Deep Learning to Estimate Model Biases in an Operational NWP Assimilation System

Abstract Model bias is one of the main obstacles to improved accuracy and reliability in numerical weather prediction conducted with state‐of‐the‐art atmospheric models. To deal with model bias, a modification of the standard four‐dimensional variational (4D‐Var) algorithm, called weak‐constraint 4D...

Full description

Saved in:
Bibliographic Details
Main Authors: Patrick Laloyaux, Thorsten Kurth, Peter Dominik Dueben, David Hall
Format: Article
Language:English
Published: American Geophysical Union (AGU) 2022-06-01
Series:Journal of Advances in Modeling Earth Systems
Subjects:
Online Access:https://doi.org/10.1029/2022MS003016
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849720149519630336
author Patrick Laloyaux
Thorsten Kurth
Peter Dominik Dueben
David Hall
author_facet Patrick Laloyaux
Thorsten Kurth
Peter Dominik Dueben
David Hall
author_sort Patrick Laloyaux
collection DOAJ
description Abstract Model bias is one of the main obstacles to improved accuracy and reliability in numerical weather prediction conducted with state‐of‐the‐art atmospheric models. To deal with model bias, a modification of the standard four‐dimensional variational (4D‐Var) algorithm, called weak‐constraint 4D‐Var, has been developed where a forcing term is introduced into the model to correct for the bias that accumulates along the model trajectory. This approach reduced the temperature bias in the stratosphere by up to 50% and is implemented in the European Centre for Medium‐Range Weather Forecasts operational forecasting system. Despite different origins and applications, data assimilation (DA) and Deep Learning are both able to learn about the Earth system from observations. In this paper, a deep learning approach for model bias correction is developed using temperature retrievals from radio occultation (RO) measurements. Neural networks (NNs) require a large number of samples to properly capture the relationship between the temperature first‐guess trajectory and the model bias. As running the Integrated Forecasting System (IFS) DA system for extended periods of time with a fixed model version and at realistic resolutions is computationally very expensive, we have chosen to train, the initial NNs are trained using the ERA5 reanalysis before using transfer learning on 1 year of the current IFS model. Preliminary results show that convolutional NNs are adequate to estimate model bias from RO temperature retrievals. The different strengths and weaknesses of both deep learning and weak constraint 4D‐Var are discussed, highlighting the potential for each method to learn model biases effectively and adaptively.
format Article
id doaj-art-53c1fbf707cd4417bbb90050daa937ba
institution DOAJ
issn 1942-2466
language English
publishDate 2022-06-01
publisher American Geophysical Union (AGU)
record_format Article
series Journal of Advances in Modeling Earth Systems
spelling doaj-art-53c1fbf707cd4417bbb90050daa937ba2025-08-20T03:12:00ZengAmerican Geophysical Union (AGU)Journal of Advances in Modeling Earth Systems1942-24662022-06-01146n/an/a10.1029/2022MS003016Deep Learning to Estimate Model Biases in an Operational NWP Assimilation SystemPatrick Laloyaux0Thorsten Kurth1Peter Dominik Dueben2David Hall3European Centre for Medium‐Range Weather Forecasts Reading UKNvidia Corporation Santa Clara CA USAEuropean Centre for Medium‐Range Weather Forecasts Reading UKNvidia Corporation Santa Clara CA USAAbstract Model bias is one of the main obstacles to improved accuracy and reliability in numerical weather prediction conducted with state‐of‐the‐art atmospheric models. To deal with model bias, a modification of the standard four‐dimensional variational (4D‐Var) algorithm, called weak‐constraint 4D‐Var, has been developed where a forcing term is introduced into the model to correct for the bias that accumulates along the model trajectory. This approach reduced the temperature bias in the stratosphere by up to 50% and is implemented in the European Centre for Medium‐Range Weather Forecasts operational forecasting system. Despite different origins and applications, data assimilation (DA) and Deep Learning are both able to learn about the Earth system from observations. In this paper, a deep learning approach for model bias correction is developed using temperature retrievals from radio occultation (RO) measurements. Neural networks (NNs) require a large number of samples to properly capture the relationship between the temperature first‐guess trajectory and the model bias. As running the Integrated Forecasting System (IFS) DA system for extended periods of time with a fixed model version and at realistic resolutions is computationally very expensive, we have chosen to train, the initial NNs are trained using the ERA5 reanalysis before using transfer learning on 1 year of the current IFS model. Preliminary results show that convolutional NNs are adequate to estimate model bias from RO temperature retrievals. The different strengths and weaknesses of both deep learning and weak constraint 4D‐Var are discussed, highlighting the potential for each method to learn model biases effectively and adaptively.https://doi.org/10.1029/2022MS003016model errorneural networksweak‐constraint 4D‐Vardata assimilationartificial intelligenceECMWF
spellingShingle Patrick Laloyaux
Thorsten Kurth
Peter Dominik Dueben
David Hall
Deep Learning to Estimate Model Biases in an Operational NWP Assimilation System
Journal of Advances in Modeling Earth Systems
model error
neural networks
weak‐constraint 4D‐Var
data assimilation
artificial intelligence
ECMWF
title Deep Learning to Estimate Model Biases in an Operational NWP Assimilation System
title_full Deep Learning to Estimate Model Biases in an Operational NWP Assimilation System
title_fullStr Deep Learning to Estimate Model Biases in an Operational NWP Assimilation System
title_full_unstemmed Deep Learning to Estimate Model Biases in an Operational NWP Assimilation System
title_short Deep Learning to Estimate Model Biases in an Operational NWP Assimilation System
title_sort deep learning to estimate model biases in an operational nwp assimilation system
topic model error
neural networks
weak‐constraint 4D‐Var
data assimilation
artificial intelligence
ECMWF
url https://doi.org/10.1029/2022MS003016
work_keys_str_mv AT patricklaloyaux deeplearningtoestimatemodelbiasesinanoperationalnwpassimilationsystem
AT thorstenkurth deeplearningtoestimatemodelbiasesinanoperationalnwpassimilationsystem
AT peterdominikdueben deeplearningtoestimatemodelbiasesinanoperationalnwpassimilationsystem
AT davidhall deeplearningtoestimatemodelbiasesinanoperationalnwpassimilationsystem