检索
E-mail
RSS

作者投稿 主编审稿 专家审稿 远程编辑

热带海洋学报

• • 上一篇    下一篇

基于全链路仿真的浅海地形SAR图像仿真研究

张颖瑞1, 马林涛2, 王小青1, 韦骏2

  

  1. 1. 中山大学电子与通信工程学院, 南方海洋科学与工程广东省实验室(珠海), 广东 深圳 518000;


    2. 中山大学大气科学学院, 广东 珠海519082


  • 收稿日期:2026-03-21 修回日期:2026-06-05 接受日期:2026-06-15
  • 通讯作者: 王小青
  • 基金资助:
    南方海洋科学与工程广东省实验室(珠海)自主科研项目(NO.SML2020SP009)

Simulation research on shallow sea topography SAR Images based on full-chain simulation

ZHANG Yingrui1, MA Lintao2, WANG Xiaoqing1, WEI Jun2   

  1. 1.School of Electronics and Communication Engineering, Sun Yat-Sen University, Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai), Shenzhen, 518000, China;


    2. School of Atmospheric Sciences, Sun Yat-Sen University, Zhuhai, 519082, China


  • Received:2026-03-21 Revised:2026-06-05 Accepted:2026-06-15
  • Supported by:
    The Project of Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai) (SML2020SP009)

PDF (PC)

赞

摘要： 浅海水下地形信息在海洋资源开发、海洋环境保障等领域具有重要应用价值。浅海地形造成的流场扰动会在海面上形成辐聚辐散效应，间接改变海面粗糙度分布，进而反映到高分辨率的SAR图像中，这使得基于合成孔径雷达(synthetic aperture radar, SAR)图像进行浅海地形的高分辨率反演成为可能。由于缺乏高分辨率的真实标签样本数据，使得基于深度学习方法进行SAR海底地形反演依赖于仿真手段构建样本。本文基于面向SAR海洋成像的全链路仿真模型，实现复杂浅海地形条件下大规模、高精度仿真数据集的快速生成。研究以江苏省太平沙地区和海南岛东北部海域为典型研究区，对3景不同时刻的仿真与真实SAR影像进行定量对比分析。定量分析结果表明，仿真影像在纹理结构、明暗分布及空间分布上与真实影像具有较高的一致性，太平沙地区和海南岛东北部海域的最高结构相似度分别达到0.62和0.73。通过对仿真结果的整体与局部分析，验证了仿真模型在不同地形动力环境下的物理一致性与泛化能力。所构建的全链路仿真模型能够有效还原水下地形到SAR图像的完整物理过程，实现从水深数据到SAR影像的正向模拟。

关键词: 水下地形反演, 全链路仿真, 波作用量方程, 海面流场调制

Abstract: Information regarding shallow-water bathymetry holds substantial practical significance for marine resource exploitation and environmental conservation. Disturbances in the current field induced by topography give rise to patterns of surface convergence and divergence, which subsequently regulate the distribution of sea-surface roughness. These regulatory effects are then clearly reflected in high-resolution synthetic aperture radar (SAR) imagery. As a result, SAR images offer a viable approach for high-resolution inversion of pre-existing submarine topography. Owing to the scarcity of high-resolution ground-truth data, SAR bathymetry inversion based on deep learning mainly depends on simulated datasets for model training. This study leverages a full- chain SAR ocean imaging simulation model to facilitate the rapid generation of large-scale, high-precision simulated datasets under complex shallow-sea topographic conditions.The Taipingsha area in Jiangsu Province and the northeastern coastal waters of Hainan Island are selected as representative study sites. A comparative analysis is conducted between simulated and real SAR images acquired at three distinct time periods. Quantitative analysis reveals that the simulated images demonstrate a high degree of consistency with the real SAR images in terms of textural structure, brightness distribution, and spatial patterns. Specifically, the maximum structural similarity indices for the Taipingsha area and the northeastern coastal waters of Hainan Island reach 0.62 and 0.73 respectively. Both global and local analyses of the simulation results further validate the physical consistency and generalization ability of the simulation model across various topographic and hydrodynamic conditions. The constructed full-chain simulation model effectively reconstructs the entire physical process from underwater topography to SAR image formation, enabling forward simulation from bathymetric data to SAR imagery.

Key words: Underwater terrain inversion, Full-chain simulation, Wave action equation, Sea surface current field modulation