A Lightweight Terrain‐Constraint Model for Wind Spatial Downscaling

Abstract High‐resolution wind fields has always been the goal of refined meteorological forecasting. Using advanced deep learning algorithms for wind downscaling is an effective approach to achieve this goal. However, the lack of physical process understanding in deep learning algorithms results in...

Full description

Saved in:
Bibliographic Details
Main Authors: Anboyu Guo, Qiyang Liu, Jiuke Wang, Yuanyong Gao, Lixin Wei, Xiaojiang Song, Chi Zhang, Kai Liu, Shuping Ma, Jiu Jiang
Format: Article
Language:English
Published: Wiley 2025-06-01
Series:Journal of Geophysical Research: Machine Learning and Computation
Subjects:
Online Access:https://doi.org/10.1029/2024JH000147
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849432434313003008
author Anboyu Guo
Qiyang Liu
Jiuke Wang
Yuanyong Gao
Lixin Wei
Xiaojiang Song
Chi Zhang
Kai Liu
Shuping Ma
Jiu Jiang
author_facet Anboyu Guo
Qiyang Liu
Jiuke Wang
Yuanyong Gao
Lixin Wei
Xiaojiang Song
Chi Zhang
Kai Liu
Shuping Ma
Jiu Jiang
author_sort Anboyu Guo
collection DOAJ
description Abstract High‐resolution wind fields has always been the goal of refined meteorological forecasting. Using advanced deep learning algorithms for wind downscaling is an effective approach to achieve this goal. However, the lack of physical process understanding in deep learning algorithms results in the inability to accurately reconstruct fine‐scale structures after downscaling. In this study, we propose a Terrain‐Constraint Wind Downscaling Model (TCWDM), a lightweight deep learning model consisting of a downscaling module and a terrain‐constraint module. By combining low‐resolution wind with high‐resolution terrain data, the model achieves a tenfold downscaling of spatial wind fields and reconstructs the detailed structure of the wind. Due to the incorporation of an attention mechanism, multi‐feature inputs, and the terrain‐constraint module, TCWDM demonstrates superior downscaling performance. Compared to traditional interpolation methods and other deep learning models, the mean absolute error is reduced by up to 49%. The terrain‐constraint module, in particular, contributes most significantly to the model's performance, especially in complex terrains, where it enables greater optimization of downscaling results. Furthermore, due to the lightweight model structure and a specific fine‐tuning strategy, TCWDM can deliver significantly better downscaling results at a lower cost across different regions, offering potential for broader applications.
format Article
id doaj-art-5bb09888e22b4191a93163288e7e43b3
institution Kabale University
issn 2993-5210
language English
publishDate 2025-06-01
publisher Wiley
record_format Article
series Journal of Geophysical Research: Machine Learning and Computation
spelling doaj-art-5bb09888e22b4191a93163288e7e43b32025-08-20T03:27:22ZengWileyJournal of Geophysical Research: Machine Learning and Computation2993-52102025-06-0122n/an/a10.1029/2024JH000147A Lightweight Terrain‐Constraint Model for Wind Spatial DownscalingAnboyu Guo0Qiyang Liu1Jiuke Wang2Yuanyong Gao3Lixin Wei4Xiaojiang Song5Chi Zhang6Kai Liu7Shuping Ma8Jiu Jiang9National Marine Environment Forecasting Center Beijing ChinaEast China Electric Power Design Institute of China Power Engineering Consulting Group Corporation Shanghai ChinaSchool of Artificial Intelligence Sun Yat sen University Zhuhai ChinaNational Marine Environment Forecasting Center Beijing ChinaNational Marine Environment Forecasting Center Beijing ChinaNational Marine Environment Forecasting Center Beijing ChinaNational Marine Environment Forecasting Center Beijing ChinaNational Marine Environment Forecasting Center Beijing ChinaDepartment of Civil and Environmental Engineering National University of Singapore Singapore SingaporeCollege of Geography and Remote Sensing Sciences Xinjiang University Urumqi ChinaAbstract High‐resolution wind fields has always been the goal of refined meteorological forecasting. Using advanced deep learning algorithms for wind downscaling is an effective approach to achieve this goal. However, the lack of physical process understanding in deep learning algorithms results in the inability to accurately reconstruct fine‐scale structures after downscaling. In this study, we propose a Terrain‐Constraint Wind Downscaling Model (TCWDM), a lightweight deep learning model consisting of a downscaling module and a terrain‐constraint module. By combining low‐resolution wind with high‐resolution terrain data, the model achieves a tenfold downscaling of spatial wind fields and reconstructs the detailed structure of the wind. Due to the incorporation of an attention mechanism, multi‐feature inputs, and the terrain‐constraint module, TCWDM demonstrates superior downscaling performance. Compared to traditional interpolation methods and other deep learning models, the mean absolute error is reduced by up to 49%. The terrain‐constraint module, in particular, contributes most significantly to the model's performance, especially in complex terrains, where it enables greater optimization of downscaling results. Furthermore, due to the lightweight model structure and a specific fine‐tuning strategy, TCWDM can deliver significantly better downscaling results at a lower cost across different regions, offering potential for broader applications.https://doi.org/10.1029/2024JH000147wind field downscalinglightweight modelcomplex terrainterrain‐constraint
spellingShingle Anboyu Guo
Qiyang Liu
Jiuke Wang
Yuanyong Gao
Lixin Wei
Xiaojiang Song
Chi Zhang
Kai Liu
Shuping Ma
Jiu Jiang
A Lightweight Terrain‐Constraint Model for Wind Spatial Downscaling
Journal of Geophysical Research: Machine Learning and Computation
wind field downscaling
lightweight model
complex terrain
terrain‐constraint
title A Lightweight Terrain‐Constraint Model for Wind Spatial Downscaling
title_full A Lightweight Terrain‐Constraint Model for Wind Spatial Downscaling
title_fullStr A Lightweight Terrain‐Constraint Model for Wind Spatial Downscaling
title_full_unstemmed A Lightweight Terrain‐Constraint Model for Wind Spatial Downscaling
title_short A Lightweight Terrain‐Constraint Model for Wind Spatial Downscaling
title_sort lightweight terrain constraint model for wind spatial downscaling
topic wind field downscaling
lightweight model
complex terrain
terrain‐constraint
url https://doi.org/10.1029/2024JH000147
work_keys_str_mv AT anboyuguo alightweightterrainconstraintmodelforwindspatialdownscaling
AT qiyangliu alightweightterrainconstraintmodelforwindspatialdownscaling
AT jiukewang alightweightterrainconstraintmodelforwindspatialdownscaling
AT yuanyonggao alightweightterrainconstraintmodelforwindspatialdownscaling
AT lixinwei alightweightterrainconstraintmodelforwindspatialdownscaling
AT xiaojiangsong alightweightterrainconstraintmodelforwindspatialdownscaling
AT chizhang alightweightterrainconstraintmodelforwindspatialdownscaling
AT kailiu alightweightterrainconstraintmodelforwindspatialdownscaling
AT shupingma alightweightterrainconstraintmodelforwindspatialdownscaling
AT jiujiang alightweightterrainconstraintmodelforwindspatialdownscaling
AT anboyuguo lightweightterrainconstraintmodelforwindspatialdownscaling
AT qiyangliu lightweightterrainconstraintmodelforwindspatialdownscaling
AT jiukewang lightweightterrainconstraintmodelforwindspatialdownscaling
AT yuanyonggao lightweightterrainconstraintmodelforwindspatialdownscaling
AT lixinwei lightweightterrainconstraintmodelforwindspatialdownscaling
AT xiaojiangsong lightweightterrainconstraintmodelforwindspatialdownscaling
AT chizhang lightweightterrainconstraintmodelforwindspatialdownscaling
AT kailiu lightweightterrainconstraintmodelforwindspatialdownscaling
AT shupingma lightweightterrainconstraintmodelforwindspatialdownscaling
AT jiujiang lightweightterrainconstraintmodelforwindspatialdownscaling