Deep Learning Regional Climate Model Emulators: A Comparison of Two Downscaling Training Frameworks
Regional climate models (RCMs) have a high computational cost due to their higher spatial resolution compared to global climate models (GCMs). Therefore, various downscaling approaches have been developed as a surrogate for the dynamical downscaling of GCMs. This study assesses the potential of usin...
Main Authors: | , , , |
---|---|
Format: | Article in Journal/Newspaper |
Language: | English |
Published: |
Wiley-Blackwell
2023
|
Subjects: | |
Online Access: | https://hdl.handle.net/20.500.11850/617206 https://doi.org/10.3929/ethz-b-000617206 |
id |
ftethz:oai:www.research-collection.ethz.ch:20.500.11850/617206 |
---|---|
record_format |
openpolar |
spelling |
ftethz:oai:www.research-collection.ethz.ch:20.500.11850/617206 2023-07-16T03:54:09+02:00 Deep Learning Regional Climate Model Emulators: A Comparison of Two Downscaling Training Frameworks van der Meer, Marijn id_orcid:0 000-0002-7604-4494 de Roda Husman, Sophie Lhermitte, Stef 2023 application/application/pdf https://hdl.handle.net/20.500.11850/617206 https://doi.org/10.3929/ethz-b-000617206 en eng Wiley-Blackwell info:eu-repo/semantics/altIdentifier/doi/10.1029/2022MS003593 info:eu-repo/semantics/altIdentifier/wos/001000548800001 http://hdl.handle.net/20.500.11850/617206 doi:10.3929/ethz-b-000617206 info:eu-repo/semantics/openAccess http://creativecommons.org/licenses/by/4.0/ Creative Commons Attribution 4.0 International Journal of Advances in Modeling Earth Systems, 15 (6) machine learning RCM-emulator GCM downscaling Antarctica info:eu-repo/semantics/article info:eu-repo/semantics/publishedVersion 2023 ftethz https://doi.org/20.500.11850/61720610.3929/ethz-b-00061720610.1029/2022MS003593 2023-06-25T23:48:59Z Regional climate models (RCMs) have a high computational cost due to their higher spatial resolution compared to global climate models (GCMs). Therefore, various downscaling approaches have been developed as a surrogate for the dynamical downscaling of GCMs. This study assesses the potential of using a cost-efficient machine learning alternative to dynamical downscaling by using the example case study of emulating surface mass balance (SMB) over the Antarctic Peninsula. More specifically, we determine the impact of the training framework by comparing two training scenarios: (a) a perfect and (b) an imperfect model framework. In the perfect model framework, the RCM-emulator learns only the downscaling function; therefore, it was trained with upscaled RCM (UPRCM) features at GCM resolution. This emulator accurately reproduced SMB when evaluated on UPRCM, but its predictions on GCM data conserved RCM-GCM inconsistencies and led to underestimation. In the imperfect model framework, the RCM-emulator was trained with GCM features and downscaled the GCM while exposed to RCM-GCM inconsistencies. This emulator predicted SMB close to the truth, showing it learned the underlying inconsistencies and dynamics. Our results suggest that a deep learning RCM-emulator can learn the proper GCM to RCM downscaling function while working directly with GCM data. Furthermore, the RCM-emulator presents a significant computational gain compared to an RCM simulation. We conclude that machine learning emulators can be applied to produce fast and fine-scaled predictions of RCM simulations from GCM data. ISSN:1942-2466 Article in Journal/Newspaper Antarc* Antarctic Antarctic Peninsula Antarctica ETH Zürich Research Collection Antarctic Antarctic Peninsula The Antarctic |
institution |
Open Polar |
collection |
ETH Zürich Research Collection |
op_collection_id |
ftethz |
language |
English |
topic |
machine learning RCM-emulator GCM downscaling Antarctica |
spellingShingle |
machine learning RCM-emulator GCM downscaling Antarctica van der Meer, Marijn id_orcid:0 000-0002-7604-4494 de Roda Husman, Sophie Lhermitte, Stef Deep Learning Regional Climate Model Emulators: A Comparison of Two Downscaling Training Frameworks |
topic_facet |
machine learning RCM-emulator GCM downscaling Antarctica |
description |
Regional climate models (RCMs) have a high computational cost due to their higher spatial resolution compared to global climate models (GCMs). Therefore, various downscaling approaches have been developed as a surrogate for the dynamical downscaling of GCMs. This study assesses the potential of using a cost-efficient machine learning alternative to dynamical downscaling by using the example case study of emulating surface mass balance (SMB) over the Antarctic Peninsula. More specifically, we determine the impact of the training framework by comparing two training scenarios: (a) a perfect and (b) an imperfect model framework. In the perfect model framework, the RCM-emulator learns only the downscaling function; therefore, it was trained with upscaled RCM (UPRCM) features at GCM resolution. This emulator accurately reproduced SMB when evaluated on UPRCM, but its predictions on GCM data conserved RCM-GCM inconsistencies and led to underestimation. In the imperfect model framework, the RCM-emulator was trained with GCM features and downscaled the GCM while exposed to RCM-GCM inconsistencies. This emulator predicted SMB close to the truth, showing it learned the underlying inconsistencies and dynamics. Our results suggest that a deep learning RCM-emulator can learn the proper GCM to RCM downscaling function while working directly with GCM data. Furthermore, the RCM-emulator presents a significant computational gain compared to an RCM simulation. We conclude that machine learning emulators can be applied to produce fast and fine-scaled predictions of RCM simulations from GCM data. ISSN:1942-2466 |
format |
Article in Journal/Newspaper |
author |
van der Meer, Marijn id_orcid:0 000-0002-7604-4494 de Roda Husman, Sophie Lhermitte, Stef |
author_facet |
van der Meer, Marijn id_orcid:0 000-0002-7604-4494 de Roda Husman, Sophie Lhermitte, Stef |
author_sort |
van der Meer, Marijn |
title |
Deep Learning Regional Climate Model Emulators: A Comparison of Two Downscaling Training Frameworks |
title_short |
Deep Learning Regional Climate Model Emulators: A Comparison of Two Downscaling Training Frameworks |
title_full |
Deep Learning Regional Climate Model Emulators: A Comparison of Two Downscaling Training Frameworks |
title_fullStr |
Deep Learning Regional Climate Model Emulators: A Comparison of Two Downscaling Training Frameworks |
title_full_unstemmed |
Deep Learning Regional Climate Model Emulators: A Comparison of Two Downscaling Training Frameworks |
title_sort |
deep learning regional climate model emulators: a comparison of two downscaling training frameworks |
publisher |
Wiley-Blackwell |
publishDate |
2023 |
url |
https://hdl.handle.net/20.500.11850/617206 https://doi.org/10.3929/ethz-b-000617206 |
geographic |
Antarctic Antarctic Peninsula The Antarctic |
geographic_facet |
Antarctic Antarctic Peninsula The Antarctic |
genre |
Antarc* Antarctic Antarctic Peninsula Antarctica |
genre_facet |
Antarc* Antarctic Antarctic Peninsula Antarctica |
op_source |
Journal of Advances in Modeling Earth Systems, 15 (6) |
op_relation |
info:eu-repo/semantics/altIdentifier/doi/10.1029/2022MS003593 info:eu-repo/semantics/altIdentifier/wos/001000548800001 http://hdl.handle.net/20.500.11850/617206 doi:10.3929/ethz-b-000617206 |
op_rights |
info:eu-repo/semantics/openAccess http://creativecommons.org/licenses/by/4.0/ Creative Commons Attribution 4.0 International |
op_doi |
https://doi.org/20.500.11850/61720610.3929/ethz-b-00061720610.1029/2022MS003593 |
_version_ |
1771551364451663872 |