MESoscale Ocean eddies and Climate Predictions (MESO-CLIP)

Lead Research Organisation: National Oceanography Centre
Department Name: Science and Technology


Mesoscale ocean eddies (MOEs) are swirls of water (typically a few hundred km in diameter) that are ubiquitous in the World Ocean. MOEs are the oceanic equivalent of weather systems in the atmosphere. In analogy to weather systems MOEs cannot be predicted a long time in advance. In computer models of the ocean MOEs can only develop if the spatial scale that the model can resolve is small enough. Typically a model needs to be able to resolve scales of about 30km (at mid-latitudes) to start generating MOEs. An ocean model is then said to be eddy-permitting. For a good representation of MOEs the resolved spatial scales need to be at least 10 km. Ocean models with that resolution are often referred to as eddy-resolving. Until recently, the grid resolution in climate models used for climate prediction was too coarse (100 km and more) for MOEs to be simulated. This is now changing and the latest generation of climate models under development use ocean components that are eddy-permitting (and soon eddy-resolving). When and where MOEs occur in high resolution models depends on initial conditions (the temperature, salinity and velocities at the beginning of the model simulation). Even small changes in initial conditions will eventually lead to different MOE fields. This is analogous to weather patterns typically adopting different patterns in a matter of days when the initial conditions are perturbed at the beginning of a forecast.

How MOEs feed back on climate variability and predictability is still largely unknown. However, some recent studies suggest that MOEs could affect ocean and atmosphere variability on interannual to decadal timescales. Cutting edge climate models currently under development use eddy-permitting (e.g. HadGEM3-H in the UK) and eddy-resolving (e.g. CM2.6 in the US) oceans and therefore there is a need to get a better understanding of how MOEs affect forecasts based on such models. The main goal of MESO-CLIP will be to determine how initial conditions (temperatures, salinities, velocities) have to be perturbed in eddy-permitting/resolving ocean models to assess the uncertainty in forecasts. We will use a hierarchy of numerical models: (i) an uncoupled global ocean model run at horizontal grid resolutions of 1/4degree (25km at Equator) and 1/12degree (9 km at Equator), (ii) the latest coupled ocean-atmosphere model currently under development at the UK MetOffice (HadGEM3-H) which uses a 1/4degree ocean component, and (iii) an eddy-resolving (1/20degree) resolution idealised coupled ocean-atmosphere model. With this set of models we will be able to address how the presence of MOEs in the ocean affect the predictability and variability of ocean and atmosphere and how important coupled processes (interactions between the ocean and the atmosphere) are likely to be. MESO-CLIP will therefore provide valuable knowledge about forecast uncertainties in present and future high resolution coupled models that will be used for climate predictions.

Planned Impact

MESO-CLIP will assess the impact of increasing the resolution in the ocean component of climate models on the variability and predictability of the climate system.
This is crucial if we want to understand the uncertainty in forecasts produced by the latest high resolution coupled climate models.

Primarily the research in MESO-CLIP will benefit developers of high-resolution models in the UK (HadGEM3-H) and abroad (e.g. EC-Earth, GFDL CM2.5/CM2.6). Our collaboration with the UK MetOffice (Adrian Hines, project partner) will ensure that the knowledge acquired in MESO-CLIP gets transferred to HadGEM3-H, the UK's most advanced coupled climate model. This will contribute to the UK's position as a leader in the development and research involving cutting edge climate models. Our results will also be relevant to operational oceanography (e.g. MERCATOR, HYCOM, FOAM) where eddying models are widely used. MESO-CLIP will provide an estimate of the importance of knowing the details of the ocean mesocale eddy field in the initial conditions.

Further down the line the results in MESO-CLIP will be relevant in the assessment of the uncertainties in climate predictions from eddy-permitting/resolving models that will feed into future IPCC reports (beyond CMIP5). The results from MESO-CLIP may also be relevant for seasonal forecasts as mesoscale eddies affect the ocean variability on a wide range of timescales. This aspect could become relevant to the (re-)insurance industry. This sector increasingly uses information from coupled climate model in its CAT (catastrophe) models used to determine the optimal insurance pricing. NOC is an associated member of the Willis Research Network (WRN) and we will make sure that WRN scientists are invited
to the project workshops in MESO-CLIP.
Description We developed a method to determine the "memory" of the surface ocean circulation. By "memory" we mean the time it takes for the information contained in anomalies of the ocean circulation at a given time (e.g. position of meanders in large currents such as the Gulf Stream or the Kuroshio or of ocean eddies) to be lost. The "memory" in the ocean is equivalent to the typical length of the time between passing weather systems in the atmosphere (e.g. a succession of high and low pressure centres) - a time often referred to as the synoptic timescale. We found the ocean "memory" of the surface currents to typically be between 10 and 30 days with the longest times of more than 70 days found in the North Pacific. We could also show that the times are similar in observations and in a state-of-the-art ocean model.

We have found a new mechanism linking the Loop Current in the Gulf of Mexico to the variability of the Gulf Stream along the U.S. east coast and into the North Atlantic. About one a year the length of the Loop Current expands until and eddy is being shed. During expansion (contraction) phases of the Loop Current length we find negative (positive) velocity anomalies in the Gulf Stream downstream of the Gulf of Mexico. Both in observations and a high resolution ocean model these anomalies are seen as pulses that within a few weeks propagate from the southern tip of Florida to Cape Hatteras along the U.S. coast.

The key goal of MESO-CLIP was to move towards a better understanding of how small scale ocean features (order of a few tens to a few hundred km) can affect the predictability of ocean and climate. In models which are used in forecasting these small scale ocean features (ocean mesoscale eddies) are very sensitive to small changes in temperature and salinity.
In MESO-CLIP we developed an approach how to optimally perturb the ocean mesoscale in high resolution ocean models which is an important step toward better exploiting these high resolution models for forecasts.
Exploitation Route Being able to quantify the ocean "memory" could have implications for operational ocean forecasting centres and for forecasting purposes in general as the method we developed can in principle easily be applied to any system that needs to be forecasted.

The optimal perturbation technique for ocean mesoscale eddies developed in MESO-CLIP could affect the way forecasting centres are initialising their models.
Sectors Environment,Other

Title Ensemble generation 
Description Using the 1/4° and 1/12° versions of the NEMO global ocean models we have generated ensemble simulations. Each ensemble member experiences the same surface forcing condition and differ from other ensemble members through small perturbations (non optimised) applied to the initial conditions. 
Type Of Material Improvements to research infrastructure 
Provided To Others? No  
Impact These ensembles will allow us to study how model trajectories start to diverge in response to subtle perturbations to the initial conditions. The 1/4° and 1/12° simulations will show whether there is a resolution dependence for the divergence between the ensemble members. Finally, the 1/4° ensemble will serve as a benchmark against which we can compare the simulations using Linear Optimal perturbations. 
Title Linear Optimal Perturbations 
Description I note that none of the above terms available above for the type of research tool/method is appropriate (it should be "Numerical modelling" or something similar). I just chose "Technology assay or reagent" to be able to complete the section Linear optimal perturbations (LOP) are used to perturb numerical models - in our case ocean models. LOPs are designed to generate the fastest possible divergence of model trajectories with respect to a given metric (e.g. sea surface temperatures, meridional overturning). In our case the metric are mesoscale ocean eddies. 
Type Of Material Technology assay or reagent 
Provided To Others? No  
Impact LOPs have been widely - and successfully use in atmospheric sciences (e.g. weather forecasting) before. 
Title Linear optimal perturbations in an eddy-permitting model 
Description The Linear Optimal Perturbation method has been set up and tested in an eddy-permitting (e.g. 0.25° resolution) version of NEMO. 
Type Of Material Improvements to research infrastructure 
Provided To Others? No  
Impact This method will allow us to generate perturbations that are optimised to assess the impact of mesoscale features on the ocean circulation. 
Title NEMO 
Description NEMO stands for "Nucleus for European Modelling of the Ocean". NEMO is the numerical ocean model of choice within the European research community and it is developed in the framework of a French-British-Italian consortium. NEMO is widely used in research centres across Europe for fundamental research as well as for operational purposes (NEMO is the ocean model that has been adopted for the forecasting models of the UK MetOffice). 
Type Of Material Computer model/algorithm 
Provided To Others? Yes  
Impact NEMO is partly being developed at NOC (NOC is a partner in the NEMO consortium) and as such developments undertaken at NOC are being used by a large community in Europe and beyond. 
Description Chaocean 
Organisation Laboratory of Glaciology and Environmental Geophysics (LGGE)
Country France 
Sector Academic/University 
PI Contribution Advice and discussions about the development of model ensemble strategies for eddying ocean models.
Collaborator Contribution Same as above since this is crucial to the success of both MESO-CLIP and CHAOCEAN.
Impact Common meetings in Vienna (April 2013) and Grenoble, France (January 2014)
Start Year 2013
Description ACSIS - OSNAP - RAPID meeting in Oxford 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other audiences
Results and Impact Presentation on chaotic vs externally forced variability of the Atlantic Meridional overturning circulation on intra and inter-annual timescales. The meeting was attended by 100+ international scientists.
Year(s) Of Engagement Activity 2017
Description DRAKKAR meeting 2017, Grenoble, France 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other audiences
Results and Impact Presentation on oceanic synoptic timescales. The meeting was attended by about 80 international scientists.
Year(s) Of Engagement Activity 2017
Description Invited Seminar - Stockholm University (Sweden) 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other audiences
Results and Impact 10-20 senior scientists attended the presentation and subsequent discussions led to ongoing collaborations.
Year(s) Of Engagement Activity 2015