Résumé
Rock glaciers (RGs) are ice and debris landforms shaped by long-term permafrost creep. Their inventory has expanded significantly in the past two decades due to their importance as water resources and indicators of climate change. Previous inventories in the Pyrenees are sparse and lack essential kinematic data, leading to an underestimation of active RGs. This study presents the first kinematic database of Pyrenean RGs, integrating European Ground Motion Service (EGMS) InSAR data with airborne Laser Imaging Detection and Ranging (LiDAR) datasets and seven years of Sentinel-1 high-resolution InSAR data processed by the SqueeSAR algorithm. The analysis focuses on three igneous plutons (Panticosa, Cauterets, and Neouvielle) in the central Pyrenees, comparing surface displacement rates from these techniques to previous geodetic measurements. A total of 733 RGs have been mapped in the Pyrenees, covering an area of 58.9 km 2 . 73 % of the mapped RGs are inactive or relict, showing no ground displacement and being partially vegetated. Only 13 % (96 RGs) remain active, covering less than 10 km 2 , and are primarily north-facing. An additional 14 % (102 RGs) lack InSAR data but they are considered potentially active based on their orientation, altitude, and remarkable morphological features. The existence of active RGs at relatively low altitudes lowers the permafrost boundary between 2100 and 2600 m on north-and south-facing slopes, respectively. Despite the limitations of InSAR to measure displacements along the north-south axis and the fact that most active RGS are oriented to the north, according to the decomposition of the LOS displacement to vertical and horizontal components and LiDAR data, the ongoing subsidence and decline in horizontal movement related to ice degradation suggest a transition of many RGs from active to relict, marking rapid permafrost degradation. Finally, the EGMS has proven inadequate for detecting slow-moving active RGs in the Pyrenees due to temporal decorrelation caused by prolonged snow cover periods, particularly when compared to the combined use of SqueeSAR and airborne LiDAR datasets. The EGMS detected ground information for only 14 out of 89 active RGs, provided inaccurate kinematic data and underestimated the decomposed vertical and horizontal velocities by up to fourfold