Met Office Academic Partnership: advancing weather and climate science
7 November 2022
The Met Office Academic Partnership (MOAP), a cluster of research excellence consisting of eight UK universities, is aimed at advancing weather and climate research. The University of Edinburgh, through EPCC, is delighted to have joined this partnership.
MOAP brings together the Met Office and leading UK institutions in weather and climate science, while reaching researchers in other sectors including health, technology, artificial intelligence, and the social sciences.
The Partnership has invested in jointly funded Chairs at UK universities, with myself appointed to the Joint Chair for the University of Edinburgh. Each Chair will act as the leader for joint research programmes and related activities within their institution and across the Partnership.
EPCC will use its supercomputing and data science expertise to help the Met Office deliver its next-generation numerical weather modelling system, and exploit the data that results from it. Our experience with ARCHER2 and designing the EIDF gives us unique experience that will support our research with the Met Office.
A key goal of the Partnership is to provide an outstanding environment to develop the science leaders of tomorrow in this very challenging area of research that is of strategic importance. This will include sponsorship of student placements, internships and staff secondments, both at the Met Office and the University.
EPCC has a long track record of working with the Met Office, starting with the parallelisation of the Unified Model more than 30 years ago. Recent collaborations include:
• EPCC worked with the Met Office to replace its Large-Eddy Model (LEM), which was fundamentally limited in its scalability, with a new code, the Met Office NERC Cloud model (MONC). MONC has been demonstrated to scale to tens of thousands of cores, enabling scientific simulations that previously were not possible. Its modular nature has made it possible to explore GPUs and even FPGAs without the need to fundamentally restructure the code. As part of the MONC development, EPCC also developed a lightweight I/O server for asynchronous writing of diagnostics and prognostics and in-situ data analytics.
• The Numerical Atmospheric Modelling Environment (NAME) code is used to simulate a wide range of atmospheric dispersion phenomena. However its performance was limited because its parallelisation was shared-memory only. EPCC collaborated with the Met Office to introduce distributed memory parallelism using MPI to complement the shared-memory parallelisation, to enable both larger simulations and improved time to solution.
MOAP brings together UCL, University of Bristol, University of Exeter, University of Leeds, University of Oxford, University of Reading, University of Edinburgh and University of Birmingham.
See the Met Office website for further information: Met Office Academic Partnerships
