A two-fold deep-learning strategy to correct and downscale winds over mountains
<p>Assessing wind fields at a local scale in mountainous terrain has long been a scientific challenge, partly because of the complex interaction between large-scale flows and local topography. Traditionally, the operational applications that require high-resolution wind forcings rely on downsc...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Copernicus Publications
2024-02-01
|
| Series: | Nonlinear Processes in Geophysics |
| Online Access: | https://npg.copernicus.org/articles/31/75/2024/npg-31-75-2024.pdf |
| _version_ | 1797315574149152768 |
|---|---|
| author | L. Le Toumelin I. Gouttevin C. Galiez N. Helbig N. Helbig |
| author_facet | L. Le Toumelin I. Gouttevin C. Galiez N. Helbig N. Helbig |
| author_sort | L. Le Toumelin |
| collection | DOAJ |
| description | <p>Assessing wind fields at a local scale in mountainous terrain has long been a scientific challenge, partly because of the complex interaction between large-scale flows and local topography. Traditionally, the operational applications that require high-resolution wind forcings rely on downscaled outputs of numerical weather prediction systems. Downscaling models either proceed from a function that links large-scale wind fields to local observations (hence including a corrective step) or use operations that account for local-scale processes, through statistics or dynamical simulations and without prior knowledge of large-scale modeling errors. This work presents a strategy to first correct and then downscale the wind fields of the numerical weather prediction model AROME (Application of Research to Operations at Mesoscale) operating at 1300 m grid spacing by using a modular architecture composed of two artificial neural networks and the DEVINE downscaling model. We show that our method is able to first correct the wind direction and speed from the large-scale model (1300 m) and then accurately downscale it to a local scale (30 m) by using the DEVINE downscaling model. The innovative aspect of our method lies in its optimization scheme that accounts for the downscaling step in the computations of the corrections of the coarse-scale wind fields. This modular architecture yields competitive results without suppressing the versatility of the DEVINE downscaling model, which remains unbounded to any wind observations.</p> |
| first_indexed | 2024-03-08T03:05:43Z |
| format | Article |
| id | doaj.art-3dc1ea2f38ca4ed4a6f70d5b8cb84ef6 |
| institution | Directory Open Access Journal |
| issn | 1023-5809 1607-7946 |
| language | English |
| last_indexed | 2024-03-08T03:05:43Z |
| publishDate | 2024-02-01 |
| publisher | Copernicus Publications |
| record_format | Article |
| series | Nonlinear Processes in Geophysics |
| spelling | doaj.art-3dc1ea2f38ca4ed4a6f70d5b8cb84ef62024-02-13T09:06:09ZengCopernicus PublicationsNonlinear Processes in Geophysics1023-58091607-79462024-02-0131759710.5194/npg-31-75-2024A two-fold deep-learning strategy to correct and downscale winds over mountainsL. Le Toumelin0I. Gouttevin1C. Galiez2N. Helbig3N. Helbig4Univ. Grenoble Alpes, Université de Toulouse, Météo-France, CNRS, CNRM, Centre d'Études de la Neige, Grenoble, FranceUniv. Grenoble Alpes, Université de Toulouse, Météo-France, CNRS, CNRM, Centre d'Études de la Neige, Grenoble, FranceUniv. Grenoble Alpes, CNRS, Grenoble INP (Institute of Engineering Univ. Grenoble Alpes), LJK, 38000 Grenoble, FranceWSL Institute for Snow and Avalanche Research SLF, Davos, SwitzerlandEastern Switzerland University of Applied Sciences, Rapperswil, Switzerland<p>Assessing wind fields at a local scale in mountainous terrain has long been a scientific challenge, partly because of the complex interaction between large-scale flows and local topography. Traditionally, the operational applications that require high-resolution wind forcings rely on downscaled outputs of numerical weather prediction systems. Downscaling models either proceed from a function that links large-scale wind fields to local observations (hence including a corrective step) or use operations that account for local-scale processes, through statistics or dynamical simulations and without prior knowledge of large-scale modeling errors. This work presents a strategy to first correct and then downscale the wind fields of the numerical weather prediction model AROME (Application of Research to Operations at Mesoscale) operating at 1300 m grid spacing by using a modular architecture composed of two artificial neural networks and the DEVINE downscaling model. We show that our method is able to first correct the wind direction and speed from the large-scale model (1300 m) and then accurately downscale it to a local scale (30 m) by using the DEVINE downscaling model. The innovative aspect of our method lies in its optimization scheme that accounts for the downscaling step in the computations of the corrections of the coarse-scale wind fields. This modular architecture yields competitive results without suppressing the versatility of the DEVINE downscaling model, which remains unbounded to any wind observations.</p>https://npg.copernicus.org/articles/31/75/2024/npg-31-75-2024.pdf |
| spellingShingle | L. Le Toumelin I. Gouttevin C. Galiez N. Helbig N. Helbig A two-fold deep-learning strategy to correct and downscale winds over mountains Nonlinear Processes in Geophysics |
| title | A two-fold deep-learning strategy to correct and downscale winds over mountains |
| title_full | A two-fold deep-learning strategy to correct and downscale winds over mountains |
| title_fullStr | A two-fold deep-learning strategy to correct and downscale winds over mountains |
| title_full_unstemmed | A two-fold deep-learning strategy to correct and downscale winds over mountains |
| title_short | A two-fold deep-learning strategy to correct and downscale winds over mountains |
| title_sort | two fold deep learning strategy to correct and downscale winds over mountains |
| url | https://npg.copernicus.org/articles/31/75/2024/npg-31-75-2024.pdf |
| work_keys_str_mv | AT lletoumelin atwofolddeeplearningstrategytocorrectanddownscalewindsovermountains AT igouttevin atwofolddeeplearningstrategytocorrectanddownscalewindsovermountains AT cgaliez atwofolddeeplearningstrategytocorrectanddownscalewindsovermountains AT nhelbig atwofolddeeplearningstrategytocorrectanddownscalewindsovermountains AT nhelbig atwofolddeeplearningstrategytocorrectanddownscalewindsovermountains AT lletoumelin twofolddeeplearningstrategytocorrectanddownscalewindsovermountains AT igouttevin twofolddeeplearningstrategytocorrectanddownscalewindsovermountains AT cgaliez twofolddeeplearningstrategytocorrectanddownscalewindsovermountains AT nhelbig twofolddeeplearningstrategytocorrectanddownscalewindsovermountains AT nhelbig twofolddeeplearningstrategytocorrectanddownscalewindsovermountains |