100 years of monitoring in the Swiss National Park reveals overall decreasing rock glacier velocities

Rock glaciers (RGs), lobed or tongue-shaped landforms composed of ice and frozen debris, are sensitive to climate change. Summer thaw is less pronounced than in glaciers due to the insulating effect of the active layer. The RGs of the Engadin, Southeastern Swiss Alps, are significant for permafrost research, boasting over 100 years of displacement rate data – the oldest globally. Emile and André Chaix initiated in-situ velocity measurements in 1918, focusing on four RGs (Val Sassa, Val da l'Acqua, Valletta, and Tantermozza). This study bridges the data gap between Chaix's early 20th-century measurements and recent observations to investigate long-term kinematic changes, focusing on the interplay between RGs and adjacent glaciers. The research seeks to understand how the degradation of adjacent glaciers, exacerbated by climate change, influences the dynamics and future behavior of these rock glaciers. The unique long-term dataset allows for an in-depth analysis of the spatiotemporal evolution of these features, contributing to a broader understanding of rock glacier response to climate change in the Alps.

Previous long-term kinematic analyses of rock glaciers typically extend back to the early to mid-20th century. The longest analysis (Fleischer et al.) integrated data from 1922, 1938, and later decades for the Innere Ölgruben rock glacier (Austria). Other studies employ historical aerial imagery and more recent GNSS measurements combined with UAV imagery, interferometric and geophysical analysis, or permafrost temperature borehole data. A consensus is emerging regarding an acceleration in rock glacier movement in recent decades, linked to increased global mean air temperature. Theoretically, acceleration can result from internal plastic deformation (enhanced by permafrost temperature increases) or basal shearing (influenced by water supply from rainfall, snowmelt, or melting ground ice). RG movement is a complex interplay of both factors, with the relative importance varying according to RG characteristics (geometry, structure, material rheology). The debate on the origin of RG ice continues, with the permafrost model and the ice-core theory representing two main theoretical end-members. Many recent studies suggest a synergy between both models is plausible where glacial-periglacial coexistence occurs.

This study utilizes photogrammetric techniques to analyze the spatiotemporal evolution of four RGs in the Swiss National Park. Data sources include: (i) Historical aerial imagery from Swisstopo (1946-2000), pre-processed to enhance detail and reduce noise; (ii) A UAV survey conducted in September 2022 in Val da l'Acqua, providing high-resolution data to capture recent changes; (iii) Historical in-situ measurements by Emile and André Chaix (1918-1942) and Heinrich Jäckli (Val da l'Acqua, 1968-1979); and (iv) Recent annual Differential Global Navigation Satellite System (DGNSS) measurements (2006-2022) from the SNP's GIS group. Photogrammetry processing (Pix4Dmapper 4.5.6) generated orthophotos and Digital Surface Models (DSMs) for each period. Horizontal surface velocities were determined using manual boulder tracking across consecutive images. Elevation changes were calculated using DSM differences. The most probable historical glacier boundaries were inferred by analyzing elevation changes and geomorphological features in the orthophotos and comparing against previous glacier inventories (1850, 1900, 1935, 1973). Volume changes were calculated using a Mean elevation difference by elevation bin method to account for voids in the DSMs. Uncertainty assessment was conducted to account for co-registration errors in photogrammetry, using random points in stable bedrock areas. The method involved creating random points in stable bedrock areas near RGs and Adjacent Glaciers, calculating horizontal and vertical differences between image pairs, interpolating these differences, computing mean and standard deviations to estimate uncertainties in horizontal displacements and vertical changes. Volume uncertainty was calculated by multiplying vertical error by the area.

All four RGs exhibit a decelerating trend in mean and maximum surface velocities since 1918 (Val Sassa and Val da l'Acqua) and 1946 (Tantermozza and Valletta). While some temporary accelerations occurred in the 1950s-1960s and 1990s, a continuous slowdown is evident from the 1960s-1970s onwards. Volume loss is substantial in Val Sassa and Valletta, while Tantermozza and Val da l'Acqua show localized elevation changes, with volume loss in the upper reaches and aggradation at the front. Adjacent glaciers (AGs) exhibit significant volume loss since 1956, retreating from positions held at the end of the Little Ice Age (LIA). Each RG is analyzed individually: **Tantermozza:** Heterogeneous kinematics across three zones (T1-T3), with the highest historical velocities in T1 (3.1 m a⁻¹). Volume loss occurred in T1, T2, and T3, with a transition from convex to flat/concave shapes in the frontal areas of the upper lobes. AG volume loss is 2.15 million m³ (1956-2015). **Valletta:** Relatively homogeneous dynamics, with a total volume loss of 0.58 million m³ (1956-2015). Highest historical velocities (2.5 m a⁻¹) occurred between 1946-1956. AG volume loss is 1.92 million m³ (1956-2015). **Val Sassa:** Largest drop in velocity and volume loss (0.60 million m³ since 1956). Three zones (S1-S3) exhibit distinct behavior, with the largest volume loss occurring in S2 and S3. AG volume loss is 3.69 million m³ (1956-2015). **Val da l'Acqua:** Most constant velocities, but with a negative trend and significant volume losses. Volume loss is 0.22 million m⁻¹ in the upper part (A2) and 0.13 million m³ gain at the front (A1). AG volume loss is 4.15 million m³ (1956-2022).

The overall decelerating trend and volume loss in the four RGs are consistent with a transition from secondary creep to degradation. The frontal deceleration, particularly noticeable in Val Sassa and Val da l'Acqua at lower elevations (2100-2300 m a.s.l.), contrasts with accelerations seen in destabilizing RGs elsewhere. This lower elevation deceleration might be due to upslope permafrost degradation, reducing the ice-rich layer at lower elevations. The observed upslope shift in velocities may be linked to a decrease in snow cover insulation, increasing annual mean ground temperatures (AMGT). A dynamical disconnection of the RG fronts from the upper rooting zones is also contributing factor in Val Sassa. The interaction with AGs may have implications for both topographic and hydrological factors. Highest velocities coincided with the post-LIA period when AG termini were close to the RG active zones; retreating glaciers reduced stress on RGs. Hydrological contributions of AGs seem to have influenced RG velocities: periods of high AG volume loss with no substantial RG ice depletion resulted in RG acceleration, while periods without significant RG volume loss and reduced AG hydrological contribution resulted in RG deceleration. Significant ice depletion in RGs, even with high water supply from adjacent glaciers, leads to deceleration, as observed by the end of the 1980s.

The study demonstrates that over a century, RGs in the Swiss National Park have transitioned from a state of secondary creep to degradation, driven primarily by permafrost degradation exacerbated by climate change. The interaction with adjacent glaciers, particularly the hydrological contribution, is a significant factor influencing RG dynamics. The findings suggest a future scenario with limited new accelerations, and a gradual slowdown of creep as permafrost degradation continues. Future research could incorporate geophysical surveys or borehole investigations to clarify ice origin and AG-RG interactions.

Uncertainties in Chaix's measurements (1918-1942) due to lack of detailed information on instruments and precision need to be acknowledged. Uncertainties related to photogrammetry are also present, influenced by image resolution and co-registration accuracy. The absence of subsurface data limits a precise assessment of volume changes and the contribution of ice loss vs. other factors. The study primarily focuses on surface kinematics and lacks detailed analysis of internal processes.