Sea Ice Motion Estimation Algorithm Using Computer Vision for The Sea Ice Dynamic Experiment (SIDEx) Field Campaign, Alaska, 2021

The Sea Ice Dynamics Experiment (SIDEx) was a field campaign in the Beaufort Sea from February to April 2021. The field experiment was designed to investigate the interaction of ice stress, strain, and fracture over meter-to-kilometer (m-km) spatial scales as ice fractured and subsequently deformed. Observations were collected in situ at an ice camp, using autonomous buoys, and with remote sensing. Observations were collected over a variety of scales, including scales larger than the target m-km scale. This dataset is one facet of the multi-modal data package collected together in this parent archive. This dataset contains horizontal displacement data for Arctic sea ice, derived from 54 RADARSAT-2 ScanSAR images. The images were downsampled to a uniform spatial resolution of 50 meters (m) and cropped to a 40 km² area centered on a drifting buoy located near the main camp. The center buoy's position at the exact time of each image was estimated by linearly interpolating between its recorded locations. Using the Scale-Invariant Feature Transform (SIFT) algorithm for feature detection and matching, followed by iterative Farneback Optical Flow, where dense motion estimation was performed for each pixel. The dataset includes NetCDF files with georeferenced displacement data at the pixel level, enabling detailed analysis and visualization of sea ice dynamics.

海冰动力学实验（Sea Ice Dynamics Experiment, SIDEx）是2021年2月至4月于波弗特海开展的野外科考项目。本次野外实验旨在探究冰体断裂并发生后续形变过程中，冰应力、冰应变与冰断裂在米至千米（m-km）空间尺度下的相互作用机制。研究团队通过冰站原位观测、自主浮标监测与遥感手段获取观测数据，观测涵盖多种尺度范围，其中亦包含超出目标m-km尺度的观测内容。本数据集为该主归档包中整合的多模态数据集的组成部分，包含源自54幅RADARSAT-2扫描合成孔径雷达（ScanSAR）图像的北极海冰水平位移数据。所有图像均被下采样至统一的50米空间分辨率，并被裁剪至以主营地附近的漂移浮标为中心的40平方千米区域。各图像拍摄时刻的中心浮标位置，通过对其已记录的位置进行线性插值得以估算。研究采用尺度不变特征变换（Scale-Invariant Feature Transform, SIFT）算法完成特征检测与匹配，随后采用迭代法尔内克光流算法对每个像素执行稠密运动估计。本数据集包含带有地理参考信息的像素级位移数据NetCDF文件，可用于开展海冰动力学过程的精细化分析与可视化研究。