Inventory of active rock glaciers in the western Nyainqêntanglha Range, Tibetan Plateau

The western Nyainqêntanglha Range on the Tibetan Plateau (TP) reaches an elevation of 7162 m and is characterized by an extensive periglacial environment. Here, we present the first rock glacier inventory of the central TP containing 1433 rock glaciers over an area of 4622 km². The rock glaciers are identified based on their surface velocity. The surface velocity is derived from Sentinel-1 satellite data of 2016 to 2019 via InSAR time series analysis. 16.4 % of the inventoried rock glaciers are classified as active with a surface velocity above 10 cmyr^-1 and 80.0 % are classified as transitional with 1 to 10 cmyr^-1. The western Nyainqêntanglha Range forms a climate divide between the dry continental climate brought by the Westerlies from the north-west and the Indian Summer Monsoon to the south. 89.7 % of all active rock glaciers and 74 % of the free ice glacial area are located on the southern side. The higher moisture availability on the southern (windward) side of the mountain range is likely the cause of a higher rock glacier occurrence and the greater activity.

Manually identifying and outlining rock glaciers is time consuming and subjective. To ensure a high reliability and comparability of our inventory, we therefore combined a manual approach with an automated classification. Three analysts worked in tandem to generate the manual outlines according to the guidelines of the IPA action group on ‘Rock glacier inventories and kinematics’. A subset of these outlines acted as training areas for a pixel-based maximum likelihood classification. Both the manual and the automated classification were performed based on DEM parameters (elevation, slope etc.), optical datasets (Sentinel-2 and NDVI) and surface velocity (generated with InSAR). 87.8 % of all manually outlined rock glaciers were identified successfully at a true positive rate of 69.5 %. 18 additional rock glaciers were added to the inventory based on the automated classification. This combined approach is therefore beneficial to generate a complete inventory. The automated classification can, however, not replace the expertise of an analyst as it greatly overestimates the actual rock glacier area.