Nicu Ionut Cristi, Rubensdotter Lena, Tanyaș Hakan, Lombardo Luigi
High North Department, Norwegian Institute for Cultural Heritage Research (NIKU), Fram Centre, N-9296, Tromsø, Norway.
Geohazard and Earth Observation Department, Geological Survey of Norway (NGU), P.O. Box 6315 Torgarden, 7491, Trondheim, Norway.
Sci Data. 2025 Jul 14;12(1):1221. doi: 10.1038/s41597-025-05592-7.
Retrogressive thaw slumps (RTS) are thermokarst landforms resulting from the thawing of ice-rich permafrost in the pan-Arctic and high mountain regions, recognized as climate-related phenomena. Using 2022 ESRI Wayback satellite imagery, we manually digitized 900 RTS polygons on the Kanin Peninsula (NW Russia), categorizing them into 633 inactive and 267 active features based on morphology and vegetation patterns. The primary goal was not to create a complete RTS inventory but to develop a highly representative dataset for training machine learning models to automatically detect and classify RTS into active and inactive categories. This approach will facilitate regional-scale monitoring of permafrost degradation and enhance understanding of RTS impacts, including their effects on indigenous Nenets reindeer herders' travel routes. By classifying RTS polygons into morphologically active and inactive features, this dataset aims to enable ongoing analysis of change patterns over time, offering critical insights in the current climate context.
退行性融冻滑塌(RTS)是北极泛区和高山地区富含冰的永久冻土融化形成的热喀斯特地貌,被认为是与气候相关的现象。利用2022年ESRI时光倒流卫星图像,我们在卡宁半岛(俄罗斯西北部)手动数字化了900个RTS多边形,根据形态和植被模式将它们分为633个非活动特征和267个活动特征。主要目标不是创建一个完整的RTS清单,而是开发一个具有高度代表性的数据集,用于训练机器学习模型,以自动检测RTS并将其分类为活动和非活动类别。这种方法将有助于对永久冻土退化进行区域尺度监测,并增进对RTS影响的理解,包括其对当地涅涅茨驯鹿牧民出行路线的影响。通过将RTS多边形分类为形态上的活动和非活动特征,该数据集旨在能够对随时间变化的模式进行持续分析,在当前气候背景下提供关键见解。