Surface velocity fields of active rock glaciers and ice-debris complexes in the Central Andes of Argentina

• Published in: Earth Surface Processes and Landforms
• Main Authors: Blöthe, Jan Henrik, Halla, Christian, Schwalbe, Ellen, Bottegal, Estefania Laura, Trombotto, Dario Tomas, Schrott, Lothar
• Format: Article in Journal/Newspaper
• Language: English
• Published: John Wiley & Sons Ltd
• Subjects: CENTRAL ANDES; CENTRELINE MODELLING; FEATURE TRACKING; ICE-DEBRIS COMPLEXES; PERMAFROST; RANDOM FORESTS; RAPIDEYE; ROCK GLACIERS; https://purl.org/becyt/ford/1.5; https://purl.org/becyt/ford/1; Argentina; Ice; permafrost
• Online Access: http://hdl.handle.net/11336/138342

Rock glaciers and transitional ice-debris complexes predominate the Central Andean landform assemblage, yet regional studies on their state of activity and their kinematics remain sparse. Here we utilize the national glacier inventory of Argentina to quantify surface velocity fields of 244 rock glaciers and 51 ice-debris complexes, located in the Cordón del Plata range, Argentina. Applying a feature-tracking approach to repeated RapidEye satellite imagery acquired between 2010 and 2017/18, we find mean displacement rates between 0.37 and 2.61 m year−1 for 149 landforms, while for the remaining 146 features, surface movement remains below our level of detection. We compare our satellite-derived velocity fields with ground-truth data from two local field sites and find closely matching results in magnitude and spatial distribution. With average displacement of one-third of the active rock glaciers and ice-debris complexes exceeding 1 m year−1, the region hosts an exceptional number of fast-flowing periglacial landforms, compared to other mountain belts. Using a random forest model, we test the predictive power of 25 morphometric and topoclimatic candidate predictors for modelling the state of activity of rock glaciers and ice-debris complexes on two different scales. For entire landforms and individual landform segments, constructed along displacement centrelines, we can predict the state of activity with overall accuracies of 70.08% (mean AUROC = 0.785) and 74.86% (mean AUROC = 0.753), respectively. While topoclimatic parameters such as solar radiation and elevation are most important for entire landforms, geometric parameters become more important at the scale of landform segments. Despite tentative correlations between local slope and surface kinematics, our results point to factors integrating slope and distance to the source to govern local deformation. We conclude that feature tracking in optical imagery is feasible for regional studies in remote regions and provides valuable insight into the current state ...