Process‐Based Climate Model Development Harnessing Machine Learning: I. A Calibration Tool for Parameterization Improvement
Abstract The development of parameterizations is a major task in the development of weather and climate models. Model improvement has been slow in the past decades, due to the difficulty of encompassing key physical processes into parameterizations, but also of calibrating or “tuning” the many free...
Main Authors: | , , , , , , , , , , , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
American Geophysical Union (AGU)
2021-03-01
|
Series: | Journal of Advances in Modeling Earth Systems |
Subjects: | |
Online Access: | https://doi.org/10.1029/2020MS002217 |
id |
doaj-5c6eea312f284b8995d3a6bb583117e3 |
---|---|
record_format |
Article |
spelling |
doaj-5c6eea312f284b8995d3a6bb583117e32021-03-26T15:36:29ZengAmerican Geophysical Union (AGU)Journal of Advances in Modeling Earth Systems1942-24662021-03-01133n/an/a10.1029/2020MS002217Process‐Based Climate Model Development Harnessing Machine Learning: I. A Calibration Tool for Parameterization ImprovementFleur Couvreux0Frédéric Hourdin1Daniel Williamson2Romain Roehrig3Victoria Volodina4Najda Villefranque5Catherine Rio6Olivier Audouin7James Salter8Eric Bazile9Florent Brient10Florence Favot11Rachel Honnert12Marie‐Pierre Lefebvre13Jean‐Baptiste Madeleine14Quentin Rodier15Wenzhe Xu16CNRM, University of Toulouse Meteo‐France CNRS Toulouse FranceLMD‐IPSL Sorbonne University CNRS Paris FranceExeter University Exeter UKCNRM, University of Toulouse Meteo‐France CNRS Toulouse FranceThe Alan Turing Institute London UKCNRM, University of Toulouse Meteo‐France CNRS Toulouse FranceCNRM, University of Toulouse Meteo‐France CNRS Toulouse FranceCNRM, University of Toulouse Meteo‐France CNRS Toulouse FranceExeter University Exeter UKCNRM, University of Toulouse Meteo‐France CNRS Toulouse FranceCNRM, University of Toulouse Meteo‐France CNRS Toulouse FranceCNRM, University of Toulouse Meteo‐France CNRS Toulouse FranceCNRM, University of Toulouse Meteo‐France CNRS Toulouse FranceCNRM, University of Toulouse Meteo‐France CNRS Toulouse FranceLMD‐IPSL Sorbonne University CNRS Paris FranceCNRM, University of Toulouse Meteo‐France CNRS Toulouse FranceExeter University Exeter UKAbstract The development of parameterizations is a major task in the development of weather and climate models. Model improvement has been slow in the past decades, due to the difficulty of encompassing key physical processes into parameterizations, but also of calibrating or “tuning” the many free parameters involved in their formulation. Machine learning techniques have been recently used for speeding up the development process. While some studies propose to replace parameterizations by data‐driven neural networks, we rather advocate that keeping physical parameterizations is key for the reliability of climate projections. In this paper we propose to harness machine learning to improve physical parameterizations. In particular, we use Gaussian process‐based methods from uncertainty quantification to calibrate the model free parameters at a process level. To achieve this, we focus on the comparison of single‐column simulations and reference large‐eddy simulations over multiple boundary‐layer cases. Our method returns all values of the free parameters consistent with the references and any structural uncertainties, allowing a reduced domain of acceptable values to be considered when tuning the three‐dimensional (3D) global model. This tool allows to disentangle deficiencies due to poor parameter calibration from intrinsic limits rooted in the parameterization formulations. This paper describes the tool and the philosophy of tuning in single‐column mode. Part 2 shows how the results from our process‐based tuning can help in the 3D global model tuning.https://doi.org/10.1029/2020MS002217calibrationlarge‐eddy simulationsphysical parameterizationsprocess‐oriented model tuningsingle‐column models |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Fleur Couvreux Frédéric Hourdin Daniel Williamson Romain Roehrig Victoria Volodina Najda Villefranque Catherine Rio Olivier Audouin James Salter Eric Bazile Florent Brient Florence Favot Rachel Honnert Marie‐Pierre Lefebvre Jean‐Baptiste Madeleine Quentin Rodier Wenzhe Xu |
spellingShingle |
Fleur Couvreux Frédéric Hourdin Daniel Williamson Romain Roehrig Victoria Volodina Najda Villefranque Catherine Rio Olivier Audouin James Salter Eric Bazile Florent Brient Florence Favot Rachel Honnert Marie‐Pierre Lefebvre Jean‐Baptiste Madeleine Quentin Rodier Wenzhe Xu Process‐Based Climate Model Development Harnessing Machine Learning: I. A Calibration Tool for Parameterization Improvement Journal of Advances in Modeling Earth Systems calibration large‐eddy simulations physical parameterizations process‐oriented model tuning single‐column models |
author_facet |
Fleur Couvreux Frédéric Hourdin Daniel Williamson Romain Roehrig Victoria Volodina Najda Villefranque Catherine Rio Olivier Audouin James Salter Eric Bazile Florent Brient Florence Favot Rachel Honnert Marie‐Pierre Lefebvre Jean‐Baptiste Madeleine Quentin Rodier Wenzhe Xu |
author_sort |
Fleur Couvreux |
title |
Process‐Based Climate Model Development Harnessing Machine Learning: I. A Calibration Tool for Parameterization Improvement |
title_short |
Process‐Based Climate Model Development Harnessing Machine Learning: I. A Calibration Tool for Parameterization Improvement |
title_full |
Process‐Based Climate Model Development Harnessing Machine Learning: I. A Calibration Tool for Parameterization Improvement |
title_fullStr |
Process‐Based Climate Model Development Harnessing Machine Learning: I. A Calibration Tool for Parameterization Improvement |
title_full_unstemmed |
Process‐Based Climate Model Development Harnessing Machine Learning: I. A Calibration Tool for Parameterization Improvement |
title_sort |
process‐based climate model development harnessing machine learning: i. a calibration tool for parameterization improvement |
publisher |
American Geophysical Union (AGU) |
series |
Journal of Advances in Modeling Earth Systems |
issn |
1942-2466 |
publishDate |
2021-03-01 |
description |
Abstract The development of parameterizations is a major task in the development of weather and climate models. Model improvement has been slow in the past decades, due to the difficulty of encompassing key physical processes into parameterizations, but also of calibrating or “tuning” the many free parameters involved in their formulation. Machine learning techniques have been recently used for speeding up the development process. While some studies propose to replace parameterizations by data‐driven neural networks, we rather advocate that keeping physical parameterizations is key for the reliability of climate projections. In this paper we propose to harness machine learning to improve physical parameterizations. In particular, we use Gaussian process‐based methods from uncertainty quantification to calibrate the model free parameters at a process level. To achieve this, we focus on the comparison of single‐column simulations and reference large‐eddy simulations over multiple boundary‐layer cases. Our method returns all values of the free parameters consistent with the references and any structural uncertainties, allowing a reduced domain of acceptable values to be considered when tuning the three‐dimensional (3D) global model. This tool allows to disentangle deficiencies due to poor parameter calibration from intrinsic limits rooted in the parameterization formulations. This paper describes the tool and the philosophy of tuning in single‐column mode. Part 2 shows how the results from our process‐based tuning can help in the 3D global model tuning. |
topic |
calibration large‐eddy simulations physical parameterizations process‐oriented model tuning single‐column models |
url |
https://doi.org/10.1029/2020MS002217 |
work_keys_str_mv |
AT fleurcouvreux processbasedclimatemodeldevelopmentharnessingmachinelearningiacalibrationtoolforparameterizationimprovement AT frederichourdin processbasedclimatemodeldevelopmentharnessingmachinelearningiacalibrationtoolforparameterizationimprovement AT danielwilliamson processbasedclimatemodeldevelopmentharnessingmachinelearningiacalibrationtoolforparameterizationimprovement AT romainroehrig processbasedclimatemodeldevelopmentharnessingmachinelearningiacalibrationtoolforparameterizationimprovement AT victoriavolodina processbasedclimatemodeldevelopmentharnessingmachinelearningiacalibrationtoolforparameterizationimprovement AT najdavillefranque processbasedclimatemodeldevelopmentharnessingmachinelearningiacalibrationtoolforparameterizationimprovement AT catherinerio processbasedclimatemodeldevelopmentharnessingmachinelearningiacalibrationtoolforparameterizationimprovement AT olivieraudouin processbasedclimatemodeldevelopmentharnessingmachinelearningiacalibrationtoolforparameterizationimprovement AT jamessalter processbasedclimatemodeldevelopmentharnessingmachinelearningiacalibrationtoolforparameterizationimprovement AT ericbazile processbasedclimatemodeldevelopmentharnessingmachinelearningiacalibrationtoolforparameterizationimprovement AT florentbrient processbasedclimatemodeldevelopmentharnessingmachinelearningiacalibrationtoolforparameterizationimprovement AT florencefavot processbasedclimatemodeldevelopmentharnessingmachinelearningiacalibrationtoolforparameterizationimprovement AT rachelhonnert processbasedclimatemodeldevelopmentharnessingmachinelearningiacalibrationtoolforparameterizationimprovement AT mariepierrelefebvre processbasedclimatemodeldevelopmentharnessingmachinelearningiacalibrationtoolforparameterizationimprovement AT jeanbaptistemadeleine processbasedclimatemodeldevelopmentharnessingmachinelearningiacalibrationtoolforparameterizationimprovement AT quentinrodier processbasedclimatemodeldevelopmentharnessingmachinelearningiacalibrationtoolforparameterizationimprovement AT wenzhexu processbasedclimatemodeldevelopmentharnessingmachinelearningiacalibrationtoolforparameterizationimprovement |
_version_ |
1724201914093010944 |