Scholarlink

<p>Sea ice drift (SID) plays a critical role in regulating the energy and mass balance of Arctic sea ice. Existing multi-decadal SID products with spatial resolutions of dozens of kilometres identified a positive trend in Arctic SID speed, highlighting increasing sea ice mobility. We present the Sea Ice Drift Computer Vision (SID-CV) dataset, a pan-Arctic product derived from Sentinel-1A&amp;B synthetic aperture radar (SAR) data spanning 2017–2021, offering an unprecedented 400-meter grid spacing. Developed using a computer vision framework combining feature tracking and pattern matching, the dataset comprises over 350k products from sequential SAR image pairs within 0–36-hour intervals. Each NetCDF file contains drift vectors, acquisition metadata, and quality flags based on algorithmic confidence metrics. Retrievals at different quality levels were systematically evaluated. Validation against drifting buoys demonstrates high accuracy, with low errors in drift speed (bias −0.02 cm/s, RMSE 0.20 cm/s), distance&#xa0;(bias −19.81 m, RMSE 198.71 m), and direction (bias 0.48 °,&#xa0;RMSE 2.94°)&#xa0;across seasons and ice regimes. Compared to existing products, SID-CV dataset offers improved spatial detail and reliability, enabling refined analyses of sea ice export, deformation and model validation.</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

Pan-Arctic sea ice drift data at 400 m grids based on spaceborne SAR

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年2月7日

查看全文
  • Yujia Qiu,
  • Xiao-Ming Li

摘要

Sea ice drift (SID) plays a critical role in regulating the energy and mass balance of Arctic sea ice. Existing multi-decadal SID products with spatial resolutions of dozens of kilometres identified a positive trend in Arctic SID speed, highlighting increasing sea ice mobility. We present the Sea Ice Drift Computer Vision (SID-CV) dataset, a pan-Arctic product derived from Sentinel-1A&B synthetic aperture radar (SAR) data spanning 2017–2021, offering an unprecedented 400-meter grid spacing. Developed using a computer vision framework combining feature tracking and pattern matching, the dataset comprises over 350k products from sequential SAR image pairs within 0–36-hour intervals. Each NetCDF file contains drift vectors, acquisition metadata, and quality flags based on algorithmic confidence metrics. Retrievals at different quality levels were systematically evaluated. Validation against drifting buoys demonstrates high accuracy, with low errors in drift speed (bias −0.02 cm/s, RMSE 0.20 cm/s), distance (bias −19.81 m, RMSE 198.71 m), and direction (bias 0.48 °, RMSE 2.94°) across seasons and ice regimes. Compared to existing products, SID-CV dataset offers improved spatial detail and reliability, enabling refined analyses of sea ice export, deformation and model validation.