Scholarlink

<p>Seamounts are distinctive features of seafloor topography. Most seafloor studies rely on the gravity-geologic method (GGM) and other fitting algorithms, often overlooking the impact of complex seamount topography on boundary regions. This study employs forward modeling of the vertical gravity gradient (VGG) and gravity anomaly (GA) to mitigate seamount effects on boundary areas, followed by bathymetric inversion via a fully connected deep neural network (FC-DNN). The GGM-based boundary bathymetry inversion shows increasing errors with increasing distance from the seamount center. To improve accuracy, the remove-and-restore method is combined with the GGM, which demonstrates that the GA outperforms the VGG for seamount boundary inversion. For a 10&#xa0;km boundary, GA inversion achieves accuracies of 23.94&#xa0;m, 28.62&#xa0;m, 31.31&#xa0;m, and 17.7&#xa0;m across four seamounts, and the accuracy improved for a 20&#xa0;km boundary to 25.81&#xa0;m, 25.39&#xa0;m, 25.59&#xa0;m, and 14&#xa0;m. A comparison with the GEBCO bathymetric model shows that despite variations in performance under different seamount complexities, over 70% of the inverted bathymetric points have errors within 30&#xa0;m, with fewer than 30% exceeding this threshold.</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

Seamount topography study based on the remove–restore method

  • 期刊论文

  • 发表日期：

    2025年6月27日

查看全文
  • Jinyang Wang,
  • Huan Xu,
  • Jianbo Wang,
  • Lina Lin,
  • Na Liu

摘要

Seamounts are distinctive features of seafloor topography. Most seafloor studies rely on the gravity-geologic method (GGM) and other fitting algorithms, often overlooking the impact of complex seamount topography on boundary regions. This study employs forward modeling of the vertical gravity gradient (VGG) and gravity anomaly (GA) to mitigate seamount effects on boundary areas, followed by bathymetric inversion via a fully connected deep neural network (FC-DNN). The GGM-based boundary bathymetry inversion shows increasing errors with increasing distance from the seamount center. To improve accuracy, the remove-and-restore method is combined with the GGM, which demonstrates that the GA outperforms the VGG for seamount boundary inversion. For a 10 km boundary, GA inversion achieves accuracies of 23.94 m, 28.62 m, 31.31 m, and 17.7 m across four seamounts, and the accuracy improved for a 20 km boundary to 25.81 m, 25.39 m, 25.59 m, and 14 m. A comparison with the GEBCO bathymetric model shows that despite variations in performance under different seamount complexities, over 70% of the inverted bathymetric points have errors within 30 m, with fewer than 30% exceeding this threshold.