Arctic sea ice cover product based on spaceborne synthetic aperture radar

The Sentinel-1 SAR based sea ice cover data in the Arctic. Based on the U-Net, one kind of architecture of deep learning, we developed a novel method of deriving sea ice cover information from spaceborne SAR data in high spatial resolution. Then we produced the relevant data product based on the ESA Sentinel-1 SAR data acquired in the Arctic using the proposed method. The Sentinel-1A and 1B SAR satellites have been extensively acquired the so-called extra-wide (EW) swath mode data in the polar regions since their launch in 2014 and 2016. The EW data have a wide coverage of approximately 400 by 400 km and a pixel size of 40 by 40 m. The twins can acquired SAR data by approximately 2500 scenes each month in the Arctic. Based on the above-mentioned method, we processed each S1 SAR image acquired in the Arctic and generated the sea ice cover data, which are in the standard CF 1.7 convention of NetCDF. Each S1 image corresponds to a NC file. The spatial resolution of the SAR-derived sea ice cover is 400 m. We have finished processing the S1 data acquired through the year 2019 to 2021 in the Arctic and upload the corresponding sea ice cover data. We are continuing to process the historical data and ongoing acquired data. Therefore, the published data will be updated often.

本数据集为基于哨兵1号（Sentinel-1）合成孔径雷达（Synthetic Aperture Radar, SAR）的北极海冰覆盖数据。本研究基于深度学习架构U-Net，开发了一种可从星载SAR数据中提取高空间分辨率海冰覆盖信息的全新方法；随后利用该方法，基于欧洲空间局（European Space Agency, ESA）在北极区域获取的Sentinel-1 SAR数据，生成了对应的海冰覆盖数据产品。哨兵1A与1B卫星自2014年、2016年发射以来，已在极地区域广泛采集所谓的超宽幅（EW）条带模式数据。该类超宽幅数据的覆盖范围约为400×400千米，像素尺寸为40×40米；这两颗双星平台每月可在北极区域采集约2500景SAR数据。基于前述方法，我们对北极区域获取的每景S1 SAR图像进行处理，生成了符合NetCDF标准CF 1.7规范的海冰覆盖数据，每景S1图像对应一个NC格式文件。该SAR反演得到的海冰覆盖数据空间分辨率为400米。目前我们已完成2019年至2021年北极区域S1数据的处理工作，并上传了对应的海冰覆盖数据；当前仍在持续处理历史数据及最新采集的在轨数据，因此已发布的数据集将持续更新。