NSFGEO-NERC Stirring at the Walls - A dynamical boundary model for the ocean

Climate change, because of its enormous social, economic and political consequences, reigns as the leading scientific problem of our times. The primary scientific tool in the study of future climate is the coupled numerical model, in which the various components of the climate system interact, producing an estimate of a future climate state. The resulting projections receive global exposure and impact global policy.

Of primary importance in the prediction of climate on interannual to centennial time scales is the ocean. Dynamically sound ocean models are integral to reliable climate forecasting, yet due to gaps in our scientific understanding of ocean function, they suffer from a fundamental weakness. Most of the kinetic energy in the ocean resides in the so-called `mesoscale', a term referring to ocean phenomenon with time scales of days to months and length scales of 10's to 100's of km. The mesoscale through its large scale feedbacks has been shown to be a major factor determining intrinsic ocean variability on interannual to decadal timescales, which covers a significant fraction of the temporal spectrum over which the ocean contributes importantly to climate. It has been estimated that up to 80% of ocean variability is due to such intrinsic processes.

This presents climate forecasting with a practical problem: the mesoscale consists of features that are small (50 km)compared to the basin scale (6000 km) and their direct numerical resolution over the entire globe for times required in climate simulations is far beyond current computer resources. Present computational resources for climate projection allow for modest but incomplete mesoscale resolution (25 km), necessitating the parametrisation of the remaining sub-grid scale dynamics. Reliable ocean models will employ parametrisations based on dynamics. This is not current modelling practice. Current practice models sub-grid scale dynamics using viscous and mixing representations and tunes the related parameters to match output to present observations, justifying this modelling by arguing that large scale low frequency winds drive the basin scale circulation that subsequently develops the mesoscale by instabilities. The amplitude of the large scale circulation is then set by balancing the energy flow into the large scale with the energy flow out of the large scale into the mesoscale. The mechanisms by which the mesoscale loses energy are not addressed directly and are not as well understood. Models are tuned to representative mesoscale energy levels so they exhibit reasonable decadal scale variability and to reproduce essential elements of the ocean circulation, such as accurate separation of the Gulf Stream from the east coast of the US. These parametrisations have, however, no basis in the dynamics of the flow. The nonlinearity of the climate system means that there is no assurance that a parametrisation tuned to present conditions will perform well when modelling a changing climate. The same difficulty arises in the modelling of palaeoclimates where the underlying flow structure is far from present day observations. A dynamically based parametrisation, especially as it addresses mesoscale dissipation, is needed to address this issue.

We argue that there is a gap between the very high spatial and temporal resolution required in global ocean models to accurately resolve the flow near ocean boundaries and the lower resolution required to resolve the motion of the ocean interior. A dynamical boundary model of the form proposed here can exploit this gap allowing more accurate simulations at lower computational cost while simultaneously increasing our knowledge of boundary mixing processes. This addresses directly the NERC priority of ``studies of water circulation in seas and oceans on a variety of temporal and spatial scales based on modelling''. This project will test a key hypothesis that, if true, will change the modelling of ocean circulation.

Scientific:
The project will test a key hypothesis that, if true, will change the modelling of ocean circulation. There is precedent for the success of physically based parametrisations in moving ocean modelling forward; the universally adopted Gent-McWilliams parametrisation captured the essential physics of the feedback of the mesoscale on the general circulation and the utility and accuracy of climate models using it experienced a quantum leap forward. The impact is thus through proving that the hypothesis is valid, that its implementation is possible and then disseminating the results as widely and rapidly as possible.

The modelling community:
Dr George Nurser of the Ocean Modelling Group at Project Partner NOC Southampton is confident that if we can produce an add-on module for the MITgcm, then he will be able to work with us to adapt it to work with community Nucleus for European Modelling of the Ocean (NEMO), state-of-the-art modelling framework of ocean related engines. Once implemented, NOC has committed to reviewing our approach for potential inclusion in the strategic NEMO model configurations in the context of the Joint Marine Modelling Programme with the UK Met Office. This programme acts to pull through developments such as a successful Dynamical Boundary Model (DBM) from the research community for inclusion in the models used for ocean prediction at a range of spatio-temporal scales. NOC has undertaken to facilitate discussion with the JMMP management group on an appropriate timeline for inclusion of the outcomes of this work into JMMP configurations. This is an extremely valuable commitment to our project as it will maximise the likelihood of the widespread acceptance of a successful DBM and thus provide a pathway for the rapid adoption of any demonstrably successful results.

The insurance industry:
Project Partner WTW will provide time and feedback from WTW staff in the Reinsurance and Analytics Technology teams for the duration of the project to take advantage of any opportunities arising from this project, and to help develop usable output for the reinsurance industry. This industry contact will provide an invaluable pathway for spreading DBM ideas into commercial sectors and will tie in closely with the work with NOC to implement a DBM in NEMO. UCL Mathematics also has close contacts with London-based Risk Management Solutions Ltd (RMS) having placed two Geophysical Fluid Dynamics PDRAs there. They run their own predictive models. If we can demonstrate a sufficiently direct method of implementing a DBM in their code we will be able to encourage them to use a DBM in their climate studies related to their insurance interests.

The Economy:
The IPCC comments that ''Accurate simulation of the ocean in climate models is essential for the correct estimation of transient heat uptake and transient climate response, ocean CO2 uptake, sea-level rise, and coupled climate modes such as ENSO''. If this project is successful then the economic impact flows from basing political decisions on more accurate climate predictions through the adoption of DBMs in climate models in the same way that improved adiabatic modelling is a consequence of the Gent-McWilliams parametrisation.

Societal - Mathematicians into Environmental Sciences:
The physical understanding and skills in analytical and numerical methods that the PDRA will learn will form an excellent basis for a career and make the PDRA highly employable in organisations like the UK Met Office and Project Partner NOC (where the PDRA will have close and regular contact) and companies like WTW and RMS. This is an important societal impact addressing the identified shortage of mathematically trained workers in environmental fields. UCL Mathematics has a strong record of placing PDRAs and PhDs into industry like those in RMS and WTW. The US graduate student will also benefit from training in state-of-the-art modelling in the environmental sciences.