Understanding and improving AMOC forecasts in inter-annual to decadal climate predictions

The Atlantic Meridional Overturning circulation (AMOC) is a key ocean current in the North Atlantic which plays an important role in Earth's climate. For example, the AMOC - and its role in transporting heat - is the main cause of warm winters in the UK and Europe compared to other continents at similar latitudes. Changes in the strength of the AMOC have also been linked with changes in regional temperatures and shifting rainfall patterns over Europe and Africa, as well as extreme weather events such as hurricanes. Paleo-proxy evidence also suggests that the AMOC has been through rapid, large, and persistent changes in the past during periods of global climate change. Therefore, the ability to predict the AMOC from years-to-decades ahead would have many benefits for society.

Unfortunately, AMOC predictions in many decadal prediction systems are hampered by poor performance and physically unrealistic behaviour such as large and substantial "drifts". Given the AMOC's important role in transporting heat and freshwater, these drifts in AMOC impact other variables in the ocean and atmosphere. Therefore, this poor performance is limiting predictability of regional climate and are a serious barrier to providing useful climate forecasts. Improving the predictions of the AMOC is crucial to improve the quality of, and the confidence in, decadal climate predictions.

The causes of the AMOC drift are currently not known and improving the predictions will require better understanding of the processes controlling the AMOC to guide development of the next generation of climate prediction systems. The representation of the AMOC in decadal predictions is limited by the ability of models to reproduce important physical processes such as air-sea interactions driving the formation of deep dense water in the North Atlantic, which is a key component of the AMOC. Biases in surface heat and freshwater fluxes or sea surface properties of temperature, salinity, and sea ice lead to errors in dense water formation. Ocean mixing processes that modify these water masses - for example, at subsurface overflows or within ocean eddies - are also poorly represented and can lead to errors in circulation. Thus, both air-sea interactions and internal ocean processes can lead to significant errors in the distribution of sub-surface ocean properties that impact the AMOC.

Therefore, the overall aim of this project is to understand how the representation of key processes in ocean models can interfere with successful reproduction and prediction of the AMOC. The student will do this by first characterizing and understanding the evolution of the AMOC in a multi-model ensemble of predictions of past climate. Through detailed process-based analysis they will explore the reasons for poor performance of AMOC predictions and understand which oceanic and atmospheric processes are key for a successful multi-annual prediction. In particular, they will use a novel application of surface water mass transformation diagnostics to address the question of whether errors in air-sea interactions or internal ocean processes dominate the drifts in AMOC in model-based predictions.

After exploring the causes of poor model performance, the student will develop specific hypotheses that can be tested in new climate model simulations. To that end, the student will design and perform modelling experiments with the Met Office coupled climate model to test for improvements in AMOC prediction in a state-of-the-art Decadal Prediction System. One key outcome of the project will then be recommendations to the Met Office, which will lead to improved models and predictions of climate in the years ahead.