A New Machine Learning Approach to Speed Up and Improve Climate Change Simulations

(NOWACK_UENV21ARIES)

A New Machine Learning Approach to Speed Up and Improve Climate Change Simulations

(NOWACK_UENV21ARIES)

Project Description

Supervisors

Dr Peer Nowack (School of Environmental Sciences, University of East Anglia) contact me

Dr Luke Abraham (Department of Chemistry, University of Cambridge)

Dr Fiona O’Connor (UK Meteorological Office)

Professor Manoj Joshi (School of Environmental Sciences, University of East Anglia)

 

Project Background

State-of-the-art computer models of Earth’s climate are key to our understanding of climate change. These Earth system models (or ESMs) are based on fundamental scientific knowledge of physical, chemical and biological processes shaping the Earth’s atmosphere and oceans. A particularly important ESM component is atmospheric chemistry, which modulates global warming by influencing atmospheric composition, and which gives rise to the existence of the stratospheric ozone layer (best known for its role in protecting us from harmful solar ultraviolet radiation). However, atmospheric chemistry also doubles the time it takes to run ESM simulations on supercomputers, and therefore pose a central bottleneck in climate change simulations.

Your project will focus on a recently developed machine learning approach to resolve this bottleneck, with the goal to substantially speed-up climate change simulations. Specifically, you will develop a new hybrid model version of the UK Earth System Model (UK-ESM; https://ukesm.ac.uk/) in which atmospheric chemistry is replaced by a machine learning model learned from NASA satellite observations and existing ESM data. You will benchmark the performance of your novel approach against established methods, compare a variety of machine learning tools, and evaluate the hybrid-model concerning its ability to realistically model the physics and chemistry of past and possible future climates.

Training

You will become an expert in the science of climate change and in advanced data science tools, including machine learning and supercomputing. You will further gain insights into the science behind the ozone hole. You will acquire these highly sought-after skills by working with leading experts at UEA, the University of Cambridge and the UK Met Office, where you will visit and work with UK-ESM developers. As a member of an interdisciplinary cohort of PhD candidates, you will develop many other transferable skills, and you will present your work at national and international conferences.

Person Specification

We are looking for a highly motivated individual with a keen interest in learning more about machine learning, programming and the science behind climate change. We welcome applications from candidates of all numerical disciplines with degrees in Mathematics, Physics, Computer Science, Chemistry, Engineering or Environmental Modelling.

References

  • 1. Nowack P, Braesicke P, Haigh J, Abraham NL, Pyle J, and Voulgarakis A. Using machine learning to build temperature-based ozone parameterizations for climate sensitivity simulations, Environmental Research Letters 13, 104016 (2018), doi.org/10.1088/1748-9326/aae2be.
  • 2. Nowack P, Abraham NL, Maycock A, Braesicke P, Gregory JM, Joshi MM, Osprey A, and Pyle J. A large ozone-circulation feedback and its implications for global warming assessments, Nature Climate Change 5, 41-45 (2015), doi.org:10.1038/nclimate2451.
  • 3. Nowack P, Ong QYE, Braesicke P, Haigh J, Abraham NL, Pyle J, Voulgarakis A. Machine learning parameterizations for ozone: climate model transferability, Conference Proceedings of the 9th International Workshop on Climate Informatics, 263-268 (2019), doi.org:10.5065/y82j-f154.
  • 4. Nowack P, Braesicke P, Abraham NL, Pyle J. On the role of ozone feedback in the ENSO amplitude response under global warming, Geophysical Research Letters 44, 3858-3866 (2017), doi.org:10.1002/2016GL072418.
  • 5. Nowack P, Abraham NL, Braesicke P, Pyle J. The impact of stratospheric ozone feedbacks on climate sensitivity estimates, Journal of Geophysical Research: Atmospheres 123, 4630-4641 (2018), doi.org:10.1002/2017JD027943.

Key Information

  • This project has been shortlisted for funding by the ARIES NERC DTP and will start on 1st October 2021. The closing date for applications is 23:59 on 12th January 2021.
  • Successful candidates who meet UKRI’s eligibility criteria will be awarded a NERC studentship, which covers fees, stipend (£15,285 p.a. for 2020-21) and research funding. For the first time in 2021/22 international applicants (EU and non-EU) will be eligible for fully-funded UKRI studentships. Please note ARIES funding does not cover visa costs (including immigration health surcharge) or other additional costs associated with relocation to the UK.
  • ARIES students benefit from bespoke graduate training and ARIES provides £2,500 to every student for access to external training, travel and conferences. Excellent applicants from quantitative disciplines with limited experience in environmental sciences may be considered for an additional 3-month stipend to take advanced-level courses in the subject area.
  • ARIES is committed to equality, diversity, widening participation and inclusion in all areas of its operation. We encourage enquiries and applications from all sections of the community regardless of gender, ethnicity, disability, age, sexual orientation and transgender status. Academic qualifications are considered alongside significant relevant non-academic experience.
  • All ARIES studentships may be undertaken on a part-time or full-time basis, visa requirements notwithstanding
  • For further information, please contact the supervisor. To apply for this Studentship click on the “Apply now” link below.

Applications are Open

Apply now