Improving Met Office Weather and Climate Forecasts with Bespoke Multigrid Solvers

Andrew Malcolm; Eike Hermann Müller; Robert Scheichl

doi:10.1007/978-3-031-48683-8_1

Improving Met Office Weather and Climate Forecasts with Bespoke Multigrid Solvers

Andrew Malcolm
, Eike Hermann Müller
, Robert Scheichl

The Met Office
Institut for Applied Mathematics
Universität Heidelberg

Research output: Chapter or section in a book/report/conference proceeding › Book chapter

11 Downloads (Pure)

Abstract

At the heart of the Met Office climate and weather forecasting capabilities lies a sophisticated numerical model which solves the equations of large-scale atmospheric flow. Since this model uses semi-implicit time-stepping, it requires the repeated solution of a large sparse system of linear equations with hundreds of millions of unknowns. This is one of the computational bottlenecks of operational forecasts and efficient numerical algorithms are crucial to ensure optimal performance. We developed and implemented a bespoke multigrid solver to address this challenge. Our solver reduces the time for solving the linear system by a factor two, compared to the previously used BiCGStab method. This leads to significant improvements of overall model performance: global forecasts can be produced 10–15% faster. Multigrid also avoids stagnating convergence of the iterative scheme in single precision. By allowing better utilisation of computational resources, our work has led to estimated annual cost savings of £300k for the Met Office.

Original language	English
Title of host publication	More UK Success Stories in Industrial Mathematics
Editors	P. J. Ashton
Place of Publication	Cham, Switzerland
Publisher	Springer
Pages	3-8
Number of pages	6
ISBN (Electronic)	9783031486838
ISBN (Print)	9783031486821
DOIs	https://doi.org/10.1007/978-3-031-48683-8_1
Publication status	Published - 23 Apr 2025

Publication series

Name	Mathematics in Industry
Volume	42
ISSN (Print)	1612-3956
ISSN (Electronic)	2198-3283

Acknowledgements

The authors thank their collaborators at the Met Office and the members of the GungHo project. In particular, we would like to acknowledge the important contributions to the early stages of this project made by Dr Markus Gross, who tragically passed away in January 2022.

Funding

This work was funded through NERC grants NE/K006754/1 and NE/J005576/1.

Funders
Natural Environment Research Council

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 13 Climate Action

ASJC Scopus subject areas

Computer Science Applications
Industrial and Manufacturing Engineering
Computational Mathematics
Applied Mathematics

Access to Document

10.1007/978-3-031-48683-8_1

Improving Met Office Weather and Climate Forecasts with Bespoke Multigrid Solvers (accepted author manuscript)
Crown Copyright, Met Office
Accepted author manuscript, 498 KB

https://arxiv.org/abs/2307.04528Licence: CC BY-NC-ND

Cite this

Improving Met Office Weather and Climate Forecasts with Bespoke Multigrid Solvers. / Malcolm, Andrew; Müller, Eike Hermann; Scheichl, Robert.
More UK Success Stories in Industrial Mathematics. ed. / P. J. Ashton. Cham, Switzerland: Springer, 2025. p. 3-8 (Mathematics in Industry; Vol. 42).

Research output: Chapter or section in a book/report/conference proceeding › Book chapter

@inbook{952a88e73e804eac890cd6425868053b,

title = "Improving Met Office Weather and Climate Forecasts with Bespoke Multigrid Solvers",

abstract = "At the heart of the Met Office climate and weather forecasting capabilities lies a sophisticated numerical model which solves the equations of large-scale atmospheric flow. Since this model uses semi-implicit time-stepping, it requires the repeated solution of a large sparse system of linear equations with hundreds of millions of unknowns. This is one of the computational bottlenecks of operational forecasts and efficient numerical algorithms are crucial to ensure optimal performance. We developed and implemented a bespoke multigrid solver to address this challenge. Our solver reduces the time for solving the linear system by a factor two, compared to the previously used BiCGStab method. This leads to significant improvements of overall model performance: global forecasts can be produced 10–15\% faster. Multigrid also avoids stagnating convergence of the iterative scheme in single precision. By allowing better utilisation of computational resources, our work has led to estimated annual cost savings of £300k for the Met Office.",

author = "Andrew Malcolm and M{\"u}ller, \{Eike Hermann\} and Robert Scheichl",

year = "2025",

month = apr,

day = "23",

doi = "10.1007/978-3-031-48683-8\_1",

language = "English",

isbn = "9783031486821",

series = "Mathematics in Industry",

publisher = "Springer",

pages = "3--8",

editor = "Ashton, \{P. J.\}",

booktitle = "More UK Success Stories in Industrial Mathematics",

}

TY - CHAP

T1 - Improving Met Office Weather and Climate Forecasts with Bespoke Multigrid Solvers

AU - Malcolm, Andrew

AU - Müller, Eike Hermann

AU - Scheichl, Robert

PY - 2025/4/23

Y1 - 2025/4/23

N2 - At the heart of the Met Office climate and weather forecasting capabilities lies a sophisticated numerical model which solves the equations of large-scale atmospheric flow. Since this model uses semi-implicit time-stepping, it requires the repeated solution of a large sparse system of linear equations with hundreds of millions of unknowns. This is one of the computational bottlenecks of operational forecasts and efficient numerical algorithms are crucial to ensure optimal performance. We developed and implemented a bespoke multigrid solver to address this challenge. Our solver reduces the time for solving the linear system by a factor two, compared to the previously used BiCGStab method. This leads to significant improvements of overall model performance: global forecasts can be produced 10–15% faster. Multigrid also avoids stagnating convergence of the iterative scheme in single precision. By allowing better utilisation of computational resources, our work has led to estimated annual cost savings of £300k for the Met Office.

AB - At the heart of the Met Office climate and weather forecasting capabilities lies a sophisticated numerical model which solves the equations of large-scale atmospheric flow. Since this model uses semi-implicit time-stepping, it requires the repeated solution of a large sparse system of linear equations with hundreds of millions of unknowns. This is one of the computational bottlenecks of operational forecasts and efficient numerical algorithms are crucial to ensure optimal performance. We developed and implemented a bespoke multigrid solver to address this challenge. Our solver reduces the time for solving the linear system by a factor two, compared to the previously used BiCGStab method. This leads to significant improvements of overall model performance: global forecasts can be produced 10–15% faster. Multigrid also avoids stagnating convergence of the iterative scheme in single precision. By allowing better utilisation of computational resources, our work has led to estimated annual cost savings of £300k for the Met Office.

UR - https://www.scopus.com/pages/publications/105003773033

U2 - 10.1007/978-3-031-48683-8_1

DO - 10.1007/978-3-031-48683-8_1

M3 - Book chapter

AN - SCOPUS:105003773033

SN - 9783031486821

T3 - Mathematics in Industry

SP - 3

EP - 8

BT - More UK Success Stories in Industrial Mathematics

A2 - Ashton, P. J.

PB - Springer

CY - Cham, Switzerland

ER -