skip to content

Cambridge NERC Doctoral Training Partnerships

Graduate Research Opportunities
Brief summary: 
This project aims at reconstructing the multidecadal variability of the Atlantic Ocean's Meridional Overturning Circulation using machine learning techniques, and assessing the relative importance of natural versus anthropogenic climate drivers during the past two millennia.
Importance of the area of research concerned: 
The Atlantic Meridional Overturning Circulation (AMOC) exerts an important control on the global climate system. On one hand, the AMOC influences ocean and atmosphere temperatures on a wide range of temporal and spatial scales, and on the other hand, it regulates variations in atmospheric CO2 by partitioning carbon between the surface and deep ocean. Monitoring and reconstructing the variability of the AMOC is thus crucial for assessing future rapid climate changes. Since mooring observations began in 2004 (Bryden et al., 2005), the AMOC have steadily weakened. However, the observational record is too short to assess whether the current AMOC slowdown is a result of multidecadal variability or an anthropogenically forced trend. Climate model studies reveal that robust physical relationships exist between specific North Atlantic oceanographic-atmospheric parameters and AMOC variations (Zhang, 2008). These relationships –or AMOC “fingerprints”– are extremely valuable as they can be exploited to provide a longer context to assess the dominant variability of the AMOC, its impact on global climate change, and to build reliable predictions of future AMOC variations.
Project summary : 
This project aims at extending the AMOC monitoring record by developing fingerprints derived from climate model simulations and observations to link AMOC with variables that are measured extensively in palaeo-climate records. The fingerprints will be inferred using machine learning techniques, which have a demonstrated effectiveness in extracting leading indicators from high-dimensional model output (Ayala-Solares et al., 2018). Specifically, machine learning algorithms will be employed to explore and identify climate variables and systems model that enable the most effective representation of AMOC as observed in the RAPID mooring programme over 2004-2017 (Bryden et al., 2005). Finally, the derived AMOC variability fingerprints will be used to reconstruct past AMOC strength over the last 2,000 years using suitable subsets of marine and terrestrial proxy-climate records.
What will the student do?: 
The candidate will apply machine learning methods to identify consistent physical relations across CMIP5 and CMIP6 climate models that will be ultimately used to produce new high-temporal resolution reconstructions of the AMOC for the past 2,000 years. The fingerprints will be constructed for natural and anthropogenically forced climate scenarios, and their sensitivity to the climate models will be quantified. The candidate will determine the optimal combination of climate variables that maximizes detectability and representation of AMOC variability. In particular, she/he will explore the sensitivity to spatial coverage and temporal sampling to resolve the most suitable proxy network for reconstructing AMOC, which will be obtained using subsets of the latest global proxy-climate data collections of the last 2,000 years.
References - references should provide further reading about the project: 
Ayala-Solares, J. R., Hua-Liang Wei, and G. R. Bigg. "The variability of the Atlantic meridional circulation since 1980, as hindcast by a data-driven nonlinear systems model." Acta Geophysica 66.4 (2018): 683-695.
Bryden, Harry L., Hannah R. Longworth, and Stuart A. Cunningham. "Slowing of the Atlantic meridional overturning circulation at 25 N." Nature 438.7068 (2005): 655.
Zhang, Rong. "Coherent surface‐subsurface fingerprint of the Atlantic meridional overturning circulation." Geophysical Research Letters 35.20 (2008).
You can find out about applying for this project on the Department of Geography page.
Dr Francesco Muschitiello
Scott Hosking
Department of Geography Graduate Administrator