Data-driven rolling model for global wave height

<p>Significant wave height (SWH) is crucial for many human activities, such as marine navigation, offshore operations, and coastal management. Traditionally, SWH is modeled using numerical wave models, which, while accurate, are computationally intensive and constrained by incomplete physical...

Full description

Saved in:
Bibliographic Details
Main Authors: X. Wang, J. Wang, W. Lu, C. Dong, H. Qin, H. Jiang
Format: Article
Language:English
Published: Copernicus Publications 2025-08-01
Series:Geoscientific Model Development
Online Access:https://gmd.copernicus.org/articles/18/5101/2025/gmd-18-5101-2025.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849329582910472192
author X. Wang
X. Wang
X. Wang
J. Wang
W. Lu
C. Dong
H. Qin
H. Jiang
H. Jiang
H. Jiang
author_facet X. Wang
X. Wang
X. Wang
J. Wang
W. Lu
C. Dong
H. Qin
H. Jiang
H. Jiang
H. Jiang
author_sort X. Wang
collection DOAJ
description <p>Significant wave height (SWH) is crucial for many human activities, such as marine navigation, offshore operations, and coastal management. Traditionally, SWH is modeled using numerical wave models, which, while accurate, are computationally intensive and constrained by incomplete physical representations of wave spectral evolution. This study introduces a simple global deep-learning-based model for SWH, which uses the current SWH field and the wind field at the next time step as inputs to predict the SWH field at the next time step. This approach mirrors the rolling prediction strategy of numerical wave models. After training on a reanalysis dataset, the errors of the model accumulate lightly with time when given a good initial field because no spectral information is used. However, after accumulating for <span class="inline-formula">∼</span> 200 h, the errors stabilize, remaining comparable to those of state-of-the-art numerical wave models. Additionally, the error accumulation can be mitigated through the assimilation of altimeter measurements. This deep learning model can not only serve as an efficient surrogate for traditional numerical wave models with respect to SWH but also provide a baseline for statistical modeling of global SWH due to its simplicity in inputs and outputs.</p>
format Article
id doaj-art-eef44d49be7e4896b6b1cde15628f0f1
institution Kabale University
issn 1991-959X
1991-9603
language English
publishDate 2025-08-01
publisher Copernicus Publications
record_format Article
series Geoscientific Model Development
spelling doaj-art-eef44d49be7e4896b6b1cde15628f0f12025-08-20T03:47:13ZengCopernicus PublicationsGeoscientific Model Development1991-959X1991-96032025-08-01185101511410.5194/gmd-18-5101-2025Data-driven rolling model for global wave heightX. Wang0X. Wang1X. Wang2J. Wang3W. Lu4C. Dong5H. Qin6H. Jiang7H. Jiang8H. Jiang9College of Life Sciences and Oceanography, Shenzhen University, Shenzhen, ChinaLaboratory for Regional Oceanography and Numerical Modeling, Qingdao Marine Science and Technology Center, Qingdao, ChinaShenzhen Research Institute, China University of Geosciences, Shenzhen, ChinaSchool of Artificial Intelligence, Sun Yat-Sen University, Zhuhai, ChinaSchool of Marine Sciences, Sun Yat-Sen University, Zhuhai, ChinaSchool of Marine Sciences, Nanjing University of Information Science & Technology, Nanjing, ChinaShenzhen Research Institute, China University of Geosciences, Shenzhen, ChinaCollege of Life Sciences and Oceanography, Shenzhen University, Shenzhen, ChinaLaboratory for Regional Oceanography and Numerical Modeling, Qingdao Marine Science and Technology Center, Qingdao, ChinaShenzhen Research Institute, China University of Geosciences, Shenzhen, China<p>Significant wave height (SWH) is crucial for many human activities, such as marine navigation, offshore operations, and coastal management. Traditionally, SWH is modeled using numerical wave models, which, while accurate, are computationally intensive and constrained by incomplete physical representations of wave spectral evolution. This study introduces a simple global deep-learning-based model for SWH, which uses the current SWH field and the wind field at the next time step as inputs to predict the SWH field at the next time step. This approach mirrors the rolling prediction strategy of numerical wave models. After training on a reanalysis dataset, the errors of the model accumulate lightly with time when given a good initial field because no spectral information is used. However, after accumulating for <span class="inline-formula">∼</span> 200 h, the errors stabilize, remaining comparable to those of state-of-the-art numerical wave models. Additionally, the error accumulation can be mitigated through the assimilation of altimeter measurements. This deep learning model can not only serve as an efficient surrogate for traditional numerical wave models with respect to SWH but also provide a baseline for statistical modeling of global SWH due to its simplicity in inputs and outputs.</p>https://gmd.copernicus.org/articles/18/5101/2025/gmd-18-5101-2025.pdf
spellingShingle X. Wang
X. Wang
X. Wang
J. Wang
W. Lu
C. Dong
H. Qin
H. Jiang
H. Jiang
H. Jiang
Data-driven rolling model for global wave height
Geoscientific Model Development
title Data-driven rolling model for global wave height
title_full Data-driven rolling model for global wave height
title_fullStr Data-driven rolling model for global wave height
title_full_unstemmed Data-driven rolling model for global wave height
title_short Data-driven rolling model for global wave height
title_sort data driven rolling model for global wave height
url https://gmd.copernicus.org/articles/18/5101/2025/gmd-18-5101-2025.pdf
work_keys_str_mv AT xwang datadrivenrollingmodelforglobalwaveheight
AT xwang datadrivenrollingmodelforglobalwaveheight
AT xwang datadrivenrollingmodelforglobalwaveheight
AT jwang datadrivenrollingmodelforglobalwaveheight
AT wlu datadrivenrollingmodelforglobalwaveheight
AT cdong datadrivenrollingmodelforglobalwaveheight
AT hqin datadrivenrollingmodelforglobalwaveheight
AT hjiang datadrivenrollingmodelforglobalwaveheight
AT hjiang datadrivenrollingmodelforglobalwaveheight
AT hjiang datadrivenrollingmodelforglobalwaveheight