物理信息人工智能与海洋混合实验建模博士研究员（科研助理）

Ocean mixing plays a fundamental role in regulating Earth's climate by controlling how heat, carbon and nutrients are transported through the deep ocean. Small scale turbulent mixing drives the buoyancy gain of dense abyssal waters and enables the return flow of the global Meridional Overturning Circulation (MOC), yet the pathways linking localized mixing processes to basin-scale circulation remain poorly resolved. Recent results from a unique 22-year tracer-release experiment in the Brazil Basin have provided the first large-scale observational evidence connecting abyssal water transformation to overturning pathways and revealed an unexpected 2000 m upwelling of abyssal waters through the Rio de Janeiro Fracture Zone, a deep passage linking the Brazil and Angola Basins.

These observations highlight the important role that fracture zones may play in controlling water-mass transformation and the return of abyssal waters toward the ocean interior, and suggest that these processes are not adequately represented in existing ocean and climate models. This project will investigate the physical mechanisms within fracture zones that drive ocean mixing, water-mass transformation, tracer horizontal and vertical transport, and exchange between overturning circulation cells. The student will combine unique tracer observations, turbulence measurements, data-science methods, and numerical modelling to quantify the mixing-driven circulation through these deep passages.

The project will build on an established framework based on the MIT General Circulation Model (MITgcm), extending previous canyon simulations to represent an open fracture-zone geometry connecting the Brazil and Angola Basins. Numerical experiments will be used to investigate how topography, stratification, and mixing influence tracer dispersion and overturning circulation. A key objective is to use these observational and modelling insights to improve the representation of ocean mixing in large-scale ocean and climate models. Existing parameterisations are largely based on local turbulence measurements or idealised process studies and do not fully account for the integrated transport pathways revealed by the long-term tracer observations. By combining observations, simulations, and statistical analysis, the project will provide new constraints on the rates and pathways of abyssal upwelling and water mass transformation. The project will further explore the use of physics-informed artificial intelligence to accelerate environmental simulation and support the development of computationally efficient modelling tools.

Machine-learning surrogate models will be trained on ensembles of MITgcm simulations to learn the relationships between topography, mixing intensity, stratification, and large-scale tracer transport. These AI emulators will reproduce the essential dynamics of the numerical model at a low computational cost, enabling rapid exploration, uncertainty quantification, and testing of new mixing parameterisations.

The project will therefore contribute to the growing integration of physical modelling, data science, and AI in environmental prediction. The project combines observational oceanography, environmental fluid dynamics, numerical simulation, data science, and artificial intelligence to address a fundamental challenge in climate prediction while exploring next-generation computational approaches for Earth-system modelling.

Funded by: EPSRC Engineering and Physical Sciences Research Council £21,805. The studentship will provide funding of fees and a stipend of £21,805 per annum for 2026/7.