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

The study addresses how and why rock glacier velocities in the Swiss National Park have changed over the last century and how adjacent, now-degrading glaciers have influenced these dynamics. The context is unique as these sites hosted the world’s first in-situ rock glacier displacement measurements by Émile and André Chaix starting in 1918. A significant gap existed between those early measurements and modern observations. The aims are: (i) to fill the historical data gap and build a continuous, century-scale kinematic record for the four rock glaciers originally studied by Chaix (Tantermozza, Valletta, Val Sassa, Val da l’Acqua); and (ii) to reconstruct the historical evolution of adjacent debris-covered glaciers and investigate their interrelationships with rock glacier kinematics. This work is important for understanding permafrost landform responses to climate change, especially the roles of hydrology and mechanical coupling with nearby glaciers.

Prior long-term rock glacier studies typically extend back to the 1940s–1950s and often rely on historical aerial photogrammetry complemented by GNSS, UAV imagery, InSAR, geophysics, and borehole temperature data. A growing body of literature reports widespread acceleration of Alpine rock glaciers in recent decades, frequently linked to atmospheric warming and permafrost temperature increases that reduce ice viscosity and strength, as well as hydrological influences on basal shear horizons. Debate persists on rock glacier ice origin (permafrost-derived versus relict glacial ice), with evidence for both endmembers and transitional zones in glacial–periglacial contexts. Historical mapping (Dufour and Siegfried maps; Swiss Glacier Inventories) documents substantial glacier retreat since the Little Ice Age, which may have modulated rock glacier dynamics via both mechanical loading/pushing and hydrological inputs. Case studies (e.g., Innere Ölgruben, Muragl, and others across the Alps and worldwide) provide methodological and conceptual precedents that this study extends by adding a century-long, observation-based kinematic record and explicit glacier–rock glacier interaction analysis.

The authors combined historical and modern datasets to create continuous kinematic and volumetric records: (1) Antecedent measurements: Chaix conducted in-situ boulder displacement measurements on Val Sassa and Val da l’Acqua in 1918–1919 and 1919–1921, with repeats in 1942. Additional measurements were reported by Jäckli at Val da l’Acqua (1968–1979). Modern DGNSS campaigns by the Swiss National Park GIS group provided nearly annual surface velocity measurements at 8 positions on Val Sassa (2006–2020) and 19 positions near the Val da l’Acqua front (2007–2022). (2) Photogrammetry: The team processed 651 historical aerial images from Swisstopo (years 1946, 1956, 1962, 1973, 1979, 1985, 1988, 1991, 1997, 2000; 2003 not usable) and a 2022 UAV survey (Ebee RTK) for Val da l’Acqua. Images were pre-processed (brightness/contrast adjustments, sharpening, level matching), and point cloud noise was manually cleaned. Ground control points (GCPs) included DGNSS-measured stable boulders (±10–40 cm) near active areas and stable outcrops upstream (XY from 2019 10 cm orthophoto; Z from 2015 SwissALTI DEM at 0.5 m; ~±2 m). Pix4Dmapper 4.5.6 was used, yielding 40 orthophotos and 40 DSMs (10 per rock glacier) with 0.2–0.7 m pixel resolution. The 2022 UAV dataset (10 cm GSD; RTK accuracy ~±2 cm) served as a reference for Val da l’Acqua co-registration. (3) Horizontal velocities: Manual boulder tracking on orthophotos defined active area limits and computed inter-epoch displacements (cm/yr) for each rock glacier. Approximately 12,000 boulders were digitized across periods, achieving homogeneous spatial coverage. Inverse Distance Weighting (IDW) produced 1 m velocity rasters for each epoch within active area boundaries. (4) Elevation change and volume: DSM differencing (DSM-d) was performed for periods with adequate coverage and quality (Tantermozza: 1956–1962–1973–1988–1991–1997–2015; Valletta and Val Sassa: 1956–1962–1973–1991–2015; Val da l’Acqua: 1956–1973–1988–1991–2000–2015–2022). Noise-prone pixels were masked using hillshade-based interpretation. Elevation changes were normalized by period length (cm/yr). Volume changes for active RG areas and inferred Adjacent Glacier (AG) areas were computed using the mean elevation difference by elevation bin method: bin size ~10% of elevation range, local polynomial interpolation to fill remaining voids, and summation of bin volumes. (5) Inference of AG extents: AG boundaries for each epoch were delineated by combining large negative DSM-d values (persistent ablation) with geomorphic evidence on orthophotos, and compared against SGI 1850 (Dufour), 1900/1935 (Siegfried), and SGI 1973 outlines. (6) Uncertainty assessment: Co-registration errors were quantified using random points on stable bedrock near target areas, deriving spatially interpolated vertical and horizontal relative error maps. Limits of detection (LoD) were typically 5–25 cm/yr for horizontal displacements and generally <20 cm/yr for vertical changes. Volume uncertainties were estimated as twice the DSM-d standard deviation multiplied by area. Historical Chaix data uncertainties were approximated at ~20 cm based on typical early 20th-century surveying precision.

• Century-scale kinematics: All four rock glaciers exhibit a long-term decelerating trend in mean and maximum surface velocities. Temporary accelerations occurred (Val Sassa and Tantermozza in the 1950s; Val da l’Acqua in the 1960s; several sites in the 1990s), but a pronounced slowdown followed from the 1960s–1970s, with a common minimum between 1985/88 and 1991 and a steady deceleration since ~2000 to record-low rates. • Site-specific velocity highlights: – Tantermozza reached maxima of ~2.5 m/yr at the T1 front (1946–1956) and ~3.1 m/yr (1956–1962) in T1, with subsequent deceleration to ~0.8 m/yr in recent decades; upper lobes (T3) slowed to ~0.5–0.6 m/yr and remained low. – Valletta recorded highest velocities ~2.5 m/yr (1946–1956), dropped to ≤1.2 m/yr (1956–1962), saw ~2.2 m/yr in the 1990s, and ~1.2 m/yr from 2000–2019. – Val Sassa peaked at ~2 m/yr (1918–1919) in S1; decelerated to ~0.7 m/yr (1962–1973) in S1 and S3 and ~0.2 m/yr in S2; since 2000, DGNSS and photogrammetry show <5 cm/yr in S1 and S2 and ~7 cm/yr in S3. – Val da l’Acqua maintained relatively stable velocities (1.5–2 m/yr) until the 1960s; localized maxima up to 3 m/yr (1962–1973); since 2000, A1 decreased to ≤0.5 m/yr while A2 remained around ~1.4 m/yr. • Volume changes in rock glaciers: – Tantermozza: Mixed pattern with elevation loss in upper zones and aggradation at the front; overall 2015 volumes comparable to 1956. – Valletta: Total RG volume loss ~0.58 ± 0.081 million m³ (1956–2015), with general flattening. – Val Sassa: Total RG volume loss ~0.60 ± 0.023 million m³ since 1956, concentrated in S2–S3; front S1 developed central depletion relative to margins in recent decades. – Val da l’Acqua: Upper A2 degraded by ~0.22 ± 0.064 million m³; front A1 aggraded by ~0.13 ± 0.027 million m³. • Adjacent Glacier (AG) depletion: Strong, continuous AG volume losses across all basins since 1956: – Tantermozza AG: −2.15 ± 0.082 million m³ (~16.99 m w.e.) (1956–2015). – Valletta AG: −1.92 ± 0.078 million m³ (~11.01 m w.e.) (1956–2015). – Val Sassa AG: −3.69 ± 0.166 million m³ (~11.83 m w.e.) (1956–2015). – Val da l’Acqua AG: −4.15 ± 0.307 million m³ (~15.80 m w.e.) (1956–2022), the largest loss among sites. • Spatial evolution and morphology: AG fronts retreated from near-LIA positions and developed concave, overdeepened longitudinal profiles. Rock glaciers show progressive flattening and signs of ice depletion, particularly in lower frontal zones that decelerated markedly, consistent with upslope permafrost degradation. • Glacier–rock glacier interactions: Peak RG velocities coincided with post-LIA configurations where AG termini contacted or closely bordered RG upper zones, implying mechanical pushing and added driving stress. Periods of high AG ablation without major simultaneous RG ice depletion correlated with RG accelerations (e.g., 1956–1973; 1990s). Conversely, reductions in AG hydrological inputs or substantial RG ice depletion coincided with RG decelerations. • Outlook: With AGs now severely depleted and increasingly debris-covered, major new accelerations are unlikely. A reduced but steady water supply may sustain slow secondary creep that will progressively diminish as permafrost continues to degrade under warming and drier conditions.

The findings answer the core questions by demonstrating that long-term deceleration and volume loss dominate the century-scale evolution of the four rock glaciers, and that their kinematics were strongly modulated by interactions with adjacent glaciers. Mechanical forcing (pushing/loading) when AGs were proximal near the end of the LIA likely enhanced driving stresses and contributed to earlier elevated velocities. Over time, progressive AG retreat reduced both mechanical coupling and hydrological supply, coinciding with RG deceleration. The most pronounced decelerations in the frontal, lower-elevation sections correspond to elevations near the lower limit of discontinuous permafrost in the region, consistent with warming-driven permafrost degradation and reduced snow insulation since the mid-1980s. The late-1980s climate anomaly (winter warming and fewer snow days) aligns with a basin-wide minimum in velocities and high ice loss (1985–1991), followed by a short-lived 1990s acceleration, likely reflecting lagged rheological and hydrological responses. This nuanced behavior contrasts with generalized Alpine trends of sustained acceleration, highlighting the dominant roles of site-specific hydrology and glacier connectivity when RGs are not undergoing severe ice depletion. Given the present, debris-covered, overdeepened, and retreating state of AGs, the study suggests limited scope for future accelerations. Instead, a gradual decline in secondary creep is expected as permafrost degradation progresses, with occasional short-term hydrologically driven speed-ups possible during periods of high precipitation.

By building the longest observation-based rock glacier kinematics record to date for the Swiss National Park and explicitly coupling it with the evolution of adjacent debris-covered glaciers, the study shows a century-scale deceleration and widespread volume loss consistent with permafrost degradation and diminishing glacier influence. Peak velocities occurred when adjacent glaciers were proximal after the LIA, and subsequent retreat reduced both mechanical and hydrological forcing on rock glaciers. With AGs now substantially depleted and increasingly debris-covered, major accelerations are unlikely; instead, slow secondary creep sustained by limited water inputs is expected to wane gradually under ongoing warming and relatively dry conditions. Future research could benefit from subsurface investigations (geophysics, boreholes) to quantify ice content, the depth and continuity of shear horizons, and hydrological pathways, as well as continuous multi-sensor monitoring to capture short-term hydrologic forcing. However, such work must balance the strict protection status of the Swiss National Park.

Uncertainties arise from: (1) Early in-situ data (1918–1942) lacking explicit error estimates; typical instrument precision suggests ~20 cm maximum error relative to 0.8–2 m/yr displacements. (2) Photogrammetry co-registration errors varying with image quality and GCP distribution, yielding LoD of ~5–25 cm/yr for horizontal displacements and generally <20 cm/yr for vertical changes. (3) Volume change interpretation in rock glaciers is complicated by sediment–ice–air mixtures; elevation change may not map linearly to ice loss due to porosity evolution, lateral strain, and compaction. (4) Absence of subsurface data (geophysics, boreholes) limits direct attribution of volume change to ice melt and constrains understanding of internal structure and basal shear. (5) Protected area restrictions prevent invasive validation in former AG–RG contact zones.