Description of the problem addressed by the scientific research project proposed herein
Background and motivation
The role of mesoscale and synoptic motions are quite different in the ocean and in the atmosphere. If one looks on arbitrary selected snapshot of the global atmosphere (Figure 1a), the general structure of the mean large scale circulation will be masked by synoptic motions largely dominating over time averaged large-scale circulation patterns, especially in the extratropics. At the same time, in the ocean large scale mean circulation patterns will be well detectable even at eddy-resolving snapshot (Figrue 1b). This reflects the matter of fact of very different characteristic length scales of the mesoscale motions in the ocean and atmosphere. In this respect ocean mesoscale variability can not be described by a dynamical analogy with the synoptic variability in the atmosphere in terms of “weather systems” and their regimes. Remarkably, in the atmosphere the totality of the meridional heat transport (MHT), especially in the extratropics is done the synoptic transient features, while in the ocean MHT is largely dominated by the contribution from poleward mean currents flowing along continents (Figures 1 c,d). This is particularly the reason why resolving the synoptic scales in ocean models for quite a time has not been a strong case as it was for atmospheric models. This highlights very different role of synoptic and mesoscale eddies in the ocean and atmosphere despite the dynamical similarities of their generating processes.
One of the most important and intriguing ocean phenomena originating from internal variability modes is the so-called Atlantic Multidecadal Variability (AMV), represented by coherent changes in sea surface temperature (SST) over the whole of the North Atlantic that are most pronounced in the extratropics with a period of 50–70 years (Latif et al. 2004, 2006, Knight et al. 2005, Gulev et al. 2013, McCarthy et al. 2015). In many climate models, AMV results from variations in the Atlantic Meridional Overturning circulation (AMOC) that are generated internally by the coupled ocean–atmosphere system itself. This strong natural variability is superimposed on the long-term global-warming trend and also causes considerable spatial variation of that trend. There is evidence from data and climate models (Sutton and Hudson 2005, Kushnir et al. 2002, Gulev et al. 2013) that AMV forces coordinated variations in European and North American climate through the advection of heat and moisture released from the North Atlantic. We therefore need to understand this variability to discriminate anthropogenic effects from natural climate forcing (Gulev and Latif 2015). Figure 2 schematically shows a general concept of climate predictability and the role of the ocean in climate predictions. On time scales of several days atmospheric predictability is considered to be an initial value problem with ocean playing a minor role. On longer scales climate predictability starts to be more dependent on the externally driven conditions (e.g. greenhouse gas concentrations) and the ocean signals. In this respect time scales from several years to a few decades are characterized by the poorest predictability, as here ocean signals (like AMV) may dominate over secular trends. Thus, climate predictability at these scales largely depends on the extent to that we can describe the mechanisms and quantify ocean variability.
These arguments underscore the need for holistic, hierarchical approach to this problem in a format of consolidated effort, involving the advances of theoretical geophysical fluid dynamics, numerical experimentation with both realistic and idealized, dynamical and empirical models, as well as experimental diagnostics. This is what we propose to undertake in our project which will, for the first time, target to merge these approaches together by utilizing the complementary expertise of theoreticians, numerical modelers and dedicated practitioners, and synergizing the potentials of all methodological avenues to the critically important problem — the role of synoptic and mesoscale processes in the variability of large-scale oceanic circulation and its impact on climate.
Problem areas: closing gaps in understanding of ocean and climate dynamics
Role of oceanic eddies in large-scale climate dynamics. There is growing evidence for the importance of mesoscale eddies (motions on spatial scales of tens to hundreds of kilometers) in the oceanic and climate dynamics. In particular, eddies can redistribute heat anomalies in the mixed layer and modulate air-sea heat fluxes (e.g. Greatbatch et al. 2007; Schucksburgh et al. 2011; Hausmann and Czaja 2012). Air-sea interactions over the oceanic fronts and eddies demonstrate that the oceans leads the atmosphere on these scales (e.g. Small et al. 2008). These effects cannot be captured by numerical models that do not resolve the mesoscale, which is illustrated by large differences in air-sea interactions between numerical simulations with and without explicit eddies (Kirtman et al. 2012). The major challenge for the oceanic and climate science communities is to examine these effects of eddies on air-sea interactions, and take steps to improve representations of these processes in climate models. This task is equally important for the IPCC-class non eddy-resolving models and emerging eddy-resolving simulations.
Figure 3 demonstrates the magnitudes of the total and intrinsic AMOC variability in the 3 long-term experiments with the NEMO model in 3 resolutions performed at LGGE/MEOM in Grenoble (Gregorio et al. 2015). Clearly eddy permitting and eddy resolving configurations capture a considerable fraction of the interannual variability in AMOC (up to 50%) while coarse resolution model contributes practically nothing to the intrinsic variability. This example shows critical importance of the accurate simulation of intrinsic variability in the models that is possible only in high resolution model set-up.
One more problem of quantifying air-sea exchanges in eddy resolving simulations is associated with the horizontal and temporal resolution of the forcing atmospheric fields used for numerical experimentation with high resolution ocean models. In most current long-term experiments with ocean GCMs (e.g. Barnier et al. 2006, Penduff et al. 2011, Danabasoglu et al. 2014), even when the ocean GCM configuration is set to high eddy resolving resolution (e.g. 1/12 degree or higher) the forcing function is typically just interpolated to such a resolution from the original resolution of reanalyses (typically 0.5-1 degree) at which it has been originally developed (e.g. Large and Yaeger 2009, Brodeau et al. 2010). To which extent that high resolution surface forcing may change the model solution, especially with respect to the dynamics of the convection in subpolar latitudes, representation of the western boundary currents and dynamics of the other key regions is unknown, since large-scale experiments with high resolution forcing were not performed yet. Statistical diagnostics of surface fluxes (e.g. Gulev and Belyaev 2012) clearly demonstrate that at small spatial and temporal scales surface fluxes may amount to very high extreme values of several thousands W/m2 due to mesoscale structures in the atmosphere (e.g. frontal jets, convective cells, etc.) which are not resolved in global reanalyses. Furthermore, interaction between surface eddy-resolving ocean and high resolution fluxes may result in climate feedbacks originating from small scales (Ma et al. 2015a, b). These effects cannot be properly treated unless eddy-resolving simulations are performed in conjunction with high resolution surface forcing conditions.
Regional effects of mesoscale and sub-mesoscale eddies in the marginal and semi-enclosed seas. In many semi-enclosed seas sub-mesoscale eddies (SE) are characterized by the Rossby number and horizontal scale smaller than the internal Rossby radius of deformation. Such small-scale vortices and associated structures have been shown to be important contributors to transferring momentum, heat, and oceanic tracers through vertical pumping and horizontal propagation (e.g., Kim, 2010). They usually persist for 1-7 days and migrate with a translation speed advected by background currents, exhibiting distinct signatures in thermohaline fields at the surface, as well as in concentrations of principal terrigenic tracers (Fig. 4). Several mechanisms have been hypothesized for SE formation, including shear instability, shedding from the mean flow at coastline inhomogeneities (Zatsepin et al., 2011, Elkin and Zatsepin, 2014), generation by spatially non-uniform wind events (Klein and Lapeyre, 2009), as well as shedding from wind-forced river plumes (e.g., Zavialov et al., 2014).
Climate predictability issues. Currently our understanding of climate predictability is primarily based on the experiments with global and regional climate models in which the atmosphere is fully eddy resolving and the ocean is not. Of 42 models which participated in CMIP5 (IPCC AR5, Flato et al. 2013) about 2/3 had spatial resolution of approximately 1 degree with the rest being coarser resolution models. This situation is not going to be seriously changed in CMIP6 in which likely very few centers will use eddy permitting resolution (1/2 to 1/4 degree) for the ocean model configurations. The described drawback also holds for the Russian national INM climate model (Volodin et al. 2010, 2013, Gusev and Diansky 2014). Higher resolution (1/10 degree) eddy resolving configuration developed by Ibrayev et al. (2012) is still used for the relatively short spin-up experiments. However, in this resolutions ocean mesoscale processes will be relatively realistically represented only in low latitudes with highly variable processes in mid and high latitudes being still unresolved. This may result event in a poorer performance of these configurations compared to the coarse resolution set-ups as it was demonstrated for instance for water mass transformation characteristics (Gulev et al. 2007). Thus, it is critically important at least try to investigate the effect of high resolution ocean on climate variability in climate models. Given the resources constrains this can be potentially done for some specific cases using regional model set-up (see section 3). In this regard, we are yet unable to investigate the most interesting part of the ocean resolution dependence based on the IPCC/CMIP experiments.
As we have specified above, mesoscale and synoptic variability in the ocean may generate intrinsic ocean variability signals at different time scales and these signals might become further responsible for the natural climate variability modes superimposed with long-term trends. In this respect the mechanisms responsible for AMV and associated AMOC variability on multidecadal time scales (Bjerknes conjecture) is a very question to be addressed through the analysis of the role of ocean eddies in large scale variability. However for this we need to close the gap between the results of climate models with coarse resolution ocean (e.g. Delworth et al. 2000, Latif et al. 2006 among many others) and the results of the decoupled experiments with ocean-only GCMs in very high resolution (Gregorio et al. 2015, Penduff et al. 2015).
What can we potentially expect from using eddy resolving ocean configurations in climate models? Biases in sea surface temperature (SST) are the most important and indicative measure for the performance of AOGCMs. Their global magnitude and large scale patterns are controlled mostly by surface and cloud albedo. In this regard, sea ice distribution could significantly influence the large scale SST biases. Eddying resolution significantly improves representation of local oceanic features, especially in the regions of strong influence of western boundary currents and coastal upwelling and in the equatorial Pacific. In such regions, high ocean resolution helps considerably to reduce SST biases. Surface current representation also affects sea ice distribution and thereby SST. The sea ice covered high latitudes are also the region of deep water formation, which is the start point of the global-scale deep overturning circulation and play a role in the global-scale heat transport. Eddying ocean resolution leads to a significant reduction of climate biases across different models in some aspects, such as those around western boundary currents and in the equatorial Pacific region. But there still remain biases uncorrected by eddying ocean resolution, such as those in the coastal upwelling regions and at deep water formation sites. For these yet-uncorrected biases, different models sometimes look to show different behavior in terms of dependence on ocean resolution (Winton et al. 2014, Saba et al. 2016). It is difficult to identify the causes of such biases from these existing model experiments because they are differently designed and analysed. Well-coordinated model experiments and metrics for analysis would help improve our understanding on the dependence of climate biases on ocean model resolution.
Description of the scientific research activities entailed by the project proposed herein
Keeping an eye on the key-mechanisms
It has been recognized now for some time that the currents associated to the mesoscale ocean circulation features are responsible for much of the global heat and salt transports, and exert a quantitative control on all the mechanisms involved in the transformation and spreading of the water masses, including the air-sea exchanges of energy and CO2 (McWilliams, 2007). Recent eddying simulations are giving evidence of the emergence of a low frequency intrinsic variability related to this mesoscale turbulence (Penduff et al., 2011, Penduff et al. 2014). Together with the increased synoptic variability in high resolution atmospheric models, these modes of variability in the ocean are posing new challenges to studying climate changes. Understanding and quantifying these mechanisms requires an accurate physical anlaysis of the energetic cycles of the ocean circulation in different resolutions along with analysing contributions from the eddy-associated processes into a variety of large scale ocean circulation processes.
It is only recently that oceanic mesoscale eddies are regarded as one possible driver of the midlatitude large-scale low-frequency (LSLF) variability (e.g., review by Kwon et al. 2010). The idea that oceanic eddies may collectively generate a significant part of the climate variability comes as a big surprise, and is likely to change significantly our understanding of the global climate variability. Therefore, studying the LSLF variability in eddy-resolving rather than eddy-parameterized models is a rapidly growing research subject. Recent joint analyses of observations and comprehensive OGCM simulations suggest that about half of the LSLF variability in the ocean is intrinsic, and is generated only when a significant fraction of mesoscale eddies is dynamically resolved rather than diffusively parameterized (Penduff et al. 2011). A very robust LSLF variability mechanism, operating in a model of eddying, wind-driven oceanic gyres, has been discovered by Berloff and McWilliams (1999) and explained later by Berloff et al. (2007a). This mechanism—referred to as the Turbulent Oscillator—is driven by competition between the eddy-induced flow rectification and the eddy-induced potential vorticity anomalies, and both of these processes require explicit dynamical representation of the eddies, as they cannot be approximated by simple diffusive processes. What is missing now is a deeper fundamental understanding of the roles of the eddies on the larger scales, and verification of the eddy-driven intrinsic variability in physically more complete and dynamically more realistic models.
Conceptual approach: think globally – act regionally
In many respects “think globally – act regionally” will be a project motto. The project will address the role of synoptic and mesoscale eddies in the large scale variability of the ocean on different time scales and the ocean’s role in climate. Definitely, this is a global challenge, as the ocean evolves globally with strong signals being identified in all basins: North Atlantic (Gulev et al. 2013, Gulev and Latif 2015, McCarthy et al. 2015, Latif et al. 2006), Indian Ocean (Lee et al. 2015), Pacific Ocean (Zhong and Liu 2009). In this respect the global signals identified in e.g. ocean heat content and sea level (Gouretski and Koltermann 2007, Balmaseda et al. 2013, Von Schuckmann et al. 2015) ingrate to a large extent effects of mesoscale and synoptic variability in the ocean. Keeping this in mind, we will target in the project the understanding of the global ocean variability. However, specific emphases will be on the key-processes and key-regions which are of particular importance, including not only specific ocean basins but also semi-enclosed seas. This approach is also justified by constrains posed by the computer resource requirements: very high resolution simulations with ocean only GCMs and especially with coupled configurations, while are in plans, may require the resources which exceed potential of even best World’s computer centers. Regarding the above mentioned key-processes, for instance the role of the increase in salinity due to a change in water mass distribution that is related to a retreat of the Labrador Current and a northerly shift of the Gulf Stream (Saba et al. 2016) is a very task for the regional high resolution model configurations to be used in conjunction with global model set-ups.
Regional focus is also critical for the analysis of new air-sea coupling mechanisms which are strongly localized, e.g. over the western boundary currents and their extensions. Allowing eddies in the ocean components of climate models reveals new air-sea coupling mechanisms, such as the influence of the Kuroshio Extension variability on the Pacific storm tracks (Taguchi, 2014). Latif and Zhou (2014) presented a new estimate of the tropospheric response to mid-latitude SST anomalies in the Pacific Ocean. These preliminary results, regarding the influence of daily SST variability, are puzzling, and confirm that more investigation with high resolution ocean-atmosphere models are needed in order to build a synthetic, consensus view. This paradigm has been recently confirmed by the experiments with regional high resolution ocean and atmospheric models in the Kuroshio current system (Ma et al. 2015).
Finally, our approach is justified by the necessity to validate model results with observations. While presently global satellite altimetry data from AVISO as well as growing ARGO data base provide much better description of the ocean upper limb than few decades before, when looking into the regional mechanisms, the use of precise in-situ observations remains critical. For instance, understanding the mechanisms driving deep convection and the formation of intermediate and deep waters in the subpolar North Atlantic requires long-term time series of highly accurate observations which are available here (Falina et al. 2012, Sarafanov et al. 2012) from the ongoing field programs jointly maintained by IORAS and French institutions. Very similarly, understanding of sub-mesoscale processes in the marginal and semi-enclosed seas requires both high resolution modelling and high resolution observations, especially for the cross-shelf exchanges and their role in the dynamics of sea basins.
Outline of the Working Plan
Our Work Plan consists in the main Tasks, assigned to the major scientific objectives and technological challenges of the project, Work Packages associated with the Working Groups of the planned Ocean Modelling Laboratory (OML), cross-cutting the Tasks, and Key phenomena and key region studies (KPKRS) which will grow with the project advancement and the needs of the end users of our study.
Task 1. Development and adaptation of the global and regional ocean GCMs
Under this task we will adapt at OML several global and regional configurations of ocean GCMs in different resolutions allowing for end-to-end understanding the role of ocean eddies in mean ocean state and long-term variability. The choice of the resolution(s) is critical and goes far beyond the technological context. Ocean mesoscale variability in the ocean appears in a variety of transient features such as eddies, meanders, rings, waves and fronts with space scales of a about 10 to 100 km and time scales of 10 to 100 days. Ocean eddies spontaneously arise from the hydrodynamic instability of the major large scale current systems, as do the atmospheric synoptic features from instability of the large scale wind systems. Ocean mesoscale turbulence is often described as being the “weather system” of the global ocean by a dynamical analogy with the synoptic features of the atmosphere (McWilliams 2007, Barnier et al., 2011). As we mentioned above, atmospheric GCMs used in global climate models are always at least “eddy-permitting” in the sense that they allow for the development of transient synoptic circulation features. However, to allow the development of turbulence, a model needs resolve the relevant scales of instability. For the ocean mesoscale, this scale is the Rossby radius of deformation (the scale of the baroclinic instability) and it is generally assumed that one needs at least 2 grid points per Rossby to allow the growth of this instability.
Task 2. Development of diagnostic tools
This task will be focused on the development and adaptation of new diagnostics for the analysis and assessment of eddy-resolving turbulent model simulations and new algorithms for the computation of air-sea energy and mass exchanges. For understanding the role of ocean eddies in forming long-term variability of the ocean and its impact on climate, adequate and well justified diagnostic tools are as important as the ocean model configurations themselves. In many respects a proper diagnostics is an inseparable part of numerical experimentation with ocean and climate models, moreover, these two research tools largely feedback on each other. Besides the standard diagnostics (vertically and zonally integrated stream function, mixed layer depth, surface water mass transformation rates, meridional heat and fresh water transports, transports of the major currents, etc., see Barnier et al. 2006, Gulev et al. 2003), our set will include special diagnostics targeting the role of eddies in large scale circulation. These will include space-time spectral characteristics, analysis of propagation patterns, analysis of eddy’s life cycles, eddy-associated kinetic energy. Also of an interest will be diagnostic tools for quantifying the role of eddies in long-term variability, especially analysis of long-term variability modes and their association with air-sea interaction patterns (Gulev et al. 2013). Special attention will be paid to the diagnostics of effects of mesoscale eddies on air-sea interactions, which will include techniques targeting high-frequency variability of surface fluxes related to mesoscale ocean eddies. These will include (but will not be limited to) the analysis of probability density functions in surface fluxes and in the eddying ocean characteristics.
Task 3. Global and basin-scale long-term experiments with different ocean GCMs
Here we will perform a number of numerical experiments targeting the role of ocean mesoscale eddies and associated ocean intrinsic variability modes in global and regional ocean variability. With the project motto outline above we will look on the global characterises of eddies in moderately long (several decades) global forced experiments (including 50 member ensembles) in resolutions from 1/4 to 1/12 degrees for what both already existing (e.g. Penduff et al. 2014, Gregorio et al. 2015, Sérazin et al., 2015) and newly performed experiments. Then same duration as well as longer term experiments will be performed for the North Atlantic – Arctic domain which will be one of our key-regions given its critical role in forming long-term internal ocean variability on multidecadal scales which is potentially associated with (and likely originating from) the ocean eddy activity. For this purpose basin-scale and regional configurations of NEMO will be used (see, section 3). Of an interest will be the use of the regional NEMO set-up covering the North Atlantic and Arctic and recently developed by Dupont et al. (2015). An important part of this Task will be development and adaptation of the forcing functions for different model experiments. For this purpose we will use our experience gained through the development of DFS forcing (Brodeau et al. 2010) as well as regional high resolution hindcasts developed at IORAS using non-hydrostatic WRF – ARM model in high resolution (Gavrikov et al. 2016). The latter will be particularly relevant for the use in conjunction with CONCEPTS NEMO high resolution configuration (Dupont et al. 2015). Interactions with the CONCEPT group in Canada and France are already existing in the context of the Drakkar consortium.
Task 4. Regional high resolution experiments in the marginal and semi-enclosed seas
Under this task we will perform high resolution experiments for the analysis of the role of sub-mesoscale eddies in the marginal and semi-enclosed seas. Of a particular interest will be modeling mesoscale eddies using high resolution configurations of NEMO as well as using already adapted at IORAS the POM and STRiPE models in conjunction with high resolution atmospheric boundary conditions. Here we will use the potential of high resolution ocean modeling for simulating sub-mesoscale eddies and specifying their role in mediated surface fluxes over e.g. Black Sea under a variety of meteorological and oceanographic conditions. For this purpose we will perform runs for specific case studies in a very high resolution (less than 1 km) and also longer-term experiments in moderately high resolution (e.g. 2-4 km). Case studies will hint on the physical mechanisms steering the life cycle of these eddies and longer-term runs will provide long-term time series of eddy statistics to be further analyzed for understanding the mechanisms of mesoscale eddy interaction with larger scale circulation elements in the sea. For both types of experiments we will produce high-resolution atmospheric boundary conditions (winds and diabatic flux terms) using WRF-ARV non-hydrostatic configuration in spatial resolutions of several hundred meters. This will allow for resolving of high resolution mesoscale atmospheric features which may affect the development of mesoscale processes in the ocean and which are not represented even in high resolution reanalyses.
Task 5. Analysis of the mechanisms of mesoscale eddy impacts on multi-scale ocean and lower atmosphere dynamics
Here we will apply advanced diagnostics (Task 2) to the results of model experiments (Tasks 3, 4) in order to get insights on the major mechanisms behind the mesoscale eddy impacts on long-term ocean circulation and its climate variability. Further we will also analyse under this task the role of mesoscale and sub-mesoscale ocean eddies on surface fluxes and lower atmosphere. For these purposes we will use capabilities of the numerical diagnostics and theory spurring the advances of both. Theoretical studies will be particularly important here, but necessarily in conjunction with detailed diagnostics of numerical experiments. Numerical experimentation with global and regional high-resolution models provides a unique information about the dependence of the large-scale circulation patterns on the extent to which the eddies are resolved, but they can barely answer the questions about the mechanisms behind the interactions of the eddies with each other and with the mean flow without the help and conceptual guidance from theoretical models. On the other hand, theoretical and idealized model results may hint on the dynamical mechanisms, but to which extent these mechanisms are really pronounced in nature should be tested in realistic simulations and observations or we will find ourselves in an unfortunate situation, where the concepts become “simpler than the nature” (Albert Einstein). Our focus will be on the development of conceptual models of the eddy impact on the ocean regional variability (first of all in the North Atlantic) and on regional responses in the surface atmospheric conditions. For these we will consider in particular the two major effects of ocean mesoscales – the so-called “pressure adjustment mechanism” (Lindzen and Nigam 1987) and “vertical mixing mechanism” (e.g. Wallace et al. 1989). The latter can lead to significant modulation of the surface winds, and small-scale (oceanic mesoscale) wind anomalies are positively correlated with sea surface temperature (SST) anomalies (Chelton et al. 2001; Small et al. 2005). Of a special interest will be also analysis of responses in precipitation and water recycling patterns over eddying ocean (e.g. Minobe et al. 2008).
Task 6. Eddying ocean in climate system
This task will be largely associated with Task 5, harvesting to a large extent from the insights of Task 5 and going further to the development of fully coupled climate configuration with the eddy resolving oceanic block. This task will be performed in a close co-operation with the Voeykov Main Geophysical Observatory (MGO) group. The main focus here is providing a modeling basis for evaluation of the effect of resolving mesoscale eddies in the ocean and the atmosphere on simulation of the observed and projected climate evolution. For this purpose a new version of the MGO AGCM (Meleshko et al. 2014) will be designed for intensive multi-processor computations. The AGCM will be used in two resolutions – the relatively low (T63, ~1.8°) and the relatively high (T106, ~1.2°), while also higher spatial resolutions might be considered. First, AGCM simulations will be perfomed using SST and sea ice concentrations prescribed externally from observations according to forcing scenario corresponding to the CMIP5 “Historical”. This will allow for evaluating the AGCM performance. Further the MGO AGCM will be used as a component of the coupled AOGCM. The latter will be using NEMO configurations and the OASIS coupler (https://verc.enes.org/Oasis). The coupled AOGCM will be used for series of “Historical” simulations, as well as RCP4.5/RCP8.5 scenarios used by IPCC in its 5th Assessment Report. Scientific focus of these experiments will be on quantification of the impact of oceanic mesoscale eddies on the atmospheric circulation patterns in middle and high latitudes of the Northern Hemisphere, particularly addressing the role of the North Atlantic mesoscale eddies in the genesis and evolution of NAO and AO persistence and strength. Another avenue will go along with evaluation of the role of oceanic mesoscale eddies in the ocean-atmosphere energy exchange and quantifying its impact on the large-scale low/high pressure systems over the Eurasia and Eurasian Arctic. This will use further extend diagnostic results from Task 5 and will also target Bjerknes conjecture analysed in Gulev et al. (2013) using observational data. Finally we will evaluate the role of oceanic signals in the genesis and evolution of high-impact mesoscale weather systems which develop mainly due to baroclinic instability. Here we will look on the role of eddying ocean in both diabatic moisture and heat sources for the atmosphere and in generating atmospheric mechanisms for transporting moisture and heat to the continents, thus forming regional weather and climate extremes.
These major Tasks will be complemented with the key phenomena and key region studies (KPKRS) targeting specific key mechanisms or/and specific regional manifestations of these mechanisms in the global and regional ocean circulation and its impact on climate variability and change. The list of KPKRS will be updated and finally developed during the first year of the project term. For now we can mention preliminary the following KPKRS:
KPKRS 1. Atlantic Meridional Overturning Circulation (AMOC) and AMO. If the existence of significant intrinsic low frequency variability (LFV) in the AMOC has been demonstrated, we still poorly understand the mechanisms that are driving it, and we have yet to identified the full range of the time and space scales it concerns. It is now important to determine the imprint of this LFV on the various components of the AMOC, i.e. its gyre and overturning components and their respective contributions to the LFV of the upper ocean heat content and meridional heat transport. A key phenomenon that will get our attention is the imprint of the ocean intrinsic LFV on the atmospheric variability which need to be quantified. This will be looked at via the sensitivity of AGCMS to the forcing by the SST intrinsic anomalies. We shall also investigate correlation of the intrinsic LFV with major climate indexes (NAO, AMO).
KPKRS 2. Atlantic – Arctic heat and freshwater exchanges on decadal and multidecadal scales. One expects that in the Arctic, global warming will drive a seasonal disappearance of the sea ice and will amplify the hydrological cycle. The storage and export of freshwater from the Arctic should consequently be modified: the net export and the routes by which it exits and the time scales at which this occurs, as well as the partitioning between liquid freshwater and sea-ice should be significantly modified. Therefore, we shall focus our studies on; the dynamics controlling the freshwater storage in the Beaufort gyre; the mechanisms by (and the time scales at) which freshwater anomalies generated in the Arctic Ocean interior are transferred into the boundary currents and propagate (distinguishing between liquid or sea-ice) along the boundaries to the exit gateways. Eddies are suspected to play a significant role here.
KPKRS 3. Gulf Stream and moisture/heat sources for the mid latitude atmosphere. The meandering/eddying of the strong temperature front of the Gulf-Stream will modify the local patterns of atmospheric convection and the vertical moisture flux which may impact on the cyclogenis and feedback on the NAO. This process suggests that the NAO could be in part driven by the ocean LFV. This has to be clarified, and the atmospheric variability should be compared under a very stable and laminar Gulf-Stream (the one simulated by present climate models) and a very turbulent one with sharp fronts.
KPKRS 4. Black Sea sub-mesoscale eddies, coastal processes and local diabatic signals. Here key phenomena are the exchange between large shelves and deep ocean, and the role of eddies in a large scale flow strongly controlled by topography, winds and buoyancy forcing. Shelves-open ocean exchanges and the associated fluxes of Carbon and pollutants must be largely eddy driven. While propagating on shelves, cyclonic and anticyclonic sub-mesoscale eddies can travel across isobaths, because the variations of their own vorticity can compensate for the depth changes, thus allowing for potential vorticity conservation. This activity will help to better understand the mechanism steering the dynamics of sub-mesoscale eddies in semi-enclosed basin.
KPKRS 5. Tracking moisture/heat transports in climate model in response to ocean signals. This will be analysed in the forced runs with atmospheric GCM first and will be precisely validated against estimates based on reanalyses and WRF simulations. Next the mechanisms of the moisture advection and its association with atmospheric synoptic transients and the mean flow state will be analysed in the coupled simulations. This will provide quantitative estimates of the role of eddying ocean in forming atmospheric water cycle and mid latitude water recycling.
Work packages and project organization within OML
To implement Tasks and KPKRSs we will organize the project into six Work Packages (WPs) which will cross-cut the Tasks with KPKRSs being built into the project structure (Figure 6). We plan the following main Work Packages of the project:
WP 2. Dynamics of the eddying ocean. This WP will focus on the adaptation and development of the relevant diagnostics and building long-term time series of the key characteristics of ocean circulation. Also here long-term time series of ocean state and circulation characteristics will be analysed using appropriate metrics quantifying the role of synoptic and mesoscale eddies in ocean dynamics.
WP3. Air-sea exchanges. This WP will be concentrated on the development of forcing functions for ocean GCM experiments (in a close co-operation with WP1) and will also provide the analysis of the role of ocean eddies in forming local anomalies of surface energy exchanges as well as analysis of long-term variations in the surface diabatic signals (in co-operation with WP2).
WP4. Model validation. This WP in a close co-operation with WP2 and WP3 will provide detailed validation of the model experiments using best available data sets of deep ocean observations and surface fluxes (see section 2.3 for details). Model validation will also include the development of additional diagnostic tools (Task 2 and WP2).
WP5. Regional processes in seas and shelf regions. This WP will concentrate on the analysis of model simulations in marginal and semi-enclosed seas using model configurations and diagnostics developed in WP1 and WP2 as well results from WP3 on air-sea interactions. Particularly WP3 will provide for this WP5 very high resolution forcing fields developed using WRF model.
WP6. Climate responses to the ocean signals. This WP will concentrate on the global and regional coupled configurations employing MGO atmospheric model and NEMO set-ups in different resolution and will provide the analysis of the impact of eddy resolving ocean on climate variability and change including new climate projections. A close co-operation between this WP and WP1 and WP3 will be essential.
Within Ocean Modelling Laboratory (OML), the WPs will be carried by 5 major Working Groups (WGs):
-> Numerical ocean modelling group
-> Process studies and theory group
-> Air-Sea Interaction group
-> Observational and model validation group
-> Climate modelling group
-> Coastal processes and regional circulation group
These Working Groups will handle different Work Packages by contributing to several of them and, thus, maintain close co-operation with each other, as shown in Figure 6. This will provide an effective work flow and deliverable exchanges between the WPs, facilitated through the close dependencies between the WGs. Further details of the Project Implementation Plan are provided in section 5.
The major benefit of the project carried out by OML for the research community will be a new suite of data archives, diagnostic tools, models and scenarios, as a working instrument to address the observed changes in the ocean general circulation with improved accuracies and uncertainty estimates. An outstanding feature of the proposed research is an integrated end-to-end approach to the analysis of the impact of synoptic and mesoscale processes on the World Ocean state and long-term variability. The major project instrument – a suite of ocean model configurations – will be made publicly available to the maximum extent possible and will be hopefully used in further national and international integrated assessments of the Earth System Change, including those developed by IPCC. We stress, that our working instrument will not be frozen and will allow including other modelling and diagnostic tools.
Benefits for end users of the OML products will be secured by a set of applications that will provide various estimates of impacts of the ocean signals on the global and regional climate variability and change with a particular focus on the Northern Extratropics and the North Atlantic. Up to the end of the project (and thereafter), we will keep the list of case studies open. Thus, after the major advances in Tasks 1 through 3, the research project anticipates “harvesting”, i.e., using its output (products) with additional efforts to address the needs of numerous regional and sub-regional applications relevant to the end-users (climate prediction, operational oceanography, development of new observational systems).
Under the broader impact of the study and benefits for IORAS and the Russian Academy as well as for Moscow State University (MSU) and MGO, we undertake intangible investments in capacity building that will remain for decades to come and will bring IORAS into the very exclusive club of the Institutions expert in eddy-resolving modelling of the global Ocean. This means building a new generation of early career scientists (master and PhD students and postdocs) participating in the project and developing high profile science. The measures of this impact will be the numbers of publications in top ranked international journals, number of PhD and doctoral defenses, the number of presentations at major international conferences and the success of the OML in fundraising at the national and international levels.
Description of the scientific techniques and methods that will be used to accomplish the project objectives
Scientific techniques and methods used in the project will consist in numerical configurations of ocean GCMs based on NEMO family, climate modelling tools, computational methods for estimating surface fluxes and boundary conditions for numerical experimentation with ocean models as well as data bases of surface fluxes and flux related variables and of observational data sets for validating models. Furthermore, these will consist of diagnostic methodologies, briefly mentioned in section 2 and partly to be developed under the project.
NEMO – based ocean model configurations
The main working horse for the project will be NEMO family of ocean general circulation models. For the last ten years, a consortium of European partners has been joining efforts in building a consistent modelling framework in which the interactions between critical fine scale processes and multi-decadal variability of the large-scale ocean circulation can be quantified, allowing a true assessment and improvement of their parameterization in climate prediction models. For that purpose the consortium, referred to as the DRAKKAR consortium (https://www.drakkar-ocean.eu/) is developing and continuously improving a state of the art hierarchy of global and regional model configurations based on the NEMO modelling framework (www.nemo-ocean.eu).
Two main reasons motivated the choice of NEMO (Nucleus for European Modelling of the Ocean, http://www.nemo-ocean.eu/). First NEMO is a very complete state-of-the-art modeling framework for oceanographic research, operational oceanography, seasonal forecast and climate studies, as it includes 5 major components: the blue ocean (ocean dynamics, NEMO-OPA), the white ocean (sea-ice, NEMO-LIM), the green ocean (biogeochemistry, NEMO-TOP), an adaptative 2-way mesh refinement software (AGRIF) that allows interactive regional zooms, and a data assimilation component (NEMO-TAM). Besides this completeness, the choice of NEMO was also made because its evolution, reliability and maintenance are organised and controlled by a European Consortium gathering six important research and operational institutions in Europe (CNRS and Mercator-Ocean in France, NERC and the Met Office in the UK, CMCC and INGV in Italy). These institutions are committed to sustain the development of the NEMO as a state-of-the-art ocean model code system suitable for both research and operational work.
An important component of the project model configurations will be the already mentioned CREG12 model which was developed as part of the CONCEPTS (Canadian Operational Network of Coupled Environmental PredicTion Systems) initiative. It represents a high-resolution (1/12°) ice–ocean regional model developed for the domain covering the North Atlantic and the Arctic oceans (Dupont et al. (2015). The global ORCA12 domain (ORCA family grid at a nominal horizontal resolution of 1/12° in both longitudinal and latitudinal directions; Drakkar Group, 2007) is used to derive a seamless (i.e., the “north-fold” discontinuity of the global grid is removed) regional domain covering the whole Arctic Ocean and parts of the North Atlantic down to 27° N. The horizontal grid consists of 1580×1817 points on which resolution varies from 8 km at the open boundary in the Atlantic Ocean to an average of 5 km in the Arctic and down to slightly below 2 km in some of the southern channels of the Canadian Arctic Archipelago (Fig. 9).
Atmospheric model and coupling
As an atmospheric model we will use a new version of the MGO AGCM (Meleshko et al. 2014). This model in both preliminary considered resolutions (T63 and T106) will be designed for intensive multi-processor computations.
The model has T63 spectral resolution corresponding to grid size of about 190 km and 25 vertical levels (T63L25). The model incorporates major physical processes governing the atmospheric circulation: solar and terrestrial radiation transfer, convection, large-scale and convective precipitation, clouds, and turbulent fluxes of sensible/latent heat in the boundary layer and in the upper atmosphere. The radiation package accounts for diurnal and annual variation of solar fluxes at the top of the atmosphere and includes greenhouse gases and aerosols. In calculating surface fluxes, a grid box considers three types of surfaces with different physical properties (water, sea ice and land). The atmospheric model is coupled to a land-surface block that takes into account processes over vegetation of different types, bare soil and snow-covered surface. The land-surface model describes heat/water transfer and water discharge in the ground of 3 m depth that is represented with 4 layers. The model has been validated and it realistically reproduces the main large-scale patterns of the atmospheric circulation through the seasonal cycle. MGO AGCM has been widely used in climate research and experimental weather prediction for decades. The MGO AGCM has also participated in a number of model intercomparisons, the most relevant to this project being the CLIVAR International Climate of the Twentieth Century Project (C20C) (Scaife et al. 2009, Kucharski et al. 2009, Zhou et al. 2009).
For coupling of MGO AGCM with different NEMO configurations the OASIS coupler will be used. The OASIS coupler was developed in the framework of the EU FP7 IS-ENES2 project. OASIS is a software allowing synchronized exchanges of coupling information between numerical codes representing different components of the climate system. Current OASIS developers are CERFACS (Toulouse, France) and Centre National de la Recherche Scientifique (Paris, France). OASIS3-MCT, the new version of the OASIS coupler interfaced with the Model Coupling Toolkit (MCT) from the Argonne National Laboratory, offers today a fully parallel implementation of coupling field, regridding and exchange. Low-intrusiveness, portability and flexibility are OASIS3-MCT key design concepts as for all previous OASIS versions. An important difference with respect to previous OASIS3.3 is that there is no longer a separate coupler executable: OASIS3-MCT is a coupling library that needs to be linked to the model components, with the main function of interpolating and exchanging the coupling fields between these components.
Air-sea flux and flux-related variables datasets and related computations
Numerical experimentation with ocean GCMs critically depends on the boundary conditions at the surface represented by fluxes of heat, moisture and momentum. Input for building these boundary conditions is typically represented by atmospheric state variables and surface fluxes available from global data sets and parameterizations of fluxes which are used for computing high resolution forcing fields. For providing surface boundary conditions for the model experiments we will use both already existing forcing functions and those which will be developed during the project term. We will employ the global DRAKKAR forcing sets (DFS), versions 4 and 5, developed and regularly updated within DRAKKAR consortium (Brodeau et al. 2010, Dussin et al., 2016). This forcing set is based on ERA-Interim reanalysis (Dee et al. 2011), incorporates many important corrections of variables and advanced bulk formulae and demonstrates in many respects better skills compared to the CORE forcing (Large and Yeager 2009). Furthermore, within the project, we will also adopt and develop several additional forcing sets for global and regional ocean simulations. These will include (but will not be limited to) centennial long forcing based on long-term reanalyses of 20th century (20CR and ERA-CLIM) as well as newly developed forcing based on JRA-55 reanalysis. For the computation of turbulent fluxes we will use advanced bulk parameterizations from the COARE family (Fairall et al. 2003, 2011) as well as radiative fluxes available from ISCCP up to date (see Bishop et al. 1997 for reference). Global precipitation data over the ocean for several decades will used from reanalyses and satellite data including GPCP, TRM and recently available GMP (Beranger et al. 2006, Sommer et al. 2016).
For the development and validation of forcing functions for ocean modelling as well as analysis of model results in conjunction with atmospheric responses we will use different data sets of surface fluxes and flux related variables available currently from reanalyses, satellite products and in-situ observations (Figure 10). In the past 10 years the number of available datasets has increased significantly compared to the preceding decade as is shown schematically in Figure 10. This increase has been driven by the advent of a wide range of higher resolution reanalysis products (e.g., CFSR, Modern Era Reanalysis for Research and Applications (MERRA)), the production of new satellite datasets the Hamburg Ocean Atmosphere Parameters and Fluxes from Satellite Data (HOAPS), and the combination of information from different sources to produce hybrid (CORE) and synthesis datasets (OAFLUX). Development of ship-based flux datasets (NOC1.1a, NOC2) remains limited by the inherent sampling problems that arise from the distribution of ship observations, with very limited amounts of information in many regions, particularly the southern hemisphere. No single product can be identified as the best available as each has its own strengths and weaknesses and the selection of dataset must be guided by the problem being tackled, for example, globally balanced products are needed for ocean model forcing. Details of these data sets (which all are available at IORAS servers) can be found in Josey et al. (2013), Valdivieso et al. (2015) and other relevant references.
For validating model results and process studies we will use a number of observational data sets targeting the dynamics of ocean processes. First of all this is satellite altimetry data from AVISO (Archiving, Validation and Interpretation of Satellite Oceanographic data) available and distributed by the Copernicus Marine and Environment Monitoring Service (CMEMS) (http://www.marine.copernicus.eu). AVISO consolidates sea surface height data at various resolutions (1/4 degree spatial resolution and 5-day time resolution being the most common for different AVISO products) from Topex/Poseidon, Jason-1 & -2, ERS-1 & -2, Envisat, Cryosat-2, and SARAL/Altika missions. Data currently cover the period from 1993 onwards and are regularly updated.
Another invaluable source of observational data is the Argo buoy archive representing the broad-scale global array of temperature/salinity profiling floats, deployed in different regions of the ocean with the amount of active floats exceeding 3300 and covering global ocean with spatially variable resolution (http://www.argo.ucsd.edu/Gridded_fields.html). These floats provide more than 100,000 temperature and salinity profiles down to 2000 meters each year. Additionally, trajectories from Argo floats drifting at depth can be used to calculate velocities which may or may not be gridded to examine the currents at 1000 dbars. Argo first deployments started in 2000, however the period of relatively dense coverage of most ocean basins with Argo starts in 2006. Both individual Argo profiles and different Argo-based gridded products are available at IORAS and will be extensively used for validating model numerical experiments especially in the regions of key-processes (deep convection, western boundary current extension regions, etc.).
Finally, for some key-regions we will extensively use data from research cruises. In this respect data from more than 20 years of monitoring of subpolar North Atlantic perfumed by Russian (IORAS) research vessels and French (IFREMER) research vessels will be of a special importance. These data cover the period from the mid 1990s onwards (Sarafanov et al. 2012, Falina et al. 2012, Mercier et al. 2015) and provide full depth high resolution cross-sections along 60N and 54N in the Atlantic as well as several surveys of the Denmark Strait. Being combined with numerous Labrador Sea surveys (Yashayaev et al. 2015), this database represents invaluable source for validating model results and analysis of key-processes, such as Atlantic-Arctic transports, dynamics of mixed layer depth and convection, formation of intermediate and deep waters – the phenomena on the core of the Atlantic climate variability. Also for the analysis of numerical experiments in the marginal and semi-enclosed seas we will use observational data available from observational campaigns of IORAS in different basins, first of all in the Black Sea. These high resolution data include instrumental measurements during multiply sub-mesoscale eddy passages along with the deployment of multiple current meters. For the period starting from 2000 these data are complimented by satellite imagery of very high resolution (MERIS/Envisa). These datasets together with the measured velocity fields will enable us to estimate the cross-shore transport of the quantities under investigation due to sub-mesoscale eddies.