Copernicus Marine Environment Monitoring Service (CMEMS) Service Evolution Strategy: R&D priorities

Transcription

1 Copernicus Marine Environment Monitoring Service (CMEMS) Service Evolution Strategy: R&D priorities Version 3 June 2017 Document prepared by the CMEMS Scientific and Technical Advisory Committee (STAC) and reviewed/endorsed by Mercator Ocean

2

3 0 0 Table of content 1 Introduction Scientific and technical backbone of CMEMS Key drivers of the Service Evolution and related R&D priorities R&D areas and required developments Circulation models for the global ocean, regional and shelf seas Sub-mesoscale - mesoscale interactions and processes Coupled ocean-marine weather information, surface currents and waves New generation of sea-ice modelling modelling and data assimilation for marine ecosystems and biogeochemistry Seamless interactions between CMEMS and coastal systems Coupled ocean-atmosphere models with assimilative capability Ocean climate products, indicators and scenarios Observation technologies and methodologies Observing systems: impact studies and optimal design Advanced assimilation for large-dimensional systems High-level data products and big data processing The service evolution roadmap for 2021 and beyond The targeted CMEMS service in and after Tier-3 R&D marine projects of interest to CMEMS Tier-2 R&D marine projects selected by CMEMS Service evolution roadmap and milestones until Appendices Service Evolution Strategic documents CMEMS Service Evolution call 21-SE-CALL1 selected projects CMEMS Service Evolution: R&D calls for Global MFc Report from the Copernicus Coastal Workshop Context and workshop objectives Main outcomes and recommendations... 36

4 1 1 1 INTRODUCTION The Copernicus Marine Environment Monitoring Service (CMEMS) provides regular and systematic reference information on the physical state, variability and dynamics of the ocean and marine ecosystems for the global ocean and the European regional seas. This capacity encompasses the description of the current ocean situation, the variability at different spatial and temporal scales, the prediction of the ocean state a few days to weeks ahead (forecast), and the provision of consistent retrospective data records for recent decades (reprocessed observations and re-analyses). CMEMS provides a sustainable response to European user needs in four areas of benefits: (i) maritime safety, (ii) marine resources, (iii) coastal and marine environment, (iv) weather, seasonal forecast and climate. A major objective of the CMEMS is to deliver and maintain a competitive and state-of-the-art European service responding to public and private intermediate user needs, and thus involving explicitly and transparently these users in the service delivery definition. A Delegation Agreement 1 has been signed between the European Commission and Mercator Ocean for the CMEMS implementation during the period. It mandates the development and maintenance of a Service Evolution Strategy (see also section 6.1) in order to respond to new needs and take advantage of improved methodologies. This strategy is the mechanism by which lessons and knowledge derived during the past, as well as new research results, are used to guide change and/or a potential service upgrade in the subsequent phases. The Service Evolution Strategy is maintained on the basis of direct feedbacks from users, scientific and technical gaps analysis of emerging and existing user requirements, and the potential to improve the Core Service elements. At the same time, the Service Evolution Strategy identifies research needs that can guide external (e.g. H2020) and internal (CMEMS) R&D priorities. Consistent updates of the Service Evolution Strategy and its R&D priorities are made on an annual basis. This document is the second update of the R&D roadmap which identifies the essential Science and Technological (S&T) developments needed for the evolution of CMEMS during the period. Different categories of activities are conducted, with different time horizons, players and objectives: long-term objectives (Tier-3, beyond 2 years up to ~10 years), corresponding to the long-term evolution of the CMEMS framework; 1 See

5 2 2 mid-term objectives (Tier-2, 1 2 year cycle) addressed by dedicated R&D project teams, for implementing and assessing major scientific upgrades of the service during Phase-I ( ) and mostly Phase-II ( ) (see technical annex of the Delegation Agreement); short-term objectives (Tier-1, several months to 1 year cycle) with activities for addressing issues requiring fast responses for rapid implementation within the existing phases of CMEMS. Long-term R&D activities, although implemented in a different framework, are as crucial as short and mid-term activities for the sustainable evolution of the Service. The short and midterm activities are addressed both through internal CMEMS activities and open R&D calls, while long-term activities are promoted in the framework of external projects (e.g. Horizon 2020 or other European and national R&D programmes).

6 3 3 2 SCIENTIFIC AND TECHNICAL BACKBONE OF CMEMS The backbone of the CMEMS relies on an architecture of production centres both for observations (Thematic Assembly Centres TACs) and modelling/assimilation (Monitoring and Forecasting Centres MFCs) and a Central Information System (CIS). The CMEMS architecture for the phase 2 ( ) consists of: Eight Thematic Assembly Centres (TACs), including six space TACs organized by ocean variables (sea surface topography, ocean colour, sea surface temperature, sea ice, waves, and winds), one TAC for in situ observations, and one TAC for multiobservations (combining both in situ and satellite observation). TACs gather observational data and generate elaborated products, e.g. multi-sensor data products, derived from these observations. TACs are fed by operators of space and in situ observation platforms and infrastructures; the detailed operational requirements for space and in situ observation data were initially specified in the Marine Core Service Strategic Implementation Plan (Ryder, 2007 GMES Fast Track Marine Core Service, Strategic Implementation Plan, Final version 24/04/2017 ). Seven Monitoring and Forecasting Centres (MFCs), distributed according to the marine area covered (Figure 1, including Global Ocean, Arctic Ocean, Baltic Sea, North Atlantic West Shelf, North Atlantic Iberia-Biscay-Ireland area, Mediterranean Sea and Black Sea). MFCs generate model-based products on the ocean physical and biogeochemical states, including forecasts, analyses and reanalyses. A Central Information System (CIS), encompassing the management and organization of the CMEMS information and products, as well as a unique User Interface. TACs, MFCs and CIS contractors for the phase 2 of CMEMS (04/ /2021) will be selected following a call for tender that will be issued during the 2 nd semester of Figure 1. Regions covered by CMEMS.

7 4 4 3 KEY DRIVERS OF THE SERVICE EVOLUTION AND RELATED R&D PRIORITIES The guidelines to identify the R&D activities to be organized within Copernicus or in coordination with the service are motivated by several key drivers: Requirements from core users in the four main areas of benefit as recalled in the Introduction, accounting for both existing needs and needs likely to emerge in the future; Requirements from the CMEMS production centres based, in particular, on their 3-6 year R&D plans, considering the limitations in the daily service due to knowledge or technological gaps identified to elaborate the products; Requirements motivated by the development of Copernicus downstream services; Requirements from European and international programmes dealing with Earth observations, operational oceanography and Copernicus; notably GODAE OceanView, the Copernicus Integrated Ground Segment (IGS) and GEOSS (Global Earth Observation System of Systems); Outcomes of EU projects implemented under the Space theme (R&D to enhance future GMES applications in the Marine and Atmosphere areas), such as MyWave, OSS2015, E-AIMS, OPEC and SANGOMA or under FP7 and H2020 projects (e.g. JERICO, PERSEUS, AtlantOS). Additional drivers that reflect a more strategic view for the service evolution are also taken into account, such as: EU directives implemented to regulate activities in the Marine sector (e.g., Marine Strategy Framework Directive, Maritime Spatial Planning); New scientific and technological opportunities in the next 5-10 years, regarding both satellite (e.g. the Sentinels suites) and in situ observations (e.g. plans for the global BGC-Argo extension), modelling and assimilation capabilities, new communication and data processing technologies, etc. The need to maintain competitiveness w.r.t. international players in operational oceanography, as keeping a high-level know-how and innovation capacity will be strategic to attract new users; Forthcoming activities involving Copernicus developments in the framework of other European (e.g. EMODnet, EuroGOOS, EOOS) and global initiatives or programmes

8 5 5 (e.g. the Global Ocean Observing System in the context of GEOSS, GEO Blue Planet, GODAE OceanView); The Sustainable Development Goals (SDG) defined as part of the UNEP 2030 Agenda (especially SDG 14 - Conserve and sustainably use the oceans, seas and marine resources for sustainable development); Outcomes of the 21 st Conference of the Parties (COP21) to the UNFCCC (United Nations Framework Convention on Climate Change) including the need for a global stocktake to support emission agreements. The constant dialogue with user communities as outlined above helped in identifying overarching themes for the Service Evolution, focusing on: i. A better description of ocean biogeochemical and ecosystem parameters in particular to support reporting on marine environmental status in the regional European seas, as required by the Marine Strategy Framework Directive (MSFD); ii. The production of consistent ocean and wave information/products resulting from the coupling between the ocean state, the atmosphere, surface wave dynamics and other air-sea interface processes; iii. A better interface between the CMEMS and coastal monitoring services operated by Member States or private groups: this will require an improved processing of satellite and in situ observations in the coastal zone as well as the provision of suitable connectivity and boundary conditions between the Marine Service and the national coastal systems; iv. A better monitoring and description of the ocean state and its variability, requiring the preparation and release of annual Ocean State Reports describing the state of the global ocean and the European regional seas, in particular for supporting the Member States in their assessment obligation; v. A better assessment of the quality of CMEMS products, which requires continuous upgrade of observation infrastructures and improved data information and access services. Moreover, it is identified that one overarching driver for service evolution remains a continuous enhancement of modelling, data assimilation and forcing techniques at both the air-sea interface and lateral boundaries (coasts, open boundaries), in order to exploit the considerable investment in recent and upcoming observation technologies (e.g. SAR, SARin, SWOT, CFOSAT) and observation infrastructures. These priorities provide guidelines to the short, medium and longer term developments outlined in section 4 below. The proposed R&D roadmap was established following an analysis of R&D priorities by the CMEMS Scientific and Technical advisory committee (STAC) set up to assist Mercator-Ocean for the CMEMS implementation. This version (V3) released in June 2017 provides an update of R&D priorities and develops the Service Evolution roadmap and plans for the coming years. It takes into account, in particular, revised 3-year and 6-year evolution plans from the present CMEMS TACs and

9 6 6 MFCs, past and on-going H2020 Copernicus projects (service evolution, downstream) referred to in section 5, and outcomes of the first call for tenders for CMEMS Service Evolution (section 6). In the prospect of proposals to be submitted in response to the second CMEMS SE call (fall 2017), it is recommended to carefully consider the scope of on-going Tier-2 or Tier-3 projects in order not to duplicate research and developments efforts already addressed by other project teams.

10 7 7 4 R&D AREAS AND REQUIRED DEVELOPMENTS In this section, the 12 R&D areas required to support the service evolution guided by user needs are described. The required developments are categorized in terms of time horizons (short- to mid-term, and long-term) and expected objectives. The developments with shortto mid-term objectives refer to R&D activities that are expected to deliver significant results in less than 2 years, while longer-term activities are thought to need more than 2 years to bear results that will impact the service. The sections addressing R&D priorities of the 4 over-arching themes that emerged from the strategic analysis made by the STAC are organized as follows: ocean circulation modelling, mesoscale and other interactions, oceanwave and ocean-ice coupling (sections 4.1 to 4.4), biogeochemistry and ecosystems in the marine environment (section 4.5), interactions with the coastal ocean (section 4.6) and ocean-atmosphere coupling, reanalysis and indicators, and climate change (sections 4.7 and 4.8). Finally, sections 4.9 to 4.12 correspond to cross-cutting activities and methodologies that will benefit the 4 overarching themes. 4.1 CIRCULATION MODELS FOR THE GLOBAL OCEAN, REGIONAL AND SHELF SEAS Expected evolutions To support Copernicus marine users and decision-makers there will be an increasing demand for model information on fine spatial scales, higher frequencies and with a more complete representation of dynamical processes of the turbulent ocean. This is relevant to all areas from the open ocean where the dynamics are significantly impacted by mesoscale eddies, to transition areas connecting coastal and shelf seas and to critical regions (e.g. straits, boundary layers) requiring high topographic resolution. Numerical models will be needed with increased resolutions to represent the dynamics captured by existing and future Earth Observation platforms (e.g. high-resolution wide-swath altimetry, geostationary sensors, direct estimations of surface currents, high resolution sea surface temperature (SST), etc.) and consistent with improved estimates of fluxes of freshwater and suspended matter at the coast. Required developments with short- to mid-term objectives Development of advanced physical parameterizations, numerical schemes and alternative grids to improve the performance of ocean models resolving the mesoscale and smaller scales;

11 8 8 Adaptation of multi-scale and seamless downscaling/nesting capabilities to achieve kilometric to sub-kilometric resolution over specific areas; Better resolution of surface currents (0-20 m depth) and higher-frequency processes (e.g. tides) including their effects (e.g. rectification) and associated uncertainties on the circulation and on the transport of tracers; Adaptation of forcing methods with consistent physical parameterizations that benefit the fine scales (fronts, filaments) of surface ocean patterns; Improved capability in 3D storm surge forecasting for global and regional models, in particular new strategies to incorporate atmospheric surface pressure data, tidal forcing and coupled ocean-wave-atmosphere effects; New strategies and algorithms to solve the model equations efficiently on next generation computing systems: this should result in code performance improvements on most intensive HPC applications, which is crucial to sustain operational production; Development of empirical or other atmospheric modelling techniques to improve the spatial and temporal resolution of atmospheric forcing. Required developments with longer-term objectives Modelling issues related to multi-scale and multi-physics and in particular model nesting and unstructured grids; Development of non-hydrostatic ocean models to represent the full evolution of finescale oceanic structures in the three spatial dimensions; Understanding of fine scale processes to guide the development of global ocean analyses and forecasts at increased resolution and complexity (e.g. including representations of tidal physics, mixing and wave-current interactions). 4.2 SUB-MESOSCALE - MESOSCALE INTERACTIONS AND PROCESSES Expected evolutions Our knowledge on the relationships between the physical, chemical and biological processes in the upper ocean is continuously improving. This is essential for understanding and predicting how the ocean and the marine ecosystems respond to ocean dynamics and changes in atmospheric forcing. Advection, mixing and enhanced vertical velocities associated with mesoscale and sub-mesoscale oceanic features such as fronts, meanders, eddies and filaments are of fundamental importance for the exchanges of heat, fresh water and for biogeochemical tracers between the surface and the ocean interior, but also for exchanges between the open oceans and shelf seas, and between the pelagic ocean and the benthos. The challenges associated with mesoscale ( km, Rossby numbers < 1, and time scales from weeks to a few months) and sub-mesoscale variability (between 1-20 km, Rossby and Richardson numbers O(1), and time scales from minutes to a few days) require high-

12 9 9 resolution observations (both in situ and satellite) and multi-sensor approaches. Accordingly, multi-platform synoptic experiments have to be designed in areas characterized by intense density gradients and strong mesoscale activity to monitor and establish the vertical exchanges associated with mesoscale and sub-mesoscale structures and their contribution to upper-ocean interior exchanges. In situ systems, including ships, profilers and drifters should be coordinated with satellite data to provide a full description of the physical and biogeochemical variability and will be combined with resolution numerical simulations. Focus should be made on a range of scales ( km) traditionally not resolved by conventional altimeters but in which the wide swath altimeter SWOT will make an unprecedented contribution. At the same time, the realistic representation of the generation and evolution of such processes in numerical ocean models remains very challenging due to the chaotic nature of the oceanic flow, making direct model-data comparisons at the smallest scales problematic. Required developments with short- to mid-term objectives Predictability studies to assess the feasibility of developing open ocean and coastal/regional seas forecasts with lead-time of a few days to weeks, together with assessment of likely prediction skill; Methods to assess the relative impact of sub-mesoscales and internal waves on sea surface height and in situ observations; Methods to account for turbulence-submesoscale interactions into numerical models, topographic interactions at small scales, mixing induced by internal waves and other interacting processes between internal waves and submesoscale eddies; Development of new metrics adapted to the increased resolution in both observations/products and models. Required developments with longer-term objectives Synthesis from multidisciplinary field experiments and numerical studies to assess (i) systematic error in ocean models induced by sub-mesoscale dynamics, (ii) impact of vertical exchanges associated with mesoscale and sub-mesoscale structures, and (iii) operational products in specific areas; Process studies focused on physical-biological interactions at sub-mesoscale, implications for biogeochemistry, productivity, export, ecosystem structure and diversity; Improved understanding of vertical exchanges associated with oceanic mesoscale and sub-mesoscale features (e.g. fronts, meanders, eddies and filaments) through the combined use of in situ, satellite data and numerical models; Assess the impact of solving the ocean dynamics at kilometric scales on the role of ocean on climate (e.g. vertical exchange of heat and carbon, representation of overflows).

13 COUPLED OCEAN-MARINE WEATHER INFORMATION, SURFACE CURRENTS AND WAVES Expected evolutions The most important ocean weather parameters delivered by CMEMS are ocean currents, temperature (in particular in the near-surface layer of the ocean), sea ice, sea level, waves and surface winds. For several marine and coastal applications, information on surface parameters, including total and non-tidal currents and waves, is highly important. One example is the off-shore industry, where information on wave height and period are fundamental in the decision making (e.g. when ocean platforms need to be evacuated during violent storms). Everyday activities along the coasts heavily depend on wave information. Information on surface waves is also an essential input to the risk analysis in storm flood events. Waves are a bridge between the ocean and the atmosphere. Therefore, surface fluxes and winds used to drive the wave and ocean circulation should be consistent. The quality of wave forecasts and analyses is strongly linked with the quality of the wind forcing. Waves are also an important mixing agent with an active role in erosion and resuspension processes. In 2012, the EU funded project MyWave was established to lay the foundation for a future Service that also includes ocean waves. Additional efforts funded under the 1 st CMEMS SE call are underway to support the production of more consistent ocean-marine weather information, including information on surface waves. However, the two-way coupling between ocean and wave processes has to be further developed and improved. Required developments with short- to mid-term objectives Improved methods for analysing and predicting upper ocean currents; Improved processing methods to retrieve wave and surface currents (total and nontidal) information from the relevant satellite data, such as wave height based on altimeters (e.g. Sentinel 3) and wave information from SAR (e.g. Sentinel 1) and CFOSAT; Improved processing methods to retrieve wave information (including integrated parameters such as significant wave height) from in situ platforms (buoys, moorings, HF radars, etc); Developments to implement fully coupled ocean-wave models in which both components will exchange information at high frequency, and development of coupling parameterizations (bulk parameterizations for instance), both in the open ocean and near the shoreline; Investigations of wave model error dependencies on swell parameters (usually source of the largest errors) and improved parameterizations; Improved wave modelling near the ice edge and in ice-infested waters.

14 11 11 Required developments with longer-term objectives Improved modelling of extreme and rogue waves ; Implementation of advanced techniques to assimilate data into fully coupled oceanwave-atmosphere model systems; Improved characterisation of the uncertainties in coupled ocean-wave analyses and forecasts. 4.4 NEW GENERATION OF SEA-ICE MODELLING Expected evolutions There is evidence of a number of shortcomings in modelling the coupled ocean-sea-iceatmosphere system at high latitudes that continue to limit the quality of operational products delivered to users (both real-time as well as reanalyses). The 10 km to 100 km wide Marginal Ice Zone where ocean waves and sea ice processes are coupled is expected to widen in a warmer Arctic, but is still not well represented, neither in data nor in operational models. Additionally, products for fisheries (having a very strong economic value) depend on several qualities of both physical models (mixing, horizontal and vertical advection of larvae) and primary and secondary production models. Some shortcomings are inherent to the traditional viscous-plastic sea-ice rheology (and elastic-viscous-plastic as well) used in the current sea-ice models, deformations of sea ice being not correctly represented and in turn the velocity of sea ice drift being also incorrect. The ability to assimilate sea ice drift is also impaired by such model biases. In addition, the inclusion of more non-linear, chaotic processes are expected to become increasingly challenging for assimilation methods, in a context where more observations will be delivered from space, especially through the Sentinel suite. Required developments with short- to mid-term objectives New methods designed to assimilate satellite sea ice observations, such as the type and thickness of sea ice and more detailed information available from SAR imagery, together with impact assessments; Sea-ice models based on a more realistic rheology (for instance rheological models based on solid mechanics rather than fluid mechanics and their inclusion into operational models); Improved coupling between ocean waves and sea-ice processes ; Improved physical modelling of mixing below sea ice; Improved sea ice thermodynamics that will include effects of surface melt ponds; Coupling with snow models to account for ageing of snow and blowing snow; Development of a new generation of sub-kilometer scale dynamic sea ice models able to resolve ice floes.

15 12 12 Required developments with longer-term objectives Representation of biological cycles in sea ice, including optical properties of sea ice and vertical migration of nutrients in sea ice; Specific development for modelling of the Marginal Ice Zone, taking into account the heterogeneity in sea ice coverage and coupling effects with ocean and atmosphere and waves at fine scales; Developments on sea-ice models and data assimilation targeted to initialize coupled ocean-atmosphere and sea-ice forecasts for a wide range of time scales. 4.5 MODELLING AND DATA ASSIMILATION FOR MARINE ECOSYSTEMS AND BIOGEOCHEMISTRY Expected evolutions Users of the future CMEMS will benefit from the gradual improvement of models and tools to monitor the biogeochemical state of the ocean and marine ecosystems in ocean basins and marginal seas, using model-data integration and assimilation methodologies. This will require very significant strengthening of the observation system of the green ocean component at a range of scales, including in regional and coastal seas. In addition to the Sentinel suite, the potential of future satellite sensors (e.g. imagers from geostationary satellites) should induce a breakthrough in the monitoring of coastal areas and the landocean interface. Moreover, monitoring the concentration and distribution of pollutants will be essential for predicting ecosystem responses to pollution. Other challenges on the marine component of the carbon/greenhouse gases cycles will also require significant R&D effort as significant gaps in monitoring tools also exist in this area. During the past years, a suite of EU funded projects have developed the processing of remote-sensing data for open ocean and coastal waters (see section 5) as well as prototype ecological marine forecast systems for ocean basins and European seas (Atlantic, Baltic, Mediterranean and Black Seas) which include hydrodynamics, lower and higher trophic levels. More recent projects funded under the 1 st CMEMS Service Evolution R&D call will deliver further advances for multi-platform biological data assimilation, data assimilation of phytoplanctonic functional types and delivery of ecosystem enriched products. There is an expectation that the integrated Atlantic Ocean observing system will increase the number and quality of in situ observations on chemistry, biology and ecology over the next decade. A co-evolution of the data use in assessment and predictive models holds great potential for new products and users. It is required that results from these projects as well as similar advances in the field be transferred to CMEMS in order to consolidate the core ecosystem component of the service.

16 13 13 Required developments with short- to mid-term objectives Improved modelling and assimilation capabilities for the representation of ocean biogeochemistry and the marine food web from primary production to higher trophic levels (plankton to fish), including estimations of uncertainty; Relationships between optical properties and biomass for direct assimilation of optical properties into biogeochemical models; Improving CMEMS monitoring (i.e. products derived from satellite and in situ observations) and modelling products to better support aquaculture management applications; Improved ph and carbon monitoring and forecasting: carbonate system observations, modelling and assessment at regional and global scales; Methods to generate operational products in addition to standard (T, S, Chl, oxygen, ph, nutrients, light, plankton biomass) in support of predictive habitat forecasts, for ecological status and fisheries modelling and risk assessment (e.g. invasive species, harmful algal blooms, marine protected area planning); Multi-objective assimilation capabilities, e.g. combining state and parameter estimation, combining ocean colour and sub-surface data from BGC-Argo, gliders and other relevant ecological observations especially in regional seas; Demonstration of consistent interfacing (nesting, downscaling) between open ocean biogeochemical models and regional/coastal ecosystem models and downstream applications; Standardized validation methods for ecosystem model products/variables (particularly related to non-assimilated observations/variables). Required developments with longer-term objectives Improved description of benthic-pelagic coupling on short-term (seasonal) and longterm (decadal) scales; identification of the impact of initial conditions; Improved methodologies for supplying operational information on sources of nutrients and pollution/chemicals to the oceans; Develop a system for routine model estimates of the dose and direct and indirect effects of pollution on communities of algae, zooplankton and fish larvae. 4.6 SEAMLESS INTERACTIONS BETWEEN CMEMS AND COASTAL SYSTEMS The coastal monitoring services operated by Member States or private groups will form an important and strategic group of users of the CMEMS. These activities will enhance the socio-economic value of CMEMS by contributing to the MSFD, spatial planning and other downstream applications (offshore operations, coastal engineering, habitat monitoring, aquaculture, harmful algal bloom monitoring, adaptation and mitigation to climate change). Downscaling also includes the integration of coastal stations, and the role of CMEMS regional products is also to fill gaps between coastal observatories.

17 14 14 However, the one-way vision of a core service delivering information to downstream users without feedback to upstream providers has a number of limitations since the coastal strip should also be considered as an active boundary layer that influences the open ocean region connected to coastal areas. Scientifically, interactions between land, littoral, shelf, regional and abyssal seas are still a major unknown, poorly characterized and modelled. It is therefore needed to develop the CMEMS in such a way as to enable more efficient interfacing with a large variety of coastal systems describing the physical and biogeochemical coastal ocean states and ecosystems. Future operational circulation models implemented in the open ocean in CMEMS should enable more flexible coupling with a variety of model codes and regional configurations specifically customized for coastal dynamics and benefiting from user experience and practices, including those interfaced with near-shore, estuary models and hydrological models, unstructured grid models in areas that require very high resolution with good representation of topographic features. These issues have been explored in a number of recent EU-funded and national projects, relying on the coordinating role of EuroGOOS with the objective to consolidate, integrate and further develop existing European coastal and regional operational observing and forecasting systems into an integrated pan-european system targeted at detecting environmental and climate changes, predicting their evolution, producing timely and quality assured forecasts, and providing marine information services (including data, information products, knowledge and scientific advices). In the ongoing H2020 CEASELESS project for instance, it will be demonstrated how the new Sentinel measurements will support the development of a coastal dimension in Copernicus by providing an unprecedented level of resolution, accuracy and continuity with respect to present products. Furthermore, it also becomes timely to explore the possible connections between the CMEMS and the Copernicus Land monitoring service as coastal users will rely on information from both services. Therefore, an expert workshop was recently organized by EEA and Mercator Ocean with the objective to better address user needs in the coastal zone (see section 6.4). The required developments listed below are partly inspired from the main conclusions of this expert workshop. Required developments with short- to mid-term objectives Improved and standardised inputs of freshwater flows and associated river inputs of particulate and dissolved matter and homogenised river forcing approaches in global, regional and coastal models; Improve the interfaces/interactions between coastal monitoring and modelling systems and CMEMS; Comprehensive impact studies of CMEMS boundary conditions on coastal systems (physics, biology) and their applications (e.g. MSFD); Data assimilation and ensemble forecasting improvements to better serve the coastal applications;

18 15 15 Development of flexible interfaces between regional and coastal models, including near-shore and estuarine models and with the capability to connect structured and unstructured grids. Required developments with longer-term objectives Adoption of robust standards to ensure compatibility between CMEMS and downstream systems; Connection and coupling with land hydrology models. 4.7 COUPLED OCEAN-ATMOSPHERE MODELS WITH ASSIMILATIVE CAPABILITY Expected evolutions While coupled ocean-atmosphere models have been developed for a while in the context of climate research and weather forecasting, the MFC systems in CMEMS are currently based on a forcing mode approach (except for the GLO-CPL forecasting system) in which the ocean is driven by surface momentum, heat and freshwater fluxes computed using specified atmospheric information at fairly coarse space-time resolution. This could be an issue due to the different scales of variability in the atmosphere and ocean systems. In particular, the lack of feedback at fine scales between ocean and atmospheric models leads to inconsistencies and has implications for the practical forecasting capability of regional and coastal environments. Furthermore, the existence of observations at the ocean-atmosphere interface is an opportunity to introduce assimilation constraints into coupled models, especially for regional/coastal applications of interest to Copernicus users. Required developments with short- to mid-term objectives Advanced assimilation methods targeted to provide improved estimations of upper ocean properties consistent with sea-surface observations and air-sea fluxes; Representation of diurnal variability and cool skin layer in forced ocean and coupled ocean-atmosphere models; Consistent numerical schemes to improve the representation of ocean-waveatmosphere interactions for regional oceanic applications; Use of ocean wave spectra to determine contributions to boundary layer momentum fluxes. Required developments with longer-term objectives Novel ocean forcing approaches that may include simplified atmospheric boundary layer dynamics and ocean feedbacks at high spatial resolution; Analyses of impacts and feedbacks resulting from coupling between ocean, atmosphere and waves on the surface ocean dynamics and biology, including in the case of extreme events; Facilitate collaboration with the atmospheric community for exploitation of the surface observations common to the coupled ocean-sea-ice-atmosphere interface,

19 16 16 such as surface winds, surface currents, SST, waves, radiative, freshwater and heat flux components, sea-ice concentration, sea-ice thickness and sea-ice temperature; Multivariate data assimilation into fully coupled ocean-wave-atmosphere dynamical systems. 4.8 OCEAN CLIMATE PRODUCTS, INDICATORS AND SCENARIOS Several CMEMS systems produce analyses and reanalyses or multi-year reprocessed observations that describe ocean variability and changes. Ocean Monitoring Indicators (OMI) can be derived from these products for tracking ocean mechanisms and/or climate modes and trends, much as we use global greenhouse gas (GHG) concentrations to follow anthropogenic effects, and globally averaged SST to monitor warming. CMEMS is developing OMIs for monitoring the ocean state at global and European scales (see the CMEMS Multi- Year Product Strategy Plan, section 6.1). Many users are also interested in future trends and evolution of the marine environment under different climate scenarios. Earth system and climate model predictions (seasonal to decadal) and longer-term projections based on GHG scenarios, together with comprehensive ocean reanalyses, can be used to infer possible changes in the ocean state at regional and coastal levels, including biogeochemical and other aspects of the marine environment (e.g., related to harmful algal blooms and other marine extreme events). In the field of ecology, our scientific understanding of ecosystems has matured to a point that we may be able to produce potential scenarios based on our knowledge of potential future change in the physical system (e.g. water quality, coral bleaching). Such outcomes require the production of improved ocean reanalyses and multi-year reprocessed observations suitable for climate change detection, with reduced systematic errors and improved dynamical and physical consistency and realism. Reanalyses product quality delivered by the CMEMS will have to be monitored on a continuous basis, using proven, verifiable and robust methodologies that can be agreed upon with external partners (see section 6.1). This activity can rely on standard metrics (e.g. as defined in the international GODAE Ocean View framework) or on new approaches. These activities will contribute to the Ocean State Reports delivered by CMEMS, and will also benefit from the development of the Copernicus Climate Change Service (C3S) line. Required developments with short- to mid-term objectives Development of methods for inferring the future state (from seasons to decades) of the marine environment (physics and biogeochemistry) at regional and coastal scales, as well as changes in ecosystems, based on climate model predictions and projections (and associated tests for quality and reliability); Development of Ocean Monitoring Indicators based on CMEMS and other products; specific needs include indicators (i) for the physical ocean state, variability and

20 17 17 change monitoring (ii) for the health of the ocean (e.g. acidification, deoxygenation, eutrophication), (iii) for fishery and aquaculture management and (iv) for other applications such as maritime transport, marine renewable energy, coastal zone management; Framework for coordination / intercomparisons with international community efforts (EuroGOOS, GODAE OceanView, CLIVAR, CMIP activities ); Methods to ensure quality, homogeneity and robust uncertainty measures in longterm time-series reconstructed from data or model reanalyses. Required developments with longer-term objectives Development of methods for inferring the future state (from seasons to decades) of the marine environment (physics and biogeochemistry) at regional and coastal scales, as well as changes in ecosystems, based on climate model predictions and projections (and associated tests for quality and reliability) [continued]. 4.9 OBSERVATION TECHNOLOGIES AND METHODOLOGIES Expected evolution The observing system, which today provides regular and systematic data on the physical state and dynamics of the ocean and marine ecosystems, is expected to evolve in several directions during the next years: The global in situ component, through coordinated programmes such as Argo and its extensions (deep ocean, polar seas and BGC), Atlantos in the Atlantic and TPOS2020 in the Pacific Ocean, is expected to more systematically collect data on new chemical and biogeochemical state variables (oxygen, bio-optical properties, nutrients, ph) in addition to conventional temperature and salinity profiles, and to increase the sampling of the deep (below 2000 m) and ice-covered seas; The European regional and coastal seas will be monitored through more coordinated, multi-sensors and multi-parameter networks including HF radars, cabled structures, fixed platforms, ferry boxes, surface drifters, gliders, and other new technologies, e.g. undertaken under EuroGOOS, EMODnet, the EOOS (European Ocean Observing System) initiative and other projects (e.g. JERICO-Next, AtlantOS); The space component will rely on a mixture of operational (e.g. the Sentinel suite) and exploratory missions, providing sea-level (both along-track and wide-swath), geoid information, SST, ocean colour, surface salinity, surface wind and currents, wave properties and sea-ice parameters; New capabilities are expanded to sample a variety of scales through multi-platform observing systems that include more and more autonomous platforms with multidisciplinary sensors; Observations delivered through Copernicus or in cooperation with other initiatives will increase, especially those dedicated to validation measurements following

21 18 18 agreed protocols (e.g. ESA protocols, quality assurance framework for Earth observation guidance); Ocean observing systems will be multi-purpose and have multiple uses, observing the ocean at multiple scales and from the coastal to the open ocean, delivering quality controlled data to science and society. New multi-platform and multi-disciplinary approaches combining both in situ (e.g. gliders, Argo, ships, drifters, fixed platforms) and satellite observations at high resolution will be needed to resolve a wide range of temporal and spatial scales, to monitor key regions or processes for the ocean state and its variability and to fill gaps in our knowledge connecting physical processes to ecosystem response (e.g. BGC Argo, TPOS2020, Atlantos, GOOS). In addition, there is a huge potential to more efficiently access/use data from different industrial platforms, including off shore power stations. The core data set assimilated in CMEMS real-time models or multi-year reanalysis systems include SST, sea-level anomalies and T/S profiles (almost systematically), sea-ice concentration (whenever possible), and chlorophyll concentration (occasionally). In addition to consolidating the assimilation of such data, a broader list of parameters should be integrated into monitoring and forecasting systems within the next 3-10 years. Required developments with short- to mid-term objectives Improved processing of Sentinel 1, 2, 3 data for coastal ocean products and applications; Adaptation of existing quality-control methods to new observing system components, both in situ (for instance new biogeochemical parameters) and satellite, with a focus on Sentinel observations and future exploratory missions (e.g. CFOSAT, SWOT); advancement and adoption of international calibration and quality-control procedures; Developments of advanced protocols for real-time quality checking and automated flagging procedures that are more consistent with nowcast/forecast information delivered by MFCs; More consistent processing and assembly of data from different, heterogeneous observation platforms and sensors for estimating derived quantities (e.g. sea-ice products, surface currents, mixed-layer depth, primary production, nutrients from temperature, salinity and O 2 vertical profiles etc.); Incremental development of assembly centres to include more observations relevant to coastal waters but also of interest to the monitoring of regional seas. Required developments with longer-term objectives Exploitation of new technologies and platforms to more systematically observe ph and pco 2 in ocean basins ; Adaptation of assembly centres procedures to take advantage of new sensors, communication technologies and unification of procedures, protocols, access and download systems across assembly centres.

22 OBSERVING SYSTEMS: IMPACT STUDIES AND OPTIMAL DESIGN Expected evolutions Regarding future observing systems (from space or in situ), the best possible basis is required in the design and exploitation of observation networks and satellite constellations. This goal can be ideally achieved through impact and design studies (Observing System Experiments OSEs; Observing System Simulation Experiments - OSSEs) and dedicated process experiments and pilots. OSEs and OSSEs represent a useful investment when expanding an existing observing system, defining a new one, or preparing the assimilation of new data types, or data with improved resolution/accuracy. They are also required for design studies to re-assess the sampling of the present in situ networks, define the required extensions, and prepare the assimilation of new in situ data types considering the synergies with satellite observations. Such approaches enable a consistent approach to the assessment of the impact of observation data, assimilation techniques and analysis/forecast systems, the definition and operation of space missions (upstream and downstream) and in situ networks. The existing TAC and MFC capabilities provide an excellent opportunity to develop a new functionality within the CMEMS, and conduct impact studies and observing system design based on rigorous and objective methodologies. This functionality will consolidate the link between CMEMS and the data providers by formalizing recommendations on future observing systems. In return, improving the design of future observing systems will be of great benefit to CMEMS itself given the likely impact on the products quality. Required developments with short- to mid-term objectives Adaptation of tools and diagnostic methods such as Degrees of Freedom Signal (DFS) and Forecast System Observation Impact (FSOI) to measure the impact of observations into ocean monitoring and forecasting systems; Adaptation of assimilation interfaces (e.g., observation operators) to new satellite sensors: wide-swath altimetry, ocean colour, SST from Sentinels, surface currents ; Development of more robust methodologies (i.e. to ensure results as independent as possible from model and error assumptions) and automatic tools to conduct impact studies (including the OSE class); Impact studies of new observation data types or products for ocean analyses, forecasts and reanalyses (e.g. SSS from space, Sea Ice thickness from Cryosat and SMOS, surface current data sets, BGC-Argo, HF Radars, etc.); Network studies that will provide guidance to deployments: this includes identifying critical ocean observation points and regions where the benefit to forecast skill and ocean analyses and reanalyses can be maximized. Required developments with longer-term objectives Design studies to re-assess the sampling of the present-day in situ T/S network (Argo including BGC-Argo, ships, buoys, moorings);

23 20 20 Guidance to the in situ observation communities on how to optimize observing strategies and the synergies with Sentinel and future wide-swath altimetry missions (SWOT), complementing the ongoing AtlantOS effort ADVANCED ASSIMILATION FOR LARGE-DIMENSIONAL SYSTEMS Expected evolutions The operational suites implemented in the CMEMS MFCs essentially rely on traditional assimilation methods inherited from the Kalman filter or the 3DVAR approach including simplifications and adaptations to make them tractable with large-dimensional systems. Today, only the Arctic MFC has explicitly implemented the EnKF to propagate the spread of an ensemble of likely states between successive updating steps, though several other MFCs intend to implement ensemble approaches in the future (Global, Mediterranean, Baltic MFCs, Figure 1). The current trend in oceanic and atmospheric data assimilation is a move toward probabilistic methods inherited from the Bayesian estimation framework, as reflected in a variety of activities conducted in recent projects (e.g. FP7 SANGOMA), or reported in workshops (e.g., GODAE Ocean View task teams) and conferences (WMO symposium on Data Assimilation). At CMEMS level, it becomes urgent to take advantage of these developments in order to (i) further develop the assimilation methods to enable implementation into systems involving different components (coupling with atmosphere, biology, biogeochemistry, cryosphere, etc.), (ii) improve the coupling between assimilative systems with different target resolutions and different assimilation parameterizations, (iii) implement an effective production of probabilistic information as needed by users to support decision-making, and (iv) improve the physical consistency of ocean state estimations, ensuring that the smallscale dynamics simulated by ocean models is in balance with the larger scales captured by the observing systems. In the NEMO framework (most MFCs systems are based on NEMO), a new assimilation component has been implemented in the reference version that includes several modules (observation operators, linear-tangent and adjoint of the circulation model, incremental updating schemes) to facilitate coupling with various assimilation systems. It is also planned to include an ensemble simulation mode in future versions of NEMO. Other efforts on modular software development have also been initiated in European institutions, such as the Object-Oriented Prediction System (OOPs) project at ECMWF. A convergence between these parallel efforts is expected in order to ensure an effective implementation of these developments into the large-scale systems based on NEMO. Required developments with short- to mid-term objectives Metrics for ocean analysis/reanalysis and forecasting produced using ensemble techniques;

24 21 21 Development and application of advanced assimilation methods, including to identify and understand model bias and forcing errors, with a focus on ocean reanalyses; Adaptations of ensemble-based assimilation/forecasting systems, and development of new techniques to increase the range and amount of observations assimilated into MFCs (e.g. ocean currents, ice thickness, ocean colour), making best use of Copernicus Sentinel missions; Common developments on assimilation modules to connect models and assimilation kernels shared between different MFCs; Extension of conventional assimilation schemes to include image data information (SST, Ocean Colour) in addition to pixel data; Efficient methods to account for complex observation and modelling error structures (incl. non-gaussian errors) in assimilation algorithms; Adaptation of analysis schemes to process different spatial and temporal scales during the analysis stage; Adaptation of analysis schemes to preserve the consistency of non-observed quantities such as vorticity, diffusion, vertical velocity, passive tracers etc.; Verification methods and intercomparison protocols (rank histograms, Continuous Rank Probability Score, ) suitable to assess the reliability of ensembles in probabilistic assimilation systems; Inverse methods that will incorporate simplified balance relationships (e.g. geostrophic and vorticity balance, omega equation) to ensure physically consistent products. Required developments with longer-term objectives Development of data assimilation tools for coupled models (ocean-atmosphere including waves, physics-biogeochemistry, ocean-shelf models), including consistent coupled initialization approaches and flux correction schemes at interfaces; Developments of software infrastructure that can accommodate different assimilation methods and facilitate the sharing of algorithms and optimization of computer codes (models, assimilation schemes) on HPC HIGH-LEVEL DATA PRODUCTS AND BIG DATA PROCESSING Expected evolutions CMEMS needs to enrich its offer with improved accessibility and quality controlled data, in agreement with scientific community standards. Users are also requesting advanced data products that should rely on more complex processing of measurements of essential ocean variables at higher resolution, including the provision of composite products and their uncertainty estimates. Those data products will feed monitoring and forecasting systems or validate model outputs. The ESA Climate Change Initiative and the Copernicus Climate

25 22 22 Change Service (C3S), for instance, will deliver advanced products that will benefit the CMEMS reanalyses production. The methodologies in place today for merging, filtering, compressing and interpolating observations (until now mostly based on statistical interpolation methods) will need to be adapted to face the challenges of big data streams and observing systems that are heterogeneous in terms of space-time sampling, resolution and accuracy. An additional issue to address is how to take advantage of the high quality delayed mode data in order to get improved reanalyses of the ocean state for assessing changes and updating climatologies. Required developments with short- to mid-term objectives New high resolution ocean colour products from Sentinel 2 and synergy with Sentinel 3 OLCI products; Development of advanced data products merging different type of observations through multivariate analysis (e.g. for chlorophyll, surface currents, ice drift and ice thickness); Reprocessing of existing data sets with more advanced quality control methods, to continuously improve the ocean reanalysis effort including for detection of change; Specific processing and mapping of data product uncertainties for both direct use, validation purposes, and data assimilation; Discrimination methods between steric and non-steric components in observed sea level anomaly signals, both for data processing and assimilation into models; Preparation of composite data products and derivation of quantities that can be used to infer transport from tracer information, or vice versa; New bathymetric data with very high resolution, and possibly information about changes in ocean bottom, including impact assessment on ocean analyses and forecasts. Required developments with longer-term objectives Data mining and image processing techniques needed to facilitate the automatic extraction and analysis of patterns from big data sets (produced by observing and/or modelling systems) and to better respond to user s needs.

26 THE SERVICE EVOLUTION ROADMAP FOR 2021 AND BEYOND In this section, the outline of the service framework anticipated in 2021 is described together with the completed and ongoing Tier-3 R&D projects to be taken into account for the definition of priorities in future calls, and the main scheduling of service evolution projects until the end of the CMEMS implementation period. The CMEMS technical annex provides the framework for the implementation of the CMEMS in terms of functionalities and operational tasks. The 12 R&D areas developed in the previous section should be covered in such a way as to ensure that CMEMS can evolve towards the targeted service lines foreseen for 2021 and after. While the 12 R&D areas are all relevant and essential at present to reach the objectives initially defined for the CMEMS, priorities are re-adjusted periodically according to updates of user requirements, the status of scientific developments achieved within and outside the CMEMS community, and to the high level CMEMS evolution strategy. 5.1 THE TARGETED CMEMS SERVICE IN AND AFTER 2021 The CMEMS is currently designed to deliver products to users in four main domains of application: i. marine safety, which requires information at very high resolution in focused areas all over the globe to support marine operations, marine weather forecasting, sea ice forecasting, combating oil spill, informing ship routing and search and rescue operations, and all activities requesting offshore operations; ii. marine resources, with applications to the sustainable management of living marine resources, through fisheries and aquaculture; iii. coastal and marine environment, with applications that require good knowledge of environmental status in coastal areas in terms of physical, chemical and biological components and their variability; iv. weather, seasonal forecasting and climate, with needs for accurate information near the surface and sub-surface ocean, on a daily or shorter time basis, in real time and delayed mode. With respect to the existing CMEMS offer and based on existing and future user needs, the following evolutions of the service in terms of technical and scientific content of its products are proposed:

27 24 24 Staying a world-leading reference source of information for operational players in the marine domain through improved product consistency (especially of reprocessed time series and reanalyses), quality accuracy and assessment, multi-data approach with scientific inter-calibration and assessment; Improving the physical information content of CMEMS products by increasing resolution in space and time of models and observations, adding the representation of physical processes, going towards more coupled systems (ocean circulation waves sea ice atmosphere biogeochemistry ecosystems) to produce consistent estimates of the full ocean state; Improving assimilation schemes, in particular for biogeochemical data, taking advantage of the considerable investment in observation infrastructure and increase of data flow, especially through the Sentinel constellation that will be in place in 2021 in tandem with the in situ component; these techniques should also take into account new HPC architectures and computing facilities available in 2021; Meeting the marine biology demand and reach the level of excellence CMEMS has now for the marine physics. This also includes providing more information on biogeochemistry and higher trophic levels (from plankton to fish) to better support fisheries management, living resources protection, sustainable exploitation of marine biomass; Enhancing ocean climate and ocean health monitoring (physics and biogeochemistry including O 2 and ph), supported by the release of annual Ocean State Reports 2, informed by annual analyses of the state of the global ocean and the European regional seas, in particular for supporting the Member States in their assessment obligation; Assessing past and future climate change impacts on the ocean environment: transform the high level CMEMS expertise on the ocean physical, biogeochemical states, ocean health and ecosystems, into a strong assessment capacity on the ocean climate, propose assessment capacity and what-if scenarios, including projections of the ocean state (physical, biogeochemical, ecosystems) at regional scales from weeks to decades for climate change scenarios; Supporting coastal area monitoring by Member States through the provision of suitable interfaces and boundary conditions by CMEMS for efficient EU leverage for supporting coastal environment knowledge and economy. Co-production and operation between the CMEMS and the national coastal systems should be encouraged, with adaptations at the level of coastal seas benefiting from user experience and practices (see also section 6.4). 2 The first issue of the Ocean State Report is available here (doi: / X ). See also section 6.1 for the CMEMS Multi-Year Product Strategy Plan.

28 25 25 These objectives will have to be covered by CMEMS Tier-2 projects (see below) and 3-6 years R&D plans of the TACs/MFCs. The CMEMS technical annex identifies additional service lines more specifically motivated by Member States requirements. The CMEMS service is operated today in an overarching framework that includes other Copernicus services, such as the Atmosphere service (CAMS, ECMWF), Land service (CLMS, EEA, JRC), Emergency service (EMS, JRC), Security service (EMSA, FRONTEX, SATCEN) and Climate Change service (C3S, ECMWF), as well as a community of new users consolidated through dedicated user uptake calls. This landscape will smoothly modify the profile of capabilities and users and service needs, including new users at the intersection between different Copernicus services. However at present, existing R&D projects and service priorities largely align with the Marine Service overarching objectives. 5.2 TIER-3 R&D MARINE PROJECTS OF INTEREST TO CMEMS This section provides a status of Tier-3 (i.e., long-term objectives) R&D projects that are identified in 2017 as relevant for the CMEMS service evolution. Completed projects A number of projects, including those implemented under the Space theme in 2012, have been completed in 2014/2015: E-AIMS: improved European contribution to the international Argo observing system in support to scientific and operational challenges for in situ monitoring of the world ocean. MyWave: preparation of the scientific grounds for a wave component of the marine service. OSS2015: consolidation of biogeochemical products, information and services based on models, in situ and satellite observations. OPEC: improvement of operational services for biochemistry and ecological parameters at time scales from days to decades, with a focus on ecological status of coastal and shelf seas. SANGOMA: preparation of the new generation of stochastic data assimilation methods for ocean physical and biological model applications. Ongoing and newly decided EU projects HIGHROC: FP7 project ( ) to conduct R&D necessary for the next generation of coastal water products and services from ocean colour space-borne data by giving an order of magnitude improvement in both temporal and spatial resolution; JERICO-NEXT: H2020 project ( ) which aims at providing a solid and transparent European network for operational services delivering high quality environmental data and information products related to marine environment in European coastal seas;

29 26 26 AtlantOS: H2020 project ( ) which aims at optimising and enhancing the integrated observing systems in the Atlantic Ocean (northern and southern basins), with a focus on in situ platforms and considering European as well as non-european projects (TPOS in the tropical Pacific, and SOOS in the southern ocean) and partners. IntarOS: a new H2020 project ( ) to build up an integrated and unified holistic monitoring system for the different parts of the Arctic, including tools for integration of data from atmosphere, ocean, cryosphere and terrestrial sciences. ODYSSEA: H2020 project (~ ) downstream of CMEMS to develop, operate and demonstrate an interoperable and cost-effective platform that fully integrates networks of observing and forecasting systems across the Mediterranean basin, addressing both the open sea and the coastal zone. SPICES: H2020 project ( ) which aims at developing new methods to retrieve sea ice parameters from existing satellite sensors to provide enhanced products for polar operators and prediction systems. CEASELESS: H2020 project ( ) that will demonstrate how the new Sentinel measurements can support the development of a coastal dimension in Copernicus. CoReSyf: H2020 project ( ) that aims at supporting the development of coastal research through satellite data by creating an online platform delivering Earth observation data for coastal water monitoring. A variety of new projects supporting mainly the development of downstream services are also of potential interest to the evolution of the CMEMS (e.g., AQUA-USERS, SAFI, SWARP, BASE-platform, EO4life, EONav, Copernicus App Lab, CYANOLAKES, EOMORES, SPACE-O and MARINE-O EO-1 &2 projects) and should be monitored to some extent by CMEMS experts. R&D gaps for future H2020 calls The guidance document published by the EC in 2015 ( Research needs of Copernicus operational services ) identified 4 key actions needed to implement the CMEMS over the period of , taking into account evolving user needs and technological developments. These 4 key actions (that were based on the first version of this roadmap) are associated to the following scientific and technical challenges: 1. Solving ocean dynamics at kilometric resolution, 2. Designing future observing systems and related assimilation methods, 3. Developing seamless information chains linking dynamics, biogeochemistry and ecosystem essential variables, 4. Seamless interactions between CMEMS and coastal monitoring systems. The table below shows that a number of H2020 projects recently implemented will address some of these challenges (mainly action 2 and 4), while other actions [(1) and (3)] are still needed to meet the objectives.

30 Solving ocean dynamics at km resolution 2. Designing future observing systems and related assimilation methods 3. Developing seamless chains linking dynamics, biogeochemistry and ecosystem 4. Seamless interactions between CMEMS and coastal monitoring systems AtlantOS, Jerico-NEXT, IntarOS, SPICES, CEASELESS CEASELESS 5.3 TIER-2 R&D MARINE PROJECTS SELECTED BY CMEMS This section provides an overview of the Tier-2 (i.e., medium-term objectives) R&D projects that were selected for the CMEMS Service Evolution call 21-SE-CALL1 released in autumn An outline of the 12 selected project including project participants and abstracts can be found online and in section 6.2. The 2015 CMEMS R&D Service Evolution call included 5 lots: Ocean circulation, ocean-wave and ocean-ice coupling (3 projects selected): proposals funded under this theme are mainly addressing sub-mesoscale processes (R&D area 4.2), ocean-wave (R&D area 4.3) and ocean-sea ice coupling (R&D area 4.4). Marine ecosystems and biogeochemistry (3 projects selected): the funded projects cover several R&D objectives identified under area 4.5, specifically modelling of higher trophic levels in ecosystems and data assimilation capabilities. Links with coastal environment (2 projects selected): one project addresses upscaling issues under R&D area 4.6, while the other focuses on regional data assimilation and ensemble forecasting; Ocean-Atmosphere coupling and climate (1 project selected): the project focuses on the momentum exchange across the ocean-atmosphere interface in regional ocean prediction systems and the value of providing CMEMS products from ocean-wave systems that are fully coupled with an atmospheric component relative to those running in forced mode (R&D area 4.7); Cross-cutting developments on observation, assimilation, and product quality improvement (3 project selected): proposals funded under this theme address (i) satellite SST assimilation with a proper account of the diurnal cycle, (ii) the integration of existing European high-frequency radar operational systems into the CMEMS, and (iii) synergistic use of satellite and in situ data to infer the state of the upper ocean and associated flux of tracers and energy at fine scale. A preliminary review of the projects took place in early 2017 to identify if all projects were on track and to assess in advance the range of results of potential interest for the Tier-1 work plan. For the next call for tenders to be released in August 2017, priorities should be targeted towards the R&D areas and objectives not covered so far or complementary to the

31 28 28 on-going R&D projects; they should also take into account results and recommendations from on-going R&D projects. 5.4 SERVICE EVOLUTION ROADMAP AND MILESTONES UNTIL 2021 The CMEMS R&D roadmap until 2021 will be implemented as illustrated below. In addition to programmatic aspects, a number of milestones relevant to the service evolution R&D will complement the roadmap, e.g. the Copernicus Marine Week to be held in Brussels (25-29 September 2017). At Tier-2 level, the 12 CMEMS R&D projects selected in early 2016 (from the fall SE- CALL1) will terminate at the end of Phase 1, so as to provide guidance to the Tier-1 R&D work plans that will have to be developed in the TACs/MFCs tenders covering phase 2. The second Tier-2 CMEMS call for tenders (66-SE-CALL2), to be released in August 2017, will mainly impact the CMEMS after 2021: this will set a clear programmatic constraint on the definition of the long-term CMEMS strategy for the post 2021 period. At Tier-3 level, ongoing H2020 projects are expected to deliver results of interest to CMEMS, but not before The newly EU-funded projects (e.g. under the Space EO ) are expected to provide elements for the future CMEMS period starting in 2021.