use of monitoring to inform decisions, however, are through improved linkage and integration of monitoring and modeling and increased focus on synthesis, including a strong data management system. These are discussed in the following sections.

Modeling to Inform Decisions

Status of Modeling Science in the Everglades

The use of models has been a key tool in the CERP, primarily for planning CERP projects. Model development has been combined with expert-level application in systems analysis to support large-scale project planning, such as the Central Everglades Planning Project (CEPP) and the COP. In such studies, the models are used to evaluate the predicted response of project alternatives on hydrology, water quality, and plant and animal abundance and distributions, typically using climatological conditions from a historical reference period. The results provide support for selecting among alternatives and thus achieving the optimal outcome. The modeling used for CERP planning is led by the Interagency Modeling Center, with additional modeling conducted by individual agencies.

Understanding of the relationships between processes that drive the outcomes of interest generally begins with conceptual models and then evolves with increased knowledge to increasingly refined mathematical models. The existing computational models represent the formalization of much of what is known about the flow and distribution of water and ecosystem response throughout the Everglades. There is currently a strong gradient in the degree of maturity, robustness, integration, and trust in results of models used for hydrology, water quality, and ecological modeling.

Conceptual understanding of hydrologic processes is mature, and mathematical models of hydrology are correspondingly robust. Hydrologic models used in Everglades restoration follow accepted approaches, with widely agreed upon modeling tools whose results are broadly trusted. However, there is a general lack of characterization of the errors and uncertainty associated with the models (see SFWMD, 2018e and USACE and SFWMD, 2014, Appendix G for prior efforts to propagate hydrologic model error through the decision process), and so the degree to which they should be relied on is largely unknown (see also Chapter 4). Uncertainties within complex multilayered numerical models can be large to the point of masking the system response to certain projects or actions. An additional challenge is that hydrologic models are commonly developed and implemented within specific geographic subregions of the Everglades system, resulting in a compartmentalized collection of models of individual subsystems.

Understanding of water quality processes is also advanced, albeit subject to much uncertainty at landscape scales. The adoption of water quality models in Everglades restoration is much less advanced and integrated than those used for hydrology, with several different approaches of varying degrees of sophistication and trust in use for different purposes. Simplified water quality models have been used for authorization and design of stormwater treatment areas (STAs),

and watershed loading models have been used to develop total maximum daily loads in sub-basins within the Everglades (see Chapter 5).

The adoption and use of ecological models in Everglades restoration is less advanced and less integrated than the hydrologic framework discussed above, although the number of ecological models and connectivity with hydrologic modeling tools continues to evolve (NASEM, 2016).¹ Ecological modeling is increasingly important in planning decisions (e.g., CEPP [see NRC, 2014], COP [see Chapter 4]).

The modeling effort has many strengths—notably the interdisciplinary nature of the modeling enterprise, including coupled hydrologic and infrastructure modeling components linked to ecological models, as well as the use of innovative machine learning techniques. Advances continue to be made in modeling, and the status of the different types of models described above is common in other systems. However, full integration of models across hydrology, water quality, and ecology (see Box 5-8) in the Everglades remains a challenge.

Using Models in New Ways to Support Restoration Planning, Implementation, and Operation

Although the use of models can be an effective approach for identifying the designs that perform best in the model versions of reality, the current usage also leaves open the question of whether the projects are well designed and selected for the conditions they will face. In addition, the potential (but largely unknown) differences between the model representation and reality raises the issue of how to evaluate project outcomes once they are implemented. Project benefits in planning that are described solely in a theoretical model future are difficult to compare with the current status of the system, which complicates general understanding of (and support for) project benefits and evaluation of actual outcomes. For example, if the results of a project differ from the modeled results, does that imply that the project is not performing as expected, or is it simply due to a gap in the model’s representation of reality? Or is the model discrepancy due to a prevailing condition that was not evaluated in the modeling exercise (e.g., the precipitation and temperature conditions occurring in real time)? These kinds of questions become central as the CERP moves from focus on project authorization to operating and adaptively managing the projects as they are implemented.

Models could be used to answer these questions by expanding the potential use and benefit of modeling. First, models could be used to extend the reach of observations, using up-to-date conditions and data assimilation to provide a consistent representation of the state of the system with which to compare moni-

___________________

¹ The ecological models are tracked at https://www.jem.gov/Modeling.

toring data and develop an improved understanding of the system. This approach could support adaptive management of CERP projects by providing the range of “expected outcomes” for a project and allowing detection of where outcomes are diverging from expectations, and thus require adaptive management action. Second, models could be used to understand the potential effects of external factors on restoration, such as sea-level rise and precipitation changes, which may in turn inform decisions related to the IDS and expectations related to restoration goals. In this section, these new modeling applications are discussed, which could enhance support for decision making.

Using models to extend the value of monitoring data.

Currently, it is difficult to deduce the status of Everglades restoration goals despite the substantial resources devoted to monitoring. This lack of a system-level view impedes clear communication about restoration progress (see Chapter 3). Understanding the status of the system is naturally difficult for such a large complex system; a number of specific factors contribute to this difficulty. First, the Everglades is subject to the random variability of natural processes, including weather and climate variability as well as the variability of ecosystem responses. These conditions make it difficult to conclude whether observations are reflecting the effects of newly implemented restoration projects, are simply arising due to random chance, or are related to factors other than restoration. Second, the large number of variables being monitored and the multipurpose nature of those variables, including economic, ecosystem, wildlife, and hydrologic objectives, complicates a simple summary of the state of the system. Furthermore, an understanding of how progress in one area or objective may affect progress in other areas in terms of trade-offs or synergies is needed to better assess the nature of restoration progress.

Existing models, supported by monitoring, provide the means to improve understanding and communication of the current status and near-term trajectory of the CERP. Unlike field observations, calibrated and verified model outputs provide a continuous and consistent representation of the state of the Everglades. Thus, the existing models could be used to provide a model-based status of the Everglades restoration. This can be achieved by simulating the current state of the system as a function of the observed external forcings on an annual or semiannual basis. The integration of models and observations can be further enhanced by using formal data assimilation methods (e.g., Kourafalou et al., 2015; Loos et al., 2020; Oke et al., 2015). By assimilating point observations, models can be used to create a coherent spatial and temporal representation of the status of the restored system promoting understanding of ecosystem dynamics beyond that possible with monitoring alone, in essence, a “nowcast” (see Box 6-2). This creates the best possible information summary using both

model output and observed data, potentially supporting multiobjective trade-off analysis to track current project effects and better understand and communicate the complex nature and status of the system. The creation of a near-real-time simulation or nowcast would allow experts a deeper consideration of the state of the Everglades and its response over the recent past to restoration actions, weather, and other external forces. The nowcast allows the articulation of status on all objectives in a consistent way, providing decision makers with a more holistic view of system response to management.

Use of models for performance assessment requires strong integration of models with observed data and also integration of modeling and monitoring teams. Members of modeling and monitoring teams may be separated within agency structures requiring a deliberate management approach to foster the interaction. As projects are implemented and data are obtained, frequent comparisons of the observed data and relative correspondence to model predictions need to be consistently made to enable reevaluation of model formulations and management strategies. With improved integration of modeling and monitoring

staff teams, updated model predictions could also be used to design more effective and efficient data collection programs. It could also allow the identification of trade-offs, synergies, and interactions between objectives.

The use of models for near-real-time simulation to better understand the effects of restoration projects can directly support adaptive management of those projects. At present, modeling to support project planning does not provide an expectation for project outcomes that can be compared with project monitoring. Without knowing whether observed deviations in project outcomes are due to the project or to external factors (e.g., anomalous rainfall), adaptive management is difficult. The effect of weather variability could be isolated from other factors by developing a set of stochastic weather time series that includes low-frequency variability due to the Atlantic Multidecadal Oscillation and the El Niño–Southern Oscillation and recent trends. It is also possible that this variability could be recreated by resampling of the historical record. This approach

allows the isolation of the effect of the particular weather that year in comparison to the range of random weather that might be experienced. The uncertainty associated with natural ecosystem response is more challenging to characterize, but if appropriate error distributions can be estimated (or assumed) this could also be incorporated within the analysis.

Using models in this way, as a strong tool for adaptive management, requires a new way of thinking commensurate with that associated with the move from project planning to system operation. Dedicating resources to a pilot application, jointly planned by managers and modelers, could demonstrate how the current use of models, based on historical hydrology, could be enhanced to understand how the system is responding to restoration efforts under current conditions.

The expanded use of models and observations represents a substantial effort, and the benefits described here must be viewed in light of the cost of achieving them. The collection and processing of current boundary conditions is a significant effort in addition to the modeling runs themselves. Although near-real-time conditions may be difficult to achieve in the short term, more frequent updating of inputs and simulation of current conditions, for example, on an annual basis, would still provide improved ability to understand restoration progress from a whole system perspective. The potential benefits to the CERP are great through the improved decision making in planning and in operations that such understanding would yield. In addition, the approaches described here can be used to enhance communication of restoration progress to the public and political leadership, improving their understanding of the nature of progress and the need for commitment to restoration.

Using models to understand the implications of an uncertain future. Models could be central tools to assess future scenarios of environmental conditions or other external drivers, such as sea-level rise and precipitation changes, that affect the entire system (as discussed in detail in NASEM, 2018), but to date the focus has been on planning and implementing the backlog of projects. As projects come online and operations influence restoration success, assessing how external drivers influence interactions among projects could provide lessons learned to inform decisions related to the IDS and expectations related to restoration goals.

CERP models could be used to better understand the effects of changing external conditions on the Everglades and the implications of those effects for restoration. For example, a few studies have used a small number of scenarios representing changes in mean precipitation, temperature, and sea-level rise to drive hydrologic and ecological models to assess potential impacts on the Everglades system (e.g., Aumen et al., 2015; Nungesser et al., 2015; Obeysekera et al., 2011, 2015). The results indicate substantial sensitivity of the Everglades

hydrology and ecology to the change scenarios, particularly the drying scenario (e.g., 10 percent reduction in average annual precipitation). However, because only a small number of possible futures have been considered, it remains difficult to deduce actionable information from the results. Given the difficulty of correctly predicting precipitation changes over the next 30 years, such analysis may have been viewed as speculative. However, working through plausible scenarios is an instructive way of anticipating potential adaptive management options in case more enduring climate changes happen (NASEM, 2018).

Recently a number of new methods for futures analysis, including vulnerability assessment, have emerged that focus on identifying ecological and water management tipping points and potential management responses (see Box 6-3). To better understand the implications of an uncertain future, RECOVER has initiated a vulnerability assessment. The ongoing vulnerability assessment is an example of how the CERP models that have been primarily used for project planning could be used to address the uncertainties of the future that threaten restoration success. However, it does not appear that the powerful modeling tools used for project development are to be used for this analysis. This would be a missed opportunity, since the models, as used in project development, are already configured to translate precipitation and temperature time series into hydrology variables and ultimately CERP objectives. Leveraging the incredible investment in these modeling tools to support vulnerability assessment ensures that outputs can be compared with planning results, and managers interested in restoration success can better understand what may lie ahead.

Communicating and Reducing Model Uncertainties

Uncertainty is a topic that is central to modeling and decision analysis and yet often the bane of decision makers. However, model uncertainty is ignored at the peril of misinformed decision making and failed restoration. The difference between model output and field observations are the errors that define the predictive uncertainty of a model. Characterization of uncertainty in model results is essential for adaptive management of CERP projects; when the observations fall outside the predicted range, managers need to understand whether this is indicative of a problem in the project operations or design or whether the results can be explained by uncertainty in the model (see also Chapter 4). Characterizing and communicating model uncertainty helps to set realistic expectations for project performance and also allows improvement of the models themselves.

Quantifying and specifying the uncertainty of model predictions will help set realistic expectations for the results of restoration actions. In the current use of models for project planning, model results are typically presented as a single

“best estimate” for the performance of each alternative over space and time for a given set of conditions. This best estimate of the effects of a particular project based on the modeling does not convey the range of possible outcomes based on both the uncertainty in the models and the difference between actual future conditions (e.g., precipitation patterns, climate) and the scenario(s) used to evaluate the alternatives. Figure 6-3 shows an example of how these differences may vary spatially. Uncertainty analysis conducted for the 2012 Louisiana Coastal Master Plan (CPRA, 2012) was conducted coastwide and by hydrologic basin (Habib and Reed, 2013). Although the coastwide values showed distinct differences between land area for Future Without Action compared to the Master Plan projects in place, the effects of uncertainty in model predictions on land area varied by basin. In some basins (e.g., Lower Terrebonne), the uncertainty in model outputs was greater than the difference between model runs with and without the projects, while in others (e.g., Mid Pontchartrain), the modeling showed greater land area with the projects than without (Figure 6-3). Decision makers should be aware of uncertainty ranges, so they can understand the range of possible outcomes of restoration for any individual aspect of the system (which may include no improvement or worse performance in some cases for some objectives). In addition, a better understanding of uncertainty can help stakeholders and decision makers better understand the trade-offs between alternatives, which may be minimal for some objectives if the difference in performance between them is small relative to the uncertainty of the estimate.

With communication of uncertainty, decision makers can also better understand the risks. For example, if the model estimate shows a slight improvement in an objective for a given alternative over the “Future Without Action” scenario, but the range of possible outcomes includes significant decreases in this objective, decision makers should be aware of this possibility (Figure 6-3). Communicating uncertainty is not easy, but acknowledging and accounting for uncertainty is the only way to improve the robustness of decisions, both to modeling uncertainty and to the other contextual uncertainties (e.g., climate change, sea-level rise) under which Everglades restoration takes place.

Uncertainties in predictions produced by mathematical models can never be fully eliminated, but predictive uncertainties can be quantified and gradually reduced, with monitoring data playing a key role. The combined use of models with observational data in a continuous process of feedback and integration, termed “living models” (Loos et al., 2020; Orouji et al., 2013; Wang et al., 2016), can improve model performance and reduce model uncertainty. Field observations are typically used to set up, calibrate, and validate numerical models; living models use new observations to reassess and improve the parametrization and validation of the numerical tools. New observations can also be used in concert with the models to explore sources of variability and, thereby, help improve understanding of complex ecosystems. Sources of differences between observation and model outputs include (1) process stochasticity, or natural variation, (2) observation error, and (3) model structure errors (Harwood and Stokes, 2003). Identifying different sources of variability is important because stochasticity is not reducible, whereas other sources (such as parameter uncertainty) may be reducible with additional measurement (Rose et al., 2015). Model errors can potentially be characterized with a stochastic error function (e.g., Vogel, 2017) to provide a realistic prediction envelope for all model results. This in turn provides the best estimate of the actual range of effects predicted by the model.

Expanding the Use of Models

Models are generally used to support project planning and design, but there is currently little evidence of consideration of their use in subsequent stages of restoration, including performance assessment and design of monitoring programs. In particular, the 2020 South Florida Environmental Report (SFWMD, 2020) describes the use of modeling to support infrastructure project planning for the C-11 impoundment and improvement of canals and the use of water quality models for STA design and watershed management planning, but does not describe uses of modeling in the restoration operational and management phases of projects.

___________________

² See https://www.aoml.noaa.gov/ocd/ocdweb/mares_reports.html.

issues. For example, Chambers et al. (2019) documented the state of knowledge of peat collapse and provided insights into the long-term dynamics of parts of the system, and Douglass et al. (2020) synthesized submerged aquatic vegetation dynamics in the Caloosahatchee. Such benchmark overviews of available information and understanding can be used to underscore interpretation of monitoring data and model outputs, support adaptive management decision making, and guide investment to address priority science needs.

Ways to Enhance Synthesis

Synthesis can take many forms. Data integration involves aggregating two or more potentially disparate data sets into an integral whole, typically to add new dimensions to the existing information or to address specific questions. Synthesis can also involve expanded and enhanced use of findings from different sources (e.g., distinct research disciplines, technologies, methodologies) in new contexts (e.g., through systematic review and meta-analysis). Conceptual synthesis bridges theories and paradigms that underpin previous studies.

For the CERP, integration and synthesis activities need to bring together not only monitoring program data, but data collected by others and relevant scientific developments in the Everglades and beyond. Conceptual synthesis tools for capturing current knowledge in a structured manner have been developed by the CERP (e.g., Ogden et al., 2005). This type of synthesis is a long-standing scientific practice. However, advances in computation and visualization techniques enable analytical approaches to synthesis that have been advanced through National Science Foundation (NSF)-funded synthesis centers at the National Center for Ecological Analysis and Synthesis (NCEAS) and the National Socio-Environmental Synthesis Center (SESYNC).³

Synthesis requires the application of disciplinary expertise and a systems perspective. Individuals who are good thinkers with considerable research experience and a good knowledge of relevant studies and system dynamics are needed, as well as those with skills in meta-analysis and other formal approaches to synthesis. Synthesis requires focus, and effective synthesis efforts typically require a strong commitment to its enterprise.

Given the level of effort involved in synthesis activities, the topics, scope, and periodicity need to be carefully considered and deliberately planned, targeted toward topics where synthesis could help the CERP move forward. For example, synthesis of research and data developed outside of the CERP on key issues (e.g., harmful algal blooms, nutrients [see Chapter 5]) can put CERP efforts in context. Over several years, and within the context of adaptive management

___________________

³ See https://www.nceas.ucsb.edu/ and https://www.sesync.org/.

and in support of CERP goals, a series of reports could be produced by experts in relevant fields for key issues (e.g., climate change, invasive species), subsystems (e.g., individual estuaries, Lake Okeechobee, STAs), or individual fauna (e.g., Cape Sable seaside sparrow) or landscape features (e.g., peat collapse). The concept is to go beyond the effects of an individual restoration project to consider emerging issues and how the system is changing and why. Insights would be gathered from available data and emerging research, and would draw in information generated by others (e.g., the Long-Term Ecological Research [LTER] program, university researchers, other state and federal agencies not directly involved in CERP). Several approaches to synthesis are highlighted in Box 6-4. These synthesis approaches build on available data and understanding to provide additional insight for use in project planning and restoration assessment, as well as other decisions.

The benefits of synthesis are worthwhile to pursue, and well-founded processes exist in the environmental science community. With a modest investment from the CERP and/or other parties, an ongoing synthesis program could

be established that would be highly beneficial to the CERP and allow for a broader understanding of natural resources in South Florida. One approach could utilize existing national synthesis centers (NCEAS or SESYNC) where staff skilled in different aspects of synthesis support synthesis projects and work with expert teams to develop synthesis products. Such an approach was used by the Bay-Delta Interagency Ecological Program, which worked with NCEAS to establish several workgroups to examine pelagic organism decline.⁴ CERP decision makers, with input from RECOVER and the Science Coordination Group, could identify priority topics for synthesis annually and work with synthesis centers to support groups of scientists to work on specific synthesis projects. Via this process, science synthesis needs can be identified, prioritized, and provisioned on a timely basis for integration into the restoration effort. The model also has the advantage of not additionally burdening staff from the CERP implementing agencies to develop and lead the synthesis work on top of existing missions. With this model, one or two high-quality synthesis outputs could be produced per year that could be highly relevant to the needs of CERP.

Data Management to Support Synthesis

Everglades researchers have collected vast amounts of different types of data that potentially can be used in synthesis. Because the data and information span a wide array of temporal and spatial scales and are provided by different agencies and principal investigators, strong data management is required to support leveraging of these data to inform decision makers on the effects of restoration. Much has been written about the characteristics of a good data management system (NASEM, 2017). Especially relevant to data synthesis are the principles in the “FAIR system” (see Box 6-5) that are intended to strengthen the ability to reuse the data in future studies. Modern databases such as the NSF DataOne⁵ and the Gulf of Mexico GRIIDC system⁶ are examples of data management systems with a focus on establishing data legacy.

The need for good data management systems is well known within the CERP and among Everglades researchers. Several good relevant systems have been developed such as EDEN (developed by the U.S. Geological Survey for automated real-time water level data), DBHYDRO (developed by the South Florida Water Management District [SFWMD] for hydrologic and water quality data), and the Florida Coastal Everglades LTER database system (part of NSF DataOne).

___________________

⁴ See https://www.nceas.ucsb.edu/workinggroups/ecosystem-analysis-pelagic-organism-declinesupper-san-francisco-estuary.

⁵ See https://www.dataone.org/.

⁶ See https://data.gulfresearchinitiative.org/.

CERPZone is a multiagency collaborative environment connected to several data management systems to enable storage, retrieval, and preservation of data and information relevant to Everglades restoration.⁷ There are also useful databases that apply only to parts of the system, such as the Coastal & Heartland National Estuary Partnership (CHNEP) Water Atlas,⁸ and other databases, such as the Watershed Information Network (WIN),⁹ that focus on specific measurements.¹⁰

A renewed commitment by all participants in CERP data collection activities to developing metadata (i.e., the data that describe the observations and observation process), use of existing databases and associated standards, and timely uploads of new or updated data, can better support the program and utility of the data in synthesis. Data management and infrastructure should be designed and maintained with a long-term vision, so that data in the system are usable to future scientists who were not involved in the collection of the data. Data management requires a commitment from program managers, staff, scientists, principal investigators, and consultants to comply with (and enforce) standards

___________________

⁷ See https://www.cerpzone.org/.

⁸ The CHNEP Water Atlas covers several northern estuaries on the west coast of Florida, including Charlotte Harbor, Estero Bay, and their contributing watersheds. See https://chnep.wateratlas.usf.edu/.

⁹ WIN is the Florida Department of Environmental Protection’s repository of environmental data from nonregulatory data providers in Florida. See http://prodenv.dep.state.fl.us/DearWin/public/welcomeGeneralPublic?calledBy=GENERALPUBLIC.

¹⁰ See also https://fcelter.fiu.edu/data/other-data-resources/index.html.

related to when data should be added to the data management system, the form of the metadata, and the quality checks performed. Although it is often the role of the investigator or laboratory to check the data quality, random checks by the CERP Quality Assurance Oversight Team (QAOT) help to strengthen the application of quality assurance protocols.

Data quality and effective management can provide both short- and long-term benefits. A solid data quality program can increase a database user’s time for analysis and interpretation and reduce the need for cleaning data. CERP’s QAOT has focused on the quality of laboratory and field measurements associated with CERP projects through documents, presentations, and laboratory and field audits. QAOT (2019) includes evaluation of quality audits for several projects, including three CERP projects, with a focus on water measurements in DBHYDRO. The importance of noting quality-related qualifiers in the metadata are noted in Table 6-1 so that those who use the data for assessments understand the limitations of the data and can screen the data appropriately. Although quality assurance is critical to the CERP data and its use in informing decision, budget cuts have reduced communication programs and restricted audits (QAOT, 2019), limiting the effectiveness of the program. For example, biological data are currently not evaluated, and laboratory checks are only for SFWMD and the U.S. Army Corps of Engineers (USACE). The EDEN program, which is based on automated sensor systems, has data checks for its real-time data.¹¹

Ready access to data without investing considerable time on cleaning and basic processing will allow it to be used in a more comprehensive and nimble manner. The recent Natural Resource Condition Assessment for Everglades National Park (Redwine et al., 2019) illustrates the considerable effort involved in assembling and synthesizing disparate data sets under the current system. The report was based on more than 100 data sets including GIS data, monitoring data, and information from publications. Although some data sets included metadata, they were often lacking in GIS data. The authors note, “The spatial scale of [the] EVER [Natural Resource Condition Assessment] makes assessment of data and synthesis among different resources more challenging, and it is this aspect of data summary that received the most effort by the NRCA ecologist.” Automating the updating of databases can improve the turn-around time for analyses and synthesis reports and increase management response times.

___________________

¹¹ See https://www.jem.gov/data/waterdepth.

or other CERP projects could be improved to better meet their objectives. The timeline of decision support for adaptive management and operational decisions will likely be shorter than the traditional 3-year project planning process and dispersed across many different projects, regions, and scales. The more nimble the decision process in adaptive management, the more quickly improved benefits can be realized.

The CERP envisioned the need for a structured approach to the integration of science and learning, and created RECOVER. RECOVER has specified roles in the assessment of monitoring data as CERP projects are implemented in support of adaptive management toward the systemwide goals of the CERP (see Box 2-1). However, with declines in staffing, RECOVER cannot meet the current demands for evaluation, assessment, and synthesis in addition to its own goals for keeping a systemwide, forward-looking vision, including vulnerability assessments and work on adaptive management (RECOVER, 2016). Recently, much of its staff time has been consumed by the needs of project planning as well as required reporting (e.g., System Status Report), with limited staff time dedicated to identification of science and monitoring needs to address changing conditions, systemwide modeling and analysis of ecosystem trends, or impactful synthesis. RECOVER leadership recognizes the current limited capacity of the RECOVER team, triggered in part by budget cuts and the loss of staff dedicated to the RECOVER program (Figure 6-4).

In a complex system such as the Everglades, experienced restoration scientists and engineers bring valuable insights and skills. Understaffing already affects the capacity of RECOVER to support decision making (Figure 6-2), and retirements and attrition pose ongoing workforce challenges. Therefore, attention is needed toward identifying the skills, capacity, and vision needed in the CERP science workforce to support decision making moving forward and developing a strategy to maintain that capacity.

This organization infrastructure should also include staff support for improved communications and science translation capacity dedicated to helping communicate results clearly and effectively to decision makers and the outside community. Although many effective science communicators exist across the restoration effort, they typically have technical responsibilities that serve as primary missions, leaving little time to serve in a communications role.

As the tasks to support decision making shift toward assessment, operations, and adaptive management, the CERP should take advantage of existing experience and knowledge. Project-level adaptive management teams should utilize, where feasible, those who previously worked on the project development teams in addition to experts in analysis of monitoring data. Extension of the role of science experts from the project delivery team through the entire

adaptive management chain could bolster learning and effectiveness. Currently within the USACE, projects are typically handed off to another agency or team for maintenance and operations after the project is built, with experience and learning developed in planning lost to the project implementation and operation.

Opportunities for using science effectively may also be hindered by organizational silos that separate CERP and non-CERP efforts. Critical learning opportunities exist within non-CERP efforts, such as the COP, which could inform CERP efforts in the central Everglades. However, the organizational infrastructure to support the COP remains undefined (see Chapter 4), which could undermine the potential outcomes of the CEPP. COP adaptive management could serve as a pilot of the organizational infrastructure needed to provide science support for adaptive management in CERP. CERP adaptive management decisions are best made in light of all opportunities for improvements, both within and outside of the CERP, if they are to achieve maximum effectiveness.

Finally, NASEM (2018) noted that the CERP could benefit from establishment of a formal central leadership with the responsibility to ensure adequate science for decision making. The report states: “Ensuring that investigative research and advances in tools and understanding are useful in a policy context requires

a programmatic approach directly linked to the CERP effort, which may be best championed by an independent Everglades Lead Scientist empowered to coordinate and promote needed scientific advances.” Although there are many capable, experienced scientists who provide insights and leadership within the restoration, the report notes: “There is no central leader to support Everglades restoration fully focused on a vision for science, its continued development, and application across agencies.” This remains the case and is further compounded by forthcoming retirement of key science leaders.

In this chapter the committee has demonstrated how monitoring, modeling, and synthesis can collectively be used to support the CERP as it moves from project planning to operations and management of the partially restored system. Centralized, focused, trusted science leadership is needed to ensure the diverse science enterprise is effective and meeting the needs of decision makers. The long-anticipated change in the program status, from planning to operations and adaptive management, requires a new approach to science leadership. The identification and prioritization of science needs to support critical restoration decisions, ensuring the adequacy and relevance of the CERP science enterprise, and fostering communication and use of science in the restoration effort, requires that the CERP identify and empower an individual or small dedicated team to lead the effort.

CONCLUSIONS AND RECOMMENDATIONS

The value of science—especially systems thinking and analysis—becomes even more important as the CERP pivots from a focus on planning and advancing individual projects to operations and management of the partially restored system. The transition from a focus almost exclusively on multiyear CERP planning efforts to providing support for ongoing adaptive management of numerous projects in parallel with ongoing planning of remaining projects will necessitate strengthened science support for decision making. CERP managers face an array of restoration decisions, including adaptive management either at the project or program level based on assessments of restoration performance, near-term operational adjustments, project sequencing, and investments in additional science. The best science should be actively integrated and synthesized to inform these decisions so that restoration benefits are maximized and opportunities for learning across both CERP and non-CERP projects are not lost. New and renewed strategies for monitoring, modeling, and synthesis can strengthen the science support for these decisions.

Some monitoring programs are falling short of their potential, and the value of data sets for decision making is being limited by lack of strategic monitor-

ing design targeted at the information most needed by decision makers. Decisions are best supported when monitoring is strategically designed to address identified management decisions and key management questions, considering natural variability and sampling constraints. Assessing how current monitoring supports decision needs (e.g., adaptive management, operations, science needs) can focus resources and ensure appropriate data are being collected as the program transitions from a focus on project planning to also support operations and management of the partially restored system.

To better support decision making, the use of models should be expanded, including applications such as assessments of restoration progress and evaluations of future scenarios and vulnerabilities. The CERP has invested significantly to develop a robust set of modeling tools to guide the restoration process, but to date these models have been used mainly for project planning. Restoration decision making would benefit if the CERP could apply its modeling tools to also investigate questions related to restoration progress, adaptive management, and potential future vulnerabilities. Consideration should be given to how these modeling tools can further benefit CERP decision making, including using models to increase understanding of the Everglades ecosystem and its response to changing external conditions. The increased use of models will require additional human and technical capacity for model application and development.

A concerted effort to systematically compare and integrate models and observations is needed to improve decision making. Observations should be compared with model results to better understand model errors and their cause, and to improve model performance. The uncertainty in model predictions should be quantified and used to assess the implications of model uncertainty on decisions. Assimilation of observations and models can also be used to create a more comprehensive view of the current state of the system and can enhance the understanding of the effects of CERP amid natural variability.

A list of priority synthesis topics should be developed annually to advance synthesis in a coordinated way and increase system understanding for management needs. The list should consider the types of synthesis needed to support decision making, the data and information expected to be available, strategies for catalyzing the synthesis, and estimates of resource needs. The skills and expertise of existing synthesis centers, as well as Everglades science experts, should be leveraged to support CERP synthesis needs.

A renewed commitment to best practices in data management from all participants in CERP data collection would better support the value of data to support decision making and promote more comprehensive and nimble synthesis efforts. The use of data to support all types of decision making depends upon effective data management, quality assurance systems, and ease of access

to a variety of users. All participants in CERP data collection activities should be required to abide by data quality assurance programs and contribute metadata and data to central and publicly accessible data management repositories in a reasonable time frame.

A nimble organizational infrastructure for science is needed to support restoration decision making in light of the CERP’s transition toward operations and adaptive management of multiple completed projects. Information alone does not guarantee effective decision making. Utilizing and integrating scientific information into decision making at appropriate times and in relevant ways is crucial. This infrastructure should include several key elements:

• Adequate staffing of appropriately trained scientists who can respond to management needs by analyzing, synthesizing, and communicating evolving relevant scientific information.
• Continuity of expertise to support adaptive management throughout the life cycle of restoration projects, bringing technical expertise developed during planning to bear on data analysis and assessment of restoration progress toward goals.
• Strong science leadership to provide an efficient and direct linkage between decision makers who need timely summaries of ongoing work and emerging issues and scientists conducting research, modeling, and monitoring. Strong science leadership is also needed to guide future investments in monitoring, modeling, and synthesis toward critical decisions and to help catalyze these efforts.

This page intentionally left blank.