Ice front positions for Greenland glaciers (2002–2021): a spatially extensive seasonal record and benchmark dataset for algorithm validation

Variations in Greenland glacier ice-front positions reflect the magnitude and trends of glacier advance and retreat and are critical for understanding ice-sheet dynamics. This dataset provides a spatially extensive and seasonally targeted record of glacier terminus positions across Greenland. It includes 19,171 ice-front delineations for multiple glacier types, including marine-, lake-, and land-terminating glaciers, spanning the period 2002–2021. The delineations were derived from multi-source satellite imagery, including Landsat 5/7/8, Sentinel-1/2, MODIS, ENVISAT ASAR, ASTER, and ERS SAR. All calving front positions were digitized using standardized workflows implemented in GEEDiT (Lea, 2018), ENVI, ArcGIS, and QGIS, and are accompanied by detailed metadata. Data quality was assessed through comparisons with TermPicks (Goliber et al., 2022), the largest compilation of manually interpreted glacier terminus traces across diverse sensors and glacier conditions, and AutoTerm (Zhang et al., 2023), an automated product with denser temporal coverage. Mean average minimum distances (AMD) of 86 m relative to TermPicks and 115 m relative to AutoTerm indicate high position precision. The dataset is distributed in GeoPackage format (total size: 26.5 MB) and includes glacier-specific centrelines for analysis. It provides a spatially extensive, high-quality benchmark for investigating Greenland glacier dynamics, refining time-varying boundary conditions in ice-sheet models, and supporting the development and validation of automated glacier front-tracking algorithms.格陵兰冰川冰缘位置的变化，反映了冰川进退的幅度与趋势，对于理解冰盖动力学机制至关重要。本数据集提供了覆盖格陵兰全域、具备季节针对性的冰川末端位置记录，包含2002年至2021年间针对海洋终止型、湖泊终止型与陆地终止型等多种冰川类型的19171处冰缘勾画数据，该数据源自Landsat 5/7/8、Sentinel-1/2、MODIS、ENVISAT ASAR、ASTER以及ERS SAR等多源卫星影像。 所有冰崩前沿位置均通过GEEDiT（Lea, 2018）、ENVI、ArcGIS与QGIS中实现的标准化工作流完成数字化，并附带详细元数据。数据质量通过与TermPicks（Goliber等，2022）及AutoTerm（Zhang等，2023）的比对进行评估：其中TermPicks是涵盖多传感器与多种冰川工况的最大规模人工解译冰川末端迹线数据集，AutoTerm则是一款时间覆盖密度更高的自动化产品。与TermPicks相比的平均最小距离（AMD）为86米，与AutoTerm相比则为115米，表明本数据集的位置精度极高。 本数据集以GeoPackage格式发布（总大小：26.5 MB），并附带用于分析的各冰川专属中心线数据。其可为格陵兰冰川动力学研究、冰盖模型中时变边界条件的优化，以及自动化冰川前端追踪算法的开发与验证，提供覆盖范围广泛、质量优异的基准参考。