CHAPTER EIGHT

Vision and Way Forward for S2S Earth System Prediction

Previous chapters in this report identified the societal value of predictions of the Earth system in the subseasonal to seasonal (S2S) time range; pointed out emerging science and technical capabilities that make advances in forecasts at these timescales possible; and identified areas that need substantial improvement. This chapter draws from the text, findings, and recommendations presented in previous chapters to develop a vision to serve as an inspirational yet possible target for a desired future state of S2S prediction in the next 10 years, and a set of research strategies to guide actions that are necessary to move toward that vision. All of the recommendations from previous chapters are organized within these strategies and together serve as the committee’s comprehensive research agenda for S2S forecasting over the next decade. Implementing the research agenda for improving S2S predictions will require collaboration between researchers and users to develop more useful forecast products, basic research to advance understanding of the processes governing predictability in the Earth system, exploiting these new discoveries in models, and melding existing with new modeling and computing capabilities. Thus the S2S research agenda should simultaneously foster work in areas that are nearing maturity with more ambitious objectives that may take a decade or more to fully realize.

VISION FOR THE NEXT DECADE

For the past several decades, weather forecasts on the scale of a few days have yielded invaluable information to improve decision-making across all sectors of society. Determining the total economic value of this forecasting information is an area of active research (Letson et al., 2007; Morss et al., 2008), but previous research indicates that a significant portion of annual U.S. gross domestic product (tens of billions or even trillions of dollars) is sensitive to fluctuations in weather (Dutton, 2002; Lazo et al., 2011; U.S. Department of Commerce, 2014). Certainly short-term forecasts play a vital role in helping society manage this economic exposure and the associated social risk. However, many critical decisions must be made several weeks to months in advance of potentially favorable or disruptive environmental conditions. As demonstrated by case studies and other information presented in Chapter 3, S2S forecasts have great

potential to inform such decisions across a wide variety of sectors. For example, it can take weeks or months to move emergency and disaster-relief supplies. Pre-staging resources to areas that are likely to experience extreme weather or an infectious disease outbreak could save lives and stretch the efficacy of limited resources. Similarly, emergency managers responding to unanticipated events such as nuclear power plant accidents or large oil spills are faced with the task of communicating the ramifications of such events on timescales that stretch well beyond a few days. There are many more such examples: naval and commercial shipping planners designate shipping routes weeks in advance, seeking to stage assets strategically, avoid hazards, and/or take advantage of favorable conditions; with improved knowledge of the likelihood of precipitation or drought, farmers can purchase seed varieties that are most likely to increase yields and reduce costs; and depending on the year, water resource managers can face a multitude of decisions about reservoir levels in the weeks, months, and seasons ahead of eventual water consumption (Table 3.1 lists additional examples).

S2S forecasts are already proving to be of value in making such decisions in sectors such as agriculture, energy, water resources management, and public health. However, many sectors have yet to exploit even the S2S information that is currently available. The Committee believes that the benefits of S2S forecasts to society will only increase as the quality of S2S forecasts improves, as more variables are represented in forecast products, and as social and computer science research and boundary institutions accelerate awareness of, access to, and use of S2S information. This potential of S2S forecasts to benefit society is only likely to grow due to the increased exposure to risk and increased severity and frequency of hazards expected with climate change and continued globalization.

Working iteratively with water resources professionals, emergency managers, military planners, and a myriad of other potential users to codesign new S2S forecast products and related decision-making tools has the potential to further expand use and enable stakeholders to derive much more value from S2S forecasts. Along with an enhanced focus on developing predictions of extreme and other disruptive events, such iterative engagement with forecast users has the potential to foster a stronger culture of planning across S2S and longer timescales, including adaptation and resilience to climate change. This could provide social and economic benefits that amplify and transcend the direct benefits of S2S forecasts themselves.

This evidence influenced the committee’s finding that more skillful and useful S2S forecasts—developed through sustained engagement with users and advances in basic knowledge and technological capabilities—could radically improve the basis for decision-making on S2S timescales. Emerging science and technical capabilities are

making rapid advances in S2S forecasts more likely than envisioned even 5 years ago. Advances both in technology (e.g., satellites, computing) and in science (e.g., model parameterizations, data assimilation techniques) are now on the horizon that make advances in S2S forecasting more feasible. Furthermore, the committee’s recommendations are targeted at areas where efforts are most needed and therefore investments are most likely to lead to advances.

Such advances now have the potential to increase the flow of benefits from S2S forecasts so that, in the committee’s view, they have high potential to outweigh the costs and effort associated with improving S2S forecasts. Thus the committee developed a vision to serve as a target for S2S predictions over the next decade: S2S forecasts will be as widely used a decade from now as weather forecasts are today. This is admittedly a bold vision because overcoming the challenges to developing S2S forecasting will take sustained effort and investment. However, the committee believes that realizing this vision is now possible within the next decade.

Achieving the committee’s vision in this report is not incompatible with other visions for Earth system prediction systems (such as for the creation of a Virtual Earth System (VES)—see Box 8.1), but it has the potential to become reality within a much shorter timeframe. The committee’s strategies and research agenda, which are presented next in this chapter, provide priority actions for moving toward this desired future state.

S2S RESEARCH STRATEGIES AND RECOMMENDATIONS

Maximizing benefits of S2S forecasts while minimizing the associated costs will be important for rapidly improving S2S forecasting. The committee drew on findings in Chapters 3 through 7 to develop four overarching research strategies to help prioritize activities in S2S forecasting and to organize activities so that they most directly support the vision to substantially expand the use of S2S forecast information in the next decade:

1. Engage Users in the Process of Developing S2S Forecast Products
2. Increase S2S Forecast Skill
3. Improve Prediction of Extreme and Disruptive Events and Consequences of Unanticipated Forcing Events
4. Include More Components of the Earth System in S2S Forecast Models

Fourteen associated recommendations derived from Chapters 3 through 6 describe research and aligned activities in the physical and social sciences that the committee has determined to have the greatest potential for advancement in each of the four

strategic directions. In addition, the committee proposes a set of supporting recommendations, derived from findings in Chapter 7, related to cyberinfrastructure and workforce. These are necessary for advancing the research strategies and achieving the committee’s vision for S2S forecasting. Additional activities envisioned by the committee to fall under each of the 16 main recommendations add further specificity and breadth to the research agenda. Although the main recommendations are placed under the research strategy or supporting activity that they primarily support, implementing each recommendation will often help to advance multiple strategies.

Collectively these strategies and recommendations constitute an S2S research agenda for the nation.

Figure 8.1 presents a schematic of the relationship between the strategies and supporting activities and the committee’s vision.

Research Strategy 1:
Engage Users in the Process of Developing S2S Forecast Products

As highlighted in Chapter 3, providing usable, valuable forecast information involves developing S2S forecast products that are more readily integrated into user decision-making. The committee envisions an S2S prediction system 10 years hence that is highly interactive with decision-makers from a wide array of sectors. To achieve this level of interaction, the user community must be brought into the research and development process sooner rather than later. In fact, a key committee finding is that the S2S research and operational prediction community would benefit from engaging in an iterative dialogue with the user community, beginning as soon as possible. Such a process can help to further prioritize the development of specific forecast variables and metrics, and ensure that data and resource-intensive retrospective forecasts, as well as the operational forecasts themselves, retain and exploit parameters that are most critical to user decision-making.

In order to maximize benefits of investments into improving S2S forecasts over time, there should be an ongoing effort to codesign forecast products on S2S timescales that match what is scientifically feasible with what users can make actionable. In many cases, this might involve a relatively straightforward extension of existing applications that have skill at shorter timescales and for which sophisticated users already exist. In other cases, there may be novel, actionable prediction products that can be identified through more extensive discussions between potential users and the developers of prediction systems and forecast products. Such discussions will be required to identify what operational S2S forecasts will look like, including how the skill of such forecasts will be verified. Public- and academic-sector boundary institutions, such as the National Oceanic and Atmospheric Administration (NOAA) Regional Integrated Sciences and Assessments Program (RISA) programs, the International Research Institute for Climate and Society (IRI) at Columbia University, and several private-sector companies, have already started these discussions. Leveraging the entire weather and climate enterprise will be necessary for further development of effective S2S products and services that maximize benefit to society.

Recommendations

Research into the use of S2S forecasts thus far indicates that users desire finer temporal and spatial resolutions, more actionable forecast variables (e.g., extreme, disruptive, and other important events as well as mean conditions), as well as a better understanding of how probabilistic S2S forecast information at varying levels of skill

can be integrated directly into decision-making. However, user needs and how they match with current forecast capabilities and barriers to use of forecasts have not been thoroughly investigated across sectors. An important first step in providing more actionable S2S forecast information is to develop a body of social and behavioral sciences research that leads to a more comprehensive understanding of the current use and barriers to use of S2S predictions. This includes a better understanding of specific aspects of products—for example, forecast variables, spatial and temporal resolutions, necessary levels of skill, formats—that would make S2S predictions more useful to different communities. This research is necessary to develop a high-level view of how S2S forecast systems and outputs might be designed to meet the basic needs of the broadest number of potential users.

Although all weather and climate forecasts are inherently probabilistic, this probabilistic nature becomes more difficult to disregard for forecasts at S2S and longer timescales than at shorter lead times. Probabilistic predictions in particular represent a significant hurdle for some forecast users, because there often are substantial differences between the large-scale probabilistic forecasts that are possible at S2S timescales and the specific information that decision-makers might currently find actionable. Research on the use of probabilistic forecast information is thus also necessary.

Recommendation A: Develop a body of social science research that leads to more comprehensive understanding of the use and barriers to use of seasonal and subseasonal Earth system predictions.

Specifically:

• Characterize current and potential users of S2S forecasts and their decision-making contexts, and identify key commonalities and differences in needs (e.g., variables, temporal and spatial scale, lead times, and forecast skill) across multiple sectors.
• Promote social and behavioral sciences research on the use of probabilistic forecast information.
• Create opportunities to share knowledge and practices among researchers working to improve the use of predictions across weather, subseasonal, and seasonal timescales.

Beyond the research recommended above, engaging the S2S research and operational prediction communities in an iterative dialogue with users is necessary to help ensure that forecasts systems, forecast products, other model output, and other decision-making tools maximize their benefit to society. This includes effective probabilistic forecasts products, verification metrics, and communication strategies. Ongoing

efforts will be needed to match what is scientifically predictable and technologically feasible at S2S timescales with what users can make actionable, as scientific skill, user needs, and user perspectives continually evolve. Such iterative efforts can also help stakeholders develop and implement decision-making strategies, such as “ready-set-go” scenarios, that utilize S2S forecasts together with shorter and longer lead information. These scenarios help organizations utilize a suite of forecasts with different lead times, promoting advance preparation for potential hazards even while forecast uncertainty is relatively high, and then adjusting actions as forecast lead times shorten and forecast uncertainty decreases. As mentioned above, private industry and “boundary organizations” within academia and the public sector (NOAA’s RISA program, the IRI at Columbia University, and many others) have already started such discussions. Efforts to further engage users in the iterative process of making S2S forecasts more actionable and used more in decision-making should build on the experience of this boundary workforce (see also section on Supporting the S2S Forecasting Enterprise below).

Recommendation B: Establish an ongoing and iterative process in which stakeholders, social and behavioral scientists, and physical scientists co-design S2S forecast products, verification metrics, and decision-making tools.

Specifically:

• Engage users with physical, social, and behavioral scientists to develop requirements for new products as advances are made in modeling technology and forecast skill, including forecasts for additional environmental variables.
• In direct collaboration with users, develop ready-set-go scenarios that incorporate S2S predictions and weather forecasts to enable advance preparation for potential hazards as timelines shorten and uncertainty decreases.
• Support boundary organizations and private-sector enterprises that act as interfaces between forecast producers and users.

Research Strategy 2: Increase S2S Forecast Skill

Operational weather and ocean forecasts have steadily increased in accuracy and lead time over the past few decades. However, there is still significant room for improving the skill of many S2S forecasts. An important prerequisite for achieving the vision of widely used S2S forecasts is to significantly improve the skill of forecasts so that users’ confidence in such predictions increases, and so that S2S forecasts can be applied to a range of decisions that requires higher forecast skill in order to act. Analogous to routine weather forecasting, there should be an emphasis on skillful, routine fore-

casts of Earth system components 2 weeks to 12 months in advance. Prediction at these timescales will necessarily be more probabilistic and less precise as to timing and spatial location than shorter-term weather forecasts, but there is strong evidence for predictability for many Earth system variables on S2S timescales. As discussed in Chapter 4, important sources of predictability on S2S timescales originate from (1) modes of variability (e.g., the El Niño Southern Oscillation [ENSO], the Madden-Julian Oscillation [MJO], the Quasi-Biennial Oscillation [QBO]), (2) slowly varying processes in the ocean and on the land surface (e.g., soil moisture, surface water, snow-pack, ocean heat content, ocean currents, eddy positions, and sea ice conditions), and (3) elements of external forcing (e.g., aerosols, greenhouse gasses).

Exploiting these sources of predictability to increase forecast skill will require developing better physical understanding of sources of S2S predictability, as well as improving all aspects of S2S forecast systems. This includes sustaining and improving the network of observations used to study predictability and to initialize models, developing improved techniques for data assimilation and uncertainty quantification in coupled Earth system models, and importantly, reducing Earth system model errors through a combination of increases in model resolution and the development of better model parameterizations to represent subgrid processes. Research to spur the development of new methods for probabilistic forecasting and probabilistic skill verification and calibration are also necessary.

With many possible avenues available for improving the skill of S2S forecasts, efforts to optimize the design of S2S forecast systems are also essential. S2S forecast systems can be configured in a wide variety of ways, and there are numerous possible selections and combinations of the design elements (“trade space”) in any forecast system. For example, what is the cost-benefit to the skill of S2S forecasts of adding more dynamical representation of different Earth system components and increasing the complexity of their coupling, versus increasing model resolution, extending retrospective forecast length or averaging period, increasing ensemble runs, and/or increasing the number of models in a multi-model ensemble system? While all may improve forecast skill, finite computing and human resources implies trade-offs in the design and implementation of any system. Thus a key part of improving and maximizing the cost-benefit relationship for producing probabilistic information 2 weeks to 12 months into the future will be to undertake a systematic exploration of the optimal use of available resources to support the development of more skillful forecast systems.

The development of a cost-effective and skillful operational multi-model ensemble forecast system is important, will require particular care and attention, and will involve the use of current operational models along with support for the research community

to actively engage in the development and validation of new or updated members of these ensemble systems (see discussion below). Similar methods for probabilistic skill verification and calibration also need to be employed and developed for the evaluation of the forecasts of probabilities. Such probabilistic skill metrics would characterize, and ultimately improve, the capability of forecasting common but also rare S2S events (Research Strategy 3).

Recommendations

Making S2S predictions relies on the identification and understanding of sources of Earth system predictability in the S2S time range. The 2010 NRC report (NRC, 2010b) identified a number of sources of predictability, including inertia in various slow-varying components of the Earth system, modes of variability in the coupled ocean-atmospheric system (e.g., ENSO, MJO), and external forcing (from either human or natural sources). Chapter 4 further explores current understanding of these sources of S2S predictability and emphasizes that much remains to be learned about these sources, especially their interactions and teleconnections. Research to advance understanding of sources and limits of predictability for specific Earth system phenomena will be critical to improving the fidelity of S2S Earth system models, as well as to improving the ability to forecast extreme or other disruptive events with longer lead times (Research Strategy 3).

Recommendation C: Identify and characterize sources of S2S predictability, including natural modes of variability (e.g., ENSO, MJO, QBO), slowly varying processes (e.g., sea ice, soil moisture, and ocean eddies), and external forcing (e.g., aerosols), and correctly represent these sources of predictability, including their interactions, in S2S forecast systems.

Specifically:

• Use long-record and process-level observations and a hierarchy of models (e.g., theory, idealized models, high-resolution models, global earth system models) to explore and characterize the physical nature of sources of predictability and their interdependencies and dependencies on the background environment and external forcing.
• Conduct comparable predictability and skill estimation studies and assess the relative importance of different sources of predictability and their interactions, using long-term observations and multi-model approaches (such as the World Meteorological Organization–lead S2S Project’s database of retrospective forecast data).

Chapter 5 emphasizes the importance of improving routine observations, developing more sophisticated data assimilation and uncertainty quantification techniques, reducing individual model errors through increased resolution and better parameterizations, and developing advanced calibration techniques and model combinations in order to develop more skillful S2S forecast systems. Routine observations are essential for initializing models to more accurately reflect the state of the Earth system and for validating model output; they can also contribute to improved understanding of the physical system and its predictability on S2S timescales (Chapter 4). Current observing networks for the atmosphere are more capable and robust than those for the other components of the Earth system. However, sustaining atmospheric observations is critical for S2S as well as weather forecasting; research to increase the use of currently available atmospheric observations, such as assimilation of satellite radiances in cloudy and precipitating areas, could unlock a wealth of new information related to representing convection and precipitation in models.

Relative to the atmosphere, the ocean, land surface, and cryosphere remain significantly under-observed, despite being major sources of S2S predictability. For the oceans, more routine and targeted observations are essential for S2S applications. In particular, sustaining and enhancing the capability to provide remotely sensed sea surface height (SSH), sea surface temperature (SST), and near-surface winds is critical, as is expanding the use of measurement arrays such as Argo floats and moored buoys to better measure key ocean properties below the surface (e.g., temperature, salinity, and current velocity). In addition to improving classic observing capabilities for the ocean, smart utilization of novel autonomous platforms could have an important impact.

Reliable and accurate year-round sea ice thickness measurements are the greatest need for improving the understanding and modeling of sea ice and its influence on the coupled system. Current (CryoSat2) and planned (ICESat2) satellite missions will help to meet this key objective. Because these satellites measure freeboard (the height of sea ice and snow above the sea level), accurate and simultaneous measurements of snow depths are also needed to solve for sea ice thickness. The procedure for solving for sea ice thickness needs to be efficient enough to be ready in about a day, so such measurements can contribute to initialization of S2S forecasts.

Land observations are critical for modeling large-scale land surface-atmosphere feedbacks and for predictions of the terrestrial water cycle. Several new satellite missions (Soil Moisture Active Passive [SMAP] and Surface Water and Ocean Topography [SWOT]) are focused on observing near-surface soil moisture and other aspects of surface hydrology that will be useful for improving S2S predictions. However, a number

of critical gaps remain. Lack of adequate precipitation measurements currently hinder S2S prediction, and measurements of soil moisture in the root zone, as well as measurements of evapotranspiration, are needed globally to better constrain hydrology and surface fluxes. Measurements of snow depth or snow water equivalent (SWE) are also critical. SWE can be estimated from existing satellite platforms; however retrieval algorithms must be improved in order to take full advantage of these observations. Because of gaps in the satellite observing network, in situ measurements of variables such as precipitation, snow depth, and land surface-atmosphere fluxes are likely to remain important and should be expanded to improve their spatial coverage.

In summary, observations of the atmosphere, ocean, land surface, and cryosphere play a critical role in building, calibrating, initializing, and evaluating the coupled Earth system models that are used to generate S2S forecasts. Better representing slow-varying processes in the Earth system—such as the ocean, cryosphere, and land surface hydrology—and their coupling to the atmosphere, as well as developing observations to inform deep convection and storm formation, are important to capturing S2S predictability, but they represent the largest gaps in the current observing network. Improved observations are also critical for improving the ability to forecast important and/or extreme events (Research Strategy 3). Including observations of phenomena that remain insufficiently observed, such as the properties of oceans or sea ice, can also facilitate the inclusion of more and more complex components of the Earth system in S2S prediction systems (Research Strategy 4).

Recommendation E: Maintain continuity of critical observations, and expand the temporal and spatial coverage of in situ and remotely sensed observations for Earth system variables that are beneficial for operational S2S prediction and for discovering and modeling new sources of S2S predictability.

Specifically:

• Maintain continuous satellite measurement records of vertical profiles of atmospheric temperature and humidity without gaps in the data collection and with increasing vertical resolution and accuracy.
• Optimize and advance observations of clouds, precipitation, wind profiles, and mesoscale storm and boundary layer structure and evolution. In particular, higher resolution observations of these quantities are needed for developing and advancing cloud-permitting components of future S2S forecast systems.
• Maintain and advance satellite and other observational capabilities (e.g., radars, drifters, and gliders) to provide continuity and better spatial coverage, resolution, and quality of key surface ocean observations (SSH, SST, and winds),

• particularly near the coasts, where predictions of oceanic conditions are of the greatest societal importance in their own right.

• Maintain and expand the network of in situ instruments providing routine real-time measurements of subsurface ocean properties, such as temperature, salinity, and currents, with increasing resolutions and accuracy. Appropriate platforms for these instruments will include arrays of moored buoys (especially in the tropics), autonomous underwater vehicles (AUVs), marine mammals, and profiling floats.
• Develop accurate and timely year-round sea ice thickness measurements; if from remote sensing of sea ice freeboard, then simultaneous snow depth measurements are needed to translate the observation of freeboard into sea ice thickness.
• Expand in situ measurements of precipitation, snow depth, soil moisture, and land-surface fluxes, and improve and/or better exploit remotely sensed soil moisture, SWE, and evapotranspiration measurements.
• Continue to invest in observations (both in situ and remotely sensed) that are important for informing fluxes between the component interfaces, including but not limited to land surface observations of temperature, moisture, and snow depth; marine surface observations from tropical moored buoys; and ocean observations of near-surface currents, temperature, salinity, ocean heat content, mixed-layer depth, and sea ice conditions.
• Apply autonomous and other new observing technologies to expand the spatial and temporal coverage of observation networks, and support the continued development of these observational methodologies.

As the scope of S2S models evolves to include and resolve more physical processes and components of the Earth system, there will be an increasing need for observations of new variables (Research Strategy 4). Furthermore, and as detailed above, current routine observations may not have sufficient resolution or coverage for S2S applications. Although it would be beneficial to expand the geographic coverage and resolution of many types of observations, cost and logistics will continue to demand that priorities are determined, and it is not always clear a priori what measurements will be most beneficial to support S2S prediction systems. Thus careful study of the improvements anticipated in S2S forecasting systems will be needed to quantify the cost-benefit ratio for various types of additional observations. Such study requires integrating ocean, land, atmosphere and sea ice modeling in the planning of observing networks. Observing system simulation experiments (OSSEs) and other sensitivity studies are powerful tools for exploring the importance of specific observations on state estimation and overall model performance, and could be better used to prioritize

improvements to observation networks (as well as model parameterizations) for S2S prediction systems.

Recommendation F: Determine priorities for observational systems and networks by developing and implementing observing system simulation experiments, observing system experiments, and other sensitivity studies using S2S forecast systems.

Many challenges are associated with integrating tens of millions of observations into the different components of an Earth system model, including ensuring that initializations are dynamically consistent and minimize the growth of errors. Given that coupling between the multiple, dynamic components of the Earth system (e.g., atmosphere, ocean, ice, land) is central to the S2S prediction problem, developing and implementing coupled data assimilation methods is at the forefront of S2S model development.

The implementation of “weakly coupled” assimilation, in which an independently coupled Earth system model is integrated forward in time as part of the assimilation process, represents an important and ongoing step in improving both weather and S2S forecast systems.“Strongly coupled” data assimilation, in which observations within one media are allowed to impact the state estimate in other components (with constraints), may allow for another important leap forward, especially for S2S systems in which the representation of the interaction between Earth system components is essential for capturing inherent predictability. However, research into the use of strongly coupled data assimilation algorithms is in its infancy, has not yet been tested on complex S2S coupled prediction models, and presently faces several barriers to implementation. Fundamental research is needed to explore and realize the potential benefits to more advanced but expensive strongly coupled data assimilation, while continuing to pursue and implement weakly coupled methods in current systems.

Efforts to improve the skill of S2S predictions will also benefit from more realistic representation of the uncertainty and statistical properties of observations and model output. Research on Bayesian data assimilation and uncertainty quantification has grown substantially in atmospheric and oceanic sciences and also in disciplines such as applied mathematics and engineering. These methods, which allow the optimal prediction and utilization of the full probabilistic information and utilize rigorous reduced-order differential equations, are strong candidates for implementation in the components of S2S prediction systems, but require more development to be implemented into operational settings.

Recommendation G: Invest in research that advances the development of strongly coupled data assimilation and quantifies the impact of such advances on operational S2S forecast systems.

Specifically:

• Continue to test and develop weakly coupled systems as operationally viable systems and as benchmarks for strongly coupled implementations.
• Further develop and evaluate hybrid assimilation methods, multiscale- and coupled-covariance update algorithms, non-Gaussian nonlinear assimilation, and rigorous reduced-order stochastic modeling.
• Optimize the use of observations collected for the ocean, land surface, and sea ice components, in part through coupled covariances and mutual information algorithms, and through autonomous adaptive sampling and observation targeting schemes.
• Further develop the joint estimation of coupled states and parameters, as well as quantitative methods that discriminate among, and learn, parameterizations.
• Develop methods and systems to fully utilize relevant satellite and in situ atmospheric information, especially for cloudy and precipitating conditions.
• Foster interactions among the growing number of science and engineering communities involved in data assimilation, Bayesian inference, and uncertainty quantification.

Systematic errors are numerous within the Earth system models used for S2S forecasting. For example, many global models produce an unrealistically strong Pacific equatorial cold tongue, a spurious double Inter Tropical Convergence Zone (ITCZ), erroneously high Indian Ocean and tropical South Atlantic SSTs, low SSTs in the tropical North Atlantic, wet or dry biases in rainfall in many parts of the world, and a bias in MJO variance. These model errors can be large compared to the predictable signals targeted by S2S forecasts.

Reducing such model errors represents one of the most important ways to improve the skill of S2S predictions (Chapter 5, models subsection). There is evidence that increasing the resolution of modeling systems (while still at resolutions that need deep convection parameterization) can reduce model errors. However, resolution is far from a panacea. Improving physical parameterizations of unresolved processes remains essential to reducing errors, even as the capability to resolve more and more processes expands. One important barrier to improving parameterizations is incomplete understanding of actual physical processes and the challenges associated with encapsulating new knowledge of these processes in (multiple, interacting) parameterizations. Coordinated, coupled field campaigns, as well as process-targeted satellite missions

and other observations, are essential for developing the understanding required to improve parameterizations. To maximize impact, field campaigns should, as far as possible, be codesigned by academics and operational centers and take full advantage of opportunities for national and international coordination.

Continuing to develop high-resolution research models will also be important for developing better parameterizations that reduce model errors. Model resolution encompasses time and space resolution, but also a balance between the order of the numerical computation and the refinement of the discretization. Further development of high-resolution models will also be extremely beneficial for examining predictability on S2S timescales, and will help pave the way for future operational use of global cloud and eddy-permitting models or cloud- and eddy-permitting meshes in critical areas. This approach is becoming more feasible as scale-aware cumulus parameterization schemes are being developed. Finally, in parallel to spatiotemporal model resolution and parameterizations, incorporating new stochastic statistical methods is also important to advance S2S forecasting. In particular, as described in Chapter 5, there are now several promising stochastic methods and reduced-order partial differential equations that could provide improved probabilistic forecasts for the same cost as running the present number of ensemble members. Furthermore, including efficient stochastic components in S2S modeling systems has the potential to increase the skill of S2S probabilistic forecasts and benefit decision making. For example, stochastic computing and stochastic parameterizations of unresolved processes (Palmer, 2014) can be used to better represent rare but significant S2S events.

To summarize, investment in research aimed at physical understanding and reducing model errors is seen by this committee as a top priority for improving the skill of S2S predictions. In addition to contributing to Research Strategy 2, reducing model errors also contributes to Strategies 3 and 4.

Recommendation H: Accelerate research to improve parameterization of unresolved (e.g., subgrid scale) processes, both within S2S system submodels and holistically across models, to better represent coupling in the Earth system.

Specifically:

• Foster long-term collaborations among scientists across academia and research and operational modeling centers, and across ocean, sea ice, land and atmospheric observation and modeling communities, to identify root causes of error in parameterization schemes, to correct these errors, and to develop, test, and optimize new (especially scale-aware or independent) parameterization schemes in a holistic manner.

• Continue to investigate the potential for reducing model errors through increases in horizontal and vertical resolutions in the atmosphere and other model components, ideally in a coupled model framework (see also Recommendation L).
• Encourage field campaigns targeted at increasing knowledge of processes that are poorly understood or poorly represented in S2S models, including tropical convection, ocean mixing, polar, sea ice and stratospheric processes, and coupling among different Earth system components (e.g., air-sea-ice-wave-land; troposphere-stratosphere; dynamics-biogeochemistry).
• Develop high-resolution (or multi-resolution) modeling systems (e.g., that permit atmospheric deep convection and non-hydrostatic ocean processes) to advance process understanding and promote the development of high-resolution operational prototypes (see also Recommendation I).

Verification metrics are important for tracking and comparing model improvements, and are also a critical part of enabling use and building trust in S2S forecasts. Understanding the different ways that users interpret forecasts and what they consider to be skillful is necessary to inform the development of better verification metrics (Recommendation B). Improving verification should also involve continued research on feature-based and two-step verification methods, along with consideration of how the design of retrospective forecasts and reanalyses can influence the ability of some users to directly evaluate the consequences of acting on forecasts at various predicted probabilities.

Recommendation J: Pursue feature-based verification techniques to more readily capture limited predictability at S2S timescales as part of a larger effort to improve S2S forecast verification.

Specifically:

• Investigate methodologies for ensemble feature verification including two-step processes linking features to critical user criterion.
• Pursue verification methodologies for rare and extreme events at S2S timescales, especially those related to multi-model ensemble predictions.
• Consider the benefits of producing more frequent reanalyses using coupled S2S forecast systems in order for the initial conditions of retrospective forecasts to be more consistent with the real-time forecasts, as well as for the purposes of predictability studies.

Multi-model ensembles (MMEs) are one of the most promising ways to account for errors associated with Earth system model formulation, and the use of MMEs is likely

to remain critical for S2S prediction. However, current MMEs are largely systems of opportunity, and research is required to develop more intentional MME forecast systems. S2S forecast systems, including the coupled Earth system model, the reanalysis, and retrospective forecasts, can be configured in a wide variety of ways. Careful optimizing of the configurations of a multi-model prediction system will include systematic exploration of the benefits and costs of adding unique models to an MME.

Today, little information is available about optimum configurations for individual or multi-model S2S ensemble forecast systems. It is likely that much can be gained in both skill and resource utilization by ascertaining which configurations produce optimum forecast systems, as defined by reliable probability forecasts and optimum levels of user-focused skill.

Forecast centers, private-sector users, and value-added providers use various calibration methods, but there has not been a comprehensive effort to compare methods or to find optimum approaches for the variables of most interest. Studies of the optimum configurations of S2S probability models (mentioned below) should include an attempt to evaluate calibration methods and ascertain whether some methods offer clear advantage over others, recognizing that some of these methods will likely be application-specific.

Exploring the “trade space,” that is, the configuration of S2S forecast systems, will be a large, complicated, and expensive endeavor, expanding as computer and Earth system modeling capabilities expand over the next decade or more, but determining how performance depends on configuration is a key task in any S2S research agenda. As such, this exploration would benefit tremendously from a central, coordinating authority and central funding, as well. Exploring the “trade space” will be important for increasing forecast skill, advancing the prediction of events (Research Strategy 3), and helping decide how to expand and design new S2S systems to include more complexity in S2S Earth system models (Research Strategy 4).

Recommendation K: Explore systematically the impact of various S2S forecast system design elements on S2S forecast skill. This includes examining the value of model diversity, as well as the impact of various selections and combinations of model resolution, number of ensemble perturbations, length of lead, averaging period, length of retrospective forecasts, and options for coupled sub-models.

Specifically:

• Design a coordinated program to assess the costs and benefits of including additional processes in S2S systems, and relate those to benefits from other investments, for example in higher resolution. In doing so, take advantage of

• the opportunity to leverage experience and codes from the climate modeling community.

• Encourage systematic studies of the costs and benefits of increasing the vertical and horizontal resolution of S2S models.
• Evaluate calibration methods and ascertain whether some methods offer clear advantage for certain applications over others, as part of studies of the optimum configurations of S2S models.
• Explore systematically how many unique models in an MME are required to predict useful S2S parameters, and whether those models require unique data assimilation, physical parameterizations, or atmosphere, ocean, land, and ice components (see also Recommendation L).

Transitioning new ideas, tools, and other technology between the S2S research community and operational centers is challenging but essential to translating research discoveries into informed decision-making. In the S2S context, one key element of this transfer will be to bring the best research to bear on developing a fully operational MME S2S forecast system. The use of MMEs in nonoperational, research, and real-time settings has demonstrated the potential for advancing S2S forecasts, for example the North American Multi-model Ensemble program (NMME) (see Chapter 6). An operational NMME relying on research institutions for funding and operations may not be a viable long-term option, but there would be great value in the development of a fully operational MME forecast system that includes the operational centers of the United States.

Developing an operational MME forecasting system will require careful optimizing of the configurations of a multi-model prediction system (Recommendation K). Test beds, such as the National Oceanic and Atmospheric Administration (NOAA) Climate Test Bed activity, provide the potential for such coordinating activities; however, the Test Bed would need significant enhancement if it were to be relied on as the primary mechanism for the development of an MME forecasting system. Although feasible, interagency and international collaborations could accelerate efforts to create an operational MME. Realistic assessment of available operational resources and centers that are able to contribute operationally rigorous prediction systems would be a useful starting point for determining the best path forward.

Recommendation L: Accelerate efforts to carefully design and create robust operational multi-model ensemble S2S forecast systems.

Specifically:

• Use test beds and interagency and international collaborations where feasible to systematically explore the impact of various S2S forecast system design elements on S2S forecast skill, in particular the question of how many unique models in a MME are required to predict operationally useful S2S parameters (see also Recommendation K).
• Assess realistically the available operational resources and centers that are able to contribute operationally rigorous prediction systems.

To make the kind of rapid improvements to operational S2S prediction systems that the committee envisions, it will be more generally important to speed the flow of information between scientists with research and operational foci. A number of mechanisms exist to improve the flow of technology into operational weather and ocean systems, including focused workshops, visiting scientist programs, special sessions at professional conferences, testbeds, and focused transition teams such as the Navy’s development-and-operations transition teams and the National Science Foundation (NSF)/NOAA’s Climate Process Teams. These mechanisms should be promoted and expanded to include more scientist involvement for plowing the new ground of S2S.

New mechanisms should also be developed especially to enhance researcher access to operational forecast data, including access to archives of ensemble forecasts themselves, retrospective forecasts, and initialization data. There are data storage challenges with such an endeavor, but it would facilitate further analyses of sources of S2S predictability and efforts to diagnose skill, among other benefits. The World Climate Research Programme/World Weather Research Programme (WCRP/WWRP) S2S Project described in Chapters 4 and 6 is already making some operational center data available to the research community to study subseasonal processes, but S2S Project data is just beginning to be explored by the research community.

In the longer term, allowing researchers to conduct or request specific experiments on operational systems would provide an additional boost to the flow of discoveries and technical advances between research and operations communities. Allowing researchers to run operational models will be a difficult challenge, one that involves making the modeling code accessible to the research community as well as ensuring access to sufficient computing power to run the code. All of these actions will require a significant effort on the part of the operational centers. To improve the flow of advances between research and operations, operational centers should work toward addressing these challenges over the next two decades.

Recommendation M: Provide mechanisms for research and operations communities to collaborate, and aid in transitioning components and parameterizations from

the research community into operational centers by increasing researcher access to operational or operational mirror systems.

Specifically:

• Increase opportunities for S2S researchers to participate in operational centers.
• Enhance interactions with the international community (e.g., the S2S Project and Asia-Pacific Climate Center [APCC]) and with the World Meteorological Organization (WMO) Lead Centers.
• Provide better access in the near term to archived data from operational systems, potentially via test centers.
• Develop, in the longer term, the ability for researchers to request reruns or perform runs themselves of operational model forecasts.
• Encourage effective partnerships with the private sector through ongoing engagement (see also Recommendation B).

Research Strategy 3:
Improve Prediction of Extreme and Disruptive Events and Consequences of Unanticipated Forcing Events

Within the efforts to improve the overall skill of S2S forecasts and to provide more actionable information to users, there are two areas that the committee believes deserve special attention (Research Strategies 3 and 4). Research Strategy 3 involves an increased focus on discrete events, and the committee made two recommendations to address this focus. The first is to emphasize the prediction of weather, climate, and other Earth system events that disrupt society’s normal functioning. Weather extremes and other relatively infrequent events can greatly disrupt society’s normal functioning and are therefore of significant concern to many users: drought and flood, strong storms with excessive precipitation, heat waves, and major wind events are all examples. A coordinated effort to improve the forecasting of these events could provide the huge benefits achieved by allowing communities more time to plan for, and mitigate the damages of, these events. Thus it is important to explore the possibilities of using model output to suggest the likelihood of such disruptive events. For some of these events, a quantitative estimate of probability would provide the opportunity to consider whether specific mitigation actions are cost-effective. However, whether action is justified would depend on the skill of the forecasts for extreme events as determined by a history of such forecasts.

Improved forecasting of extreme or disruptive events may entail an emphasis on

forecasts of opportunity—windows in time when expected skill for predicting specific events is high because of the presence of certain features in the Earth system—rather than simply predicting average conditions for given time periods, as is done today. Skillful extended-range prediction of such events may only be possible for certain phases of large-scale climate patterns, such as the seasonal cycle, ENSO, or MJO, or North Atlantic Oscillation (NAO), or may be contingent on interactions between these modes and other slowly varying processes. Moreover, skillful prediction of the probabilities of some types of disruptive events will be possible at these timescales, whereas others may not. Examples of events for which there is good evidence for predictive skill at S2S timescales include regional drought; watershed-scale melt-driven flooding; and significant shifts in hurricane tracks or land-falling events in various ocean basins (Vecchi et al., 2011). More research is needed to investigate the potential skill for forecasts of different types of disruptive events, with a focus on discovering the potential for so-called forecasts of opportunity.

In addition to the improved prediction of events within the Earth system, there are events driven by outside forces that have major—and potentially predictable—consequences on the Earth system. Such outside forces include volcanoes, meteor impacts, and human actions (e.g., aerosol emissions, widespread fires, large oil spills, certain acts of war, or climate intervention). Over the past 25 years, a number of these unusual natural or human-caused events have had, or were initially feared to have, the possibility of large-scale consequences to the Earth system (Chapters 3 and 6) and accompanying adverse impacts to a wide range of human activities.

Some consequences of high-impact events are predictable on timescales of weeks to a year. These events are unusual because they are of a nature or magnitude not represented within the recent past (for example, since the start of observational satellite climate records in about 1979), and hence do not have well-observed analogs that can be used to validate prediction systems. Moreover, depending on the nature of the event, the operational forecast systems may not be suited for predicting the event’s consequences. Although some of these events, such as the 1991 eruption of Mt. Pinatubo, had clear consequences for the global system for a year and beyond (0.3°C global-mean cooling averaged over the 3 years following the Pinatubo eruption), many other events had much smaller impacts than originally projected. However, these were notable in raising significant public concern that might have required action by decision-makers.

Recommendations

The improved prediction of extreme or disruptive events on S2S timescales is an extension of improvements in S2S forecasting skill (Research Strategy 2). But given the importance of having actionable information about these events for users and decision-makers (Research Strategy 1), the committee believes it is important to highlight the prediction of events as a separate strategy. Improving the prediction of such events will involve improved understanding of sources of predictability of extreme and disruptive events in the S2S time range. It will also involve ensuring that all relevant sources of predictability and their interactions are represented in Earth system models (Chapter 4).

Recommendation D: Focus predictability studies, process exploration, model development, and forecast skill advancements on high-impact S2S“forecasts of opportunity” that in particular target disruptive and extreme events.

Specifically:

• Determine how predictability sources (e.g., natural modes of variability, slowly varying processes, external forcing) and their multiscale interactions can influence the occurrence, evolution, and amplitude of extreme and disruptive events using long-record and process-level observations.
• Ensure the relationships between disruptive and extreme weather/environmental events—or their proxies—and sources of S2S predictability (e.g., modes of natural variability and slowly varying processes) are represented in S2S forecast systems.
• Investigate and estimate the predictability and prediction skill of disruptive and extreme events through utilization and further development of forecast and retrospective forecast databases, such as those from the S2S Project and NMME.

The second part of this research strategy involves using S2S forecast systems to predict the consequences of disruptive events caused by an unusual Earth system event, such as a volcanic eruption or a major oil spill. Such an outside event generates an immediate demand for scientific guidance for the public and policymakers about potential consequences. A flexible system for estimating Earth system consequences of such unusual forcing events would address a national need that has become evident several times over the past few decades.

The nation should develop a capability for estimating the range of possible impacts and consequences of unexpected but critical events such as volcanic eruptions,

nuclear detonations, widespread fires, or large spills of toxic materials (Chapters 3 and 6). Such a capability would need to be mobilized within 1 week and return preliminary results for S2S timescales and beyond as appropriate.

Performing regular full-scale exercises in collaboration with the response community in the spirit of war games would help improve this capability and maintain it for any new event. This could serve as a focal point and help to improve the connections of advances in the academic sector in modeling unexpected events with operational model development.

Recommendation N: Develop a national capability to forecast the consequences of unanticipated forcing events.

Specifically:

• Improve the coordination of government agencies and academia to enable rapid response to unanticipated events and to provide S2S forecasts using the unanticipated events as sources of predictability.
• Utilize emerging applications of Earth system models for long-range transport and dispersion processes (e.g., of aerosols).
• Increase research on the generation, validation, and verification of forecasts for the aftermath of unanticipated forcing events.

Research Strategy 4:
Include More Components of the Earth System in S2S Forecast Models

The other area that the committee believes requires more focused attention is accelerating the development of Earth system model components outside the troposphere—Research Strategy 4. As mentioned above, representing oceans, sea ice, land surface and hydrology, and biogeochemical cycles (including aerosol and air quality) in coupled Earth system models is more important for S2S predictions than for traditional weather prediction, because much of the predictability of the Earth system on these timescales arises from conditions outside the troposphere or from interactions between Earth system components. Operational S2S forecast systems increasingly utilize coupled Earth system models that include major Earth system components (e.g., ocean, atmosphere, ice, land) (Brassington et al., 2015; Brunet et al., 2010). However, the representation of processes outside the troposphere has generally been less well developed. Improving model representation of land surface and terrestrial hydrology, ocean, sea ice, and upper atmosphere—including fluxes and feedbacks between these components—will be important for increasing the skill of S2S forecasts. This includes

advancing the observations, modeling, data assimilation, and integrated prediction capabilities in those components. Those components that have significant interactions with the weather and climate system as a whole will need to be dynamically integrated into the operational forecasting systems. Other fields that do not contribute substantially to the evolution of the rest of the system could be predicted by post-processing operations or by independent activities after the primary forecasts have been carried out. However, as demand grows for forecasts of phenomena that are predictable on S2S timescales but that do not feedback strongly to the atmosphere, improving the dynamical representation of many of these Earth system processes in S2S prediction systems may also become important in its own right.

Representing interactions between the various Earth system components has become increasingly important for climate projections. Comprehensive Earth system models, which include composition, aerosols, vegetation, and snow and glaciers, are increasingly being used to provide projections on decadal to centennial timescales (e.g., Coupled Model Intercomparison Project Phase 5 [CMIP5]; IPCC, 2013; Taylor et al., 2012). The extension of operational S2S forecasts toward dynamic predictions of more of the Earth system will be carried out most effectively by leveraging these existing efforts.

Recommendations

Improving the representation of more components and variables of the Earth system in S2S forecasts, including the ocean, sea ice, biogeochemistry, and land surface, will produce information applicable to a new and wider range of decisions. Iterative interaction with forecast users (Research Strategy 1) can help determine what processes and variables are most important to include in coupled S2S systems as these systems evolve. Expanding the comprehensiveness of such component models and advancing their coupling in Earth system models will also help improve the overall skill of forecasts (Research Strategy 2).

Priorities for improving ocean models include both fundamental numerical capabilities and improved depictions of important oceanic phenomena that are currently omitted from most S2S forecasting systems, for example, tides and their interactions with storm surges, and oceanic mixing of nutrients. The dynamics of the near surface ocean are of particular importance for the coupled ocean at S2S timescales, so the representation of ocean boundary-layer turbulence and its interactions with waves and sea ice are a promising subject of study for improving S2S forecasts. But the most important limitation on oceanic S2S forecasts arises from the global influence of the

ocean at these timescales, along with the need to accurately represent many important oceanic phenomena at relatively small scales to capture this influence. Implementing a regionally eddy-resolving ocean component along with additional research on parameterizing the effects of unresolved baroclinic and submesoscale oceanic eddies would thus help improve S2S coupled prediction models.

Sea ice models used for S2S often contain only rudimentary thermodynamics and dynamics. Connecting advances in cutting-edge sea ice models (including more sophisticated physics representations of ice-thickness distribution, melt ponds, biogeochemistry, and divergence/convergence, as well as new methods to account for wave-floe interactions, blowing snow, and ice microstructure) with sea ice models used in S2S forecast system could advance S2S predictions of the atmosphere through improved representation of radiative and ocean feedbacks, as well as advancing S2S prediction of sea ice and polar ocean conditions.

Similarly, land surface models used for S2S prediction need to improve treatment of the hydrological cycle and aspects of the land surface that are coupled to hydrology, such as vegetation. Effort is needed to incorporate surface and underground water storage and river routing in models, including the role of human water management and use. These important aspects of the land system have been implemented in “off-line” hydrologic forecast systems, but they are usually oversimplified or neglected altogether in fully coupled S2S forecast systems. Improving the representation of land surface processes such as soil moisture storage and snow in such fully coupled systems will be important for predicting events such as heat waves, cold surges, and storm formation. In addition, predicting runoff may also help to enable S2S forecasts of flooding and lake and coastal hypoxia.

Additional strong candidates for improvements to existing practice for operational S2S forecasting systems include advancing the observations, modeling, data assimilation, and integrated prediction capabilities of aerosols and air quality, and aquatic and marine ecosystems.

Beyond advancing the representation of the land surface, hydrology, stratosphere, sea ice, ocean, and biogeochemical models and translating these advancements to the coupled Earth system models used for S2S forecasting, efforts are needed to pave the way toward global cloud-/eddy-resolving atmosphere-ocean-land-sea ice coupled models, which will one day become operational for S2S prediction. Although this goal is unlikely to be reached in the next decade, revolutions in the computing industry may shorten the distance between now and the otherwise long way to go, and the S2S research community needs to be proactive and poised if/when that happens.

Recommendation I: Pursue next-generation ocean, sea ice, wave, biogeochemistry, and land surface/hydrologic as well as atmospheric model capability in fully coupled Earth system models used in S2S forecast systems.

Specifically:

• Build a robust research program to explore potential benefits (to S2S predictive skill and to forecast users) from adding more advanced Earth system components in forecast systems.
• Initiate new efficient partnerships between academics and operational centers to create the next generation model components that can be easily integrated into coupled S2S Earth system models.
• Support and expand model coupling frameworks to link ocean-atmosphere-land-wave-ice models interoperably for rapid and easy exchange of flux and variable information.
• Develop a strategy to transition high-resolution (cloud-/eddy-resolving) atmosphere-ocean-land-sea ice coupled models to operations, including strategies for new parameterization schemes, data assimilation procedures, and multi-model ensembles.

Supporting the S2S Forecasting Enterprise

It is essential to highlight two specific cross-cutting challenges that must be met to support the four research strategies for reaching the committee’s vision for S2S prediction. These are (1) ensuring that the computational infrastructure is sufficient to support the S2S forecasting enterprise and (2) developing and maintaining the workforce that will be needed to realize potential advances in S2S forecasting. These challenges are not necessarily unique to the S2S enterprise—they are also faced by the numerical weather prediction and climate modeling communities, and indeed, across many other technical enterprises.

Recommendations

The volume of observational data, data assimilation steps, model outputs, and reanalysis and retrospective forecasts involved in S2S forecasting means that the S2S modeling process is extremely data intensive. S2S prediction systems test the limits of current cyberinfrastructure, as do weather forecasting and climate modeling. Advances in S2S forecast models (e.g., higher resolutions, increased complexity, generation and retention of long retrospective forecasts), will require dramatically increased comput-

ing capacities (perhaps 1,000 times) and similar expansion of related storage and data transport capacities.

That said, today’s Earth system models are not taking full advantage of current computing architectures, and improving their performance will likely require new algorithms that do more to work on data locally before transporting it to those analyzing them, as well as significant refactoring of existing algorithms to exploit more parallelism. To compound these challenges, the transition to new computing hardware and software through the next decade will be highly disruptive. This transition will not involve faster processing elements, but rather more processors with considerably more complex embodiments of concurrency. In addition, future storage technology will be more complex and varied than it is today, and leveraging these innovations will require fundamental software changes.

An integrative modeling environment presents an appealing future option for facing some of the large uncertainty about the evolution of hardware and programming models over the next two decades. New approaches to data-centric workflow software that incorporates parallelism, remote analysis, and data compression will be required to keep up with the demands of the S2S forecasting community.

Recommendation O: Develop a national plan and investment strategy for S2S prediction to take better advantage of current hardware and software and to meet the challenges in the evolution of new hardware and software for all stages of the prediction process, including data assimilation, operation of high-resolution coupled Earth system models, and storage and management of results.

Specifically:

• Redesign and recode S2S models and data assimilation systems so that they will be capable of exploiting current and future massively parallel computational capabilities; this will require a significant and long-term investment in computer scientists, software engineers, applied mathematicians, and statistics researchers in partnership with the S2S researchers.
• Increase efforts to achieve an integrated modeling environment using the opportunity of S2S and seamless prediction to bring operational agency groups (e.g., the Earth System Prediction Capability [ESPC]) and integrated modeling efforts (e.g., the Interagency Group on Integrative Modeling [IGIM]) together to create common software infrastructure and standards for component interfaces.
• Provide larger and dedicated supercomputing and storage resources.
• Resolve the emerging challenges around S2S big data, including development

• and deployment of integrated data-intensive cyberinfrastructure, utilization of efficient data-centric workflows, reduction of stored data volumes, and deployment of data serving and analysis capabilities for users outside the research/operational community.

• Further develop techniques for high-volume data processing and in-line data volume reduction.
• Continue to develop dynamic model cores that take the advantage of new computer technology.

As highlighted in Chapters 3 and 6, the committee believes there are significant challenges in maintaining a pipeline of talented workers in the S2S enterprise. S2S is complex and involves working across computing–Earth science boundaries to develop and improve S2S models and working across science–user decision boundaries to better design and communicate forecast products and decision tools.

From the limited data available, it appears that the cadre of trained S2S modelers is not growing robustly in the United States and is not keeping pace with this rapidly evolving field (Chapter 7). Given the importance of S2S predictions to the nation, a concerted effort is needed to entrain, develop, and retain a robust S2S workforce.

Similar to weather forecasting, S2S forecasts are used or have the potential to be used by many people to make important decisions. Because S2S connects in a very public way to risk management, many opportunities will exist within the S2S enterprise to help society better manage risks. These factors can be exploited to entrain more talented and mission-driven people into the field.

One possible concrete step forward would be a series of workshops to explore how to feature S2S in more undergraduate and graduate curriculums, how to identify and connect with organizations that can help with this effort (e.g., the National Science Teachers Association), and how to interact with the private sector to help understand what skills are needed. Other entities such as the American Meteorological Society (AMS) or the NSF could also play an important role in coordinating the entrainment of talented young people.

Recommendation P: Pursue a collection of actions to address workforce development that removes barriers that exist across the entire workforce pipeline and increases the diversity of scientists and engineers involved in advancing S2S forecasting and the component and coupled systems.

Specifically:

• Gather quantitative information about workforce requirements and the

• expertise base to support S2S modeling to more fully develop such a training program and workforce pipeline.

• Improve incentives and funding to support existing professionals and to attract new professionals to the S2S research community, especially in model development and improvement, and for those who bridge scientific disciplines and/or work at component interfaces.
• Expand interdisciplinary programs to train a more robust workforce to be employed in boundary organizations that work in between S2S model developers and the users of forecasts.
• Integrate basic meteorology and climatology into academic disciplines, such as business and engineering, to improve the capacity within operational agencies and businesses to create new opportunities for the use of S2S information.
• Provide more graduate and postgraduate training opportunities, enhanced professional recognition and career advancement, and adequate incentives to encourage top students in relevant scientific and computer programming disciplines to choose S2S model development and research as a career.

CONCLUSION

This report envisions a substantial improvement in S2S prediction capability and expects valuable benefits to flow from these improvements to a wide range of public and private activities. It sets forth a research agenda that describes what must be done—with observations, data management, computer modeling, and interactions with users—to advance prediction capability and improve societal benefits.

Despite the specificity of the report in recommending what should be done, it does not address the challenging issues of how the agenda should actually be pursued—that is, who will do what and how the work will be supported financially. Given that this research agenda significantly expands the scope of the current S2S efforts, the committee believes that some progress can be made with current levels of support and within current organizational structures, but fully achieving the S2S vision will likely require additional resources for basic and applied research, observations, and forecast operations. The scope of the research agenda will also require closer collaboration between federal agencies and international partners, better flow of ideas and data between the research and operational forecasting communities, and engagement of the entire weather and climate enterprise.

The four research strategies provide broad guidance for how to focus effort, and the recommendations under each strategy in and of themselves represent the committee’s

view of the most important actions to advance S2S forecasting, presented without any prioritization or sequencing. The technological, political, and financial environment in which the research agenda will be implemented is constantly changing and will continue to be fluid, and multiple pathways to success exist. As such, the committee believes it was more important to provide a list of the most important areas where progress can be made toward improving S2S forecasts without overly prescribing the sequence or priority in which they should be addressed. All of these actions can improve S2S forecasting, and the more that is done to implement these recommendations, the more advances can be made.

To help agencies and other actors within the weather/climate enterprise select specific parts of the research agenda to pursue, Table 8.1 provides additional details about both the main recommendations and more specific or related activities that the committee envisions to be part of implementing each main recommendation: whether they involve basic or applied research; which are expected to have short-term benefits; which might require a new initiative; and which have a scope that calls for international collaboration to leverage U.S. effort. Although recognizing that it might not be possible to pursue all of these actions simultaneously, the committee hopes that these strategies, recommendations, and designations can help to guide progress across the span of recommended S2S research and forecasting activities.

The vision for the future of S2S forecasting can be achieved with a national will to pursue this research agenda and to convert the results into daily operations. The more that can be pursued within this research agenda, the closer the nation can be toward realizing the full potential of S2S forecasting and the more benefits that can be produced for a wide range of users and the nation as a whole.

TABLE 8.1 The committee’s 16 main recommendations—lettered in the order they appear in the report—are shown in bold typeface along with information to help guide their implementation. The committee sometimes recommends more specific or related activities that it envisions to be part of implementing each main recommendation. These are listed in plain text under each main recommendation. The second column indicates the research strategy that each recommendation and associated activity primarily supports (colors are the same as in Figure 8.1). Additional research strategies (1-4) supported by each recommendation are indicated by numbers. The final columns contain the committee’s opinion on whether each recommendation will involve mainly basic or applied research/operational activities, or both; whether a short-term return on investment is likely (≤ 5 years); and whether a new initiative or program, or a significant expansion of a program, may be necessary to implement each recommendation. The last column indicates recommendations for which the committee believes that international collaboration and coordination is particularly important.

Recommendation	Research Strategies	Basic Research	Applied Research / Operational	Benefits Likely in the Short Term	May Need New Initiative	International Collab. Critical
Chapter 3
A: Develop a body of social science research that leads to more comprehensive and systematic understanding of the use and barriers to use of seasonal and subseasonal Earth system predictions.	1, 4			∎
Characterize current and potential users of S2S forecasts and their decision-making contexts, and identify key commonalities and differences in needs (e.g., variables, temporal and spatial scale, lead times, and forecast skill) across multiple sectors.	1, 4		∎	∎
Promote social and behavioral science research on the use of probabilistic forecast information.	1	∎		∎
Create opportunities to share knowledge and practices among researchers working to improve the use of predictions across weather, subseasonal, and seasonal timescales.	1			∎