Scholarlink

<p>Accurate and efficient prediction of ocean waves and storm surges induced by tropical cyclones is essential for coastal hazard mitigation and maritime safety. While traditional numerical models are reliable, they require substantial computational resources. Deep learning (DL) offers an alternative approach for modeling these phenomena. Although existing DL models typically address ocean waves or storm surges separately, both phenomena represent responses to atmospheric forcing, suggesting that they can be modeled using identical underlying principles within a data-driven framework. This study, guided by the physics of ocean wave and storm surge generation and evolution, presents a unified DL model that simultaneously predicts both phenomena using current and historical wind and sea level pressure field data. The model effectively captures the complex nonlinear relationships between meteorological inputs and hydrodynamic outputs with high accuracy. Results demonstrate that the proposed model accurately predicts both significant wave height (SWH) and storm surge height, showing strong correlation with numerical models and validating our hypothesis that these phenomena can be modeled collectively using identical input parameters. The model demonstrates superior computational efficiency compared with traditional numerical models while maintaining high accuracy, making it particularly suitable for real-time operational forecasting and climate research of ocean waves and storm surges.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Physics-Guided Deep Learning Model for Ocean Waves and Storm Surges Induced by Tropical Cyclones with Unified Underlying Logic

  • 期刊论文

  • 发表日期：

    2026年2月16日

查看全文
  • Ao Li,
  • Cheng-cheng Qian,
  • Hao-yu Jiang,
  • Xing-ru Feng,
  • Wen-fang Lu,
  • Lin Mu,
  • Chang-ming Dong

摘要

Accurate and efficient prediction of ocean waves and storm surges induced by tropical cyclones is essential for coastal hazard mitigation and maritime safety. While traditional numerical models are reliable, they require substantial computational resources. Deep learning (DL) offers an alternative approach for modeling these phenomena. Although existing DL models typically address ocean waves or storm surges separately, both phenomena represent responses to atmospheric forcing, suggesting that they can be modeled using identical underlying principles within a data-driven framework. This study, guided by the physics of ocean wave and storm surge generation and evolution, presents a unified DL model that simultaneously predicts both phenomena using current and historical wind and sea level pressure field data. The model effectively captures the complex nonlinear relationships between meteorological inputs and hydrodynamic outputs with high accuracy. Results demonstrate that the proposed model accurately predicts both significant wave height (SWH) and storm surge height, showing strong correlation with numerical models and validating our hypothesis that these phenomena can be modeled collectively using identical input parameters. The model demonstrates superior computational efficiency compared with traditional numerical models while maintaining high accuracy, making it particularly suitable for real-time operational forecasting and climate research of ocean waves and storm surges.