The two landslides, Jettan and Gámanjunni, form a natural experiment. Jettan is a complex slide in garnet-quartz-mica schist and calcite marble that lies below the modeled permafrost boundary, creeping at an average of about eight millimeters per year. Gámanjunni, a rotational slide in mica schist, sits above the permafrost boundary and moves more than three times faster, at roughly twenty-eight millimeters per year. Jettan holds an estimated six million cubic meters of rock above Lyngsfjorden, where an abrupt failure could generate a devastating tsunami; Gámanjunni’s twenty-five million cubic meters could dam the river below it, threatening catastrophic flooding if the dam collapsed. Both carry annual theoretical failure probabilities between one in one hundred and one in one thousand, which is why the Norwegian Water Resources and Energy Directorate has instrumented them so intensively.

The monitoring network at Jettan is among the most complete of any Arctic landslide. Ten GNSS receivers track surface motion with millimeter precision, while three boreholes, each about one hundred meters deep, contain strings of bi-axial inclinometers that measure deformation at every meter of depth, alongside piezometers that record groundwater pressure hourly. A broadband seismic station operated by NORSAR has captured ambient noise since 2016, allowing researchers to compute subtle changes in the apparent velocity of seismic waves traveling through the slide. At Gámanjunni, only surface GNSS and meteorological data exist, because drilling there was deemed prohibitively expensive. This asymmetry means Jettan provides the subsurface story, while Gámanjunni offers the permafrost-affected comparison.

Despite their very different thermal settings, the two landslides behave remarkably alike at the surface. Both accelerate in late spring and again in autumn, slow through summer, and reach a minimum in winter. The spring acceleration at Jettan precedes Gámanjunni’s by about seventeen days on average, most likely because Jettan sits at a lower elevation and its snowpack melts earlier. The correlation between the two velocity records is moderate to strong, and in spring the velocities track the combined input of rain and snowmelt closely: at Jettan the lag is only about two days, with a correlation coefficient of 0.71, while at Gámanjunni the lag stretches to roughly twenty days. In autumn, after snowmelt has ended, Jettan’s velocity rises in step with rainfall events, a relationship that is absent at Gámanjunni.

The borehole data from Jettan reveal something the surface instruments cannot see. Most of the landslide’s deformation is distributed through the rock volume rather than concentrated on a single discrete slip surface, a pattern the researchers attribute to the rock’s many planes of weakness and its layered lithology. Yet within this distributed deformation, localized shear zones are emerging and evolving. Borehole BH-1, which had moved steadily for years, began accelerating in 2020 and accelerated again in 2022, reaching twenty-one millimeters per year. The seasonal amplitude of the upper shear zone in borehole BH-2 nearly doubled between 2019 and 2022, from three to six millimeters per year, and nearly tripled to nine millimeters per year by 2025. These accelerations coincided with years of deeper snowpacks and more intense melt, and statistical modeling confirmed that the shear zones are becoming increasingly localized and increasingly sensitive to pore-water pressure.

To disentangle cause and effect, the team applied autoregressive distributed lag models, a statistical framework that relates present sliding velocity to its own past values and to current and lagged values of external drivers such as groundwater pressure, air temperature, precipitation, and seismic velocity change. The models, selected by the Bayesian Information Criterion and validated against held-out data, outperformed simpler models by thirty-seven to forty-three percent in out-of-sample error. For the upper shear zone of BH-2, groundwater pressure and the fourteen-day sum of rain and snowmelt showed strong correlations with slip velocity, with lags of just a few days. For BH-1, the dominant control was the borehole’s own displacement history, suggesting that progressive structural damage within the shear zone, rather than weather alone, explains why accelerations began only after 2020. The researchers interpret this as evidence of a maturing failure surface, in which local asperities are progressively breaking down and the slide is inching toward a more connected, through-going shear plane.

The seismic data add an entirely independent dimension. By cross-correlating ambient seismic noise recorded at the JETT station between 2016 and 2025, the team computed relative changes in surface wave velocity, a quantity sensitive to the rigidity, water content, and temperature of the upper tens of meters of the landslide. The signal follows a clear annual cycle: velocity drops every spring, reaches minima in summer and autumn, and recovers toward baseline in mid-winter. The drops correlate strongly and negatively with shallow ground temperature, with rain and snowmelt, and with slip velocity in the boreholes, with correlation coefficients as strong as negative 0.98. The researchers interpret this as an annual rheological transition: as meltwater and rain infiltrate the fractured rock, the landslide mass softens, shifting from rigid, block-like behavior toward compliant, volumetric deformation.

Notably, the timing of the two annual seismic velocity minima maps onto two different deformation styles. The first drop coincides with peak velocities in both the GNSS record and the borehole shear zones, indicating localized shear deformation in spring. The second drop, in autumn, aligns with surface acceleration but not with borehole acceleration, pointing instead to distributed volumetric deformation of the slide body. In other words, spring and autumn accelerations, which look similar from the surface, are mechanically distinct events happening at different depths and in different ways. This is precisely the kind of insight that only a combined surface and subsurface, geodetic and seismic dataset can deliver.

There are also warning signs in the long-term trends. The seismic velocity within the landslide has been declining at a rate of about 0.23 percent per year since 2019, which the authors suggest may reflect progressive internal changes such as increasing water content or reduced structural integrity. At other landslides worldwide, velocity drops of four to seven percent have preceded collapse or major acceleration, and at Jettan, four percent drops do precede acceleration events, although the standard threshold signals used elsewhere do not appear to be reliable failure precursors here. The authors are careful to note that the trend applies only to the observed period and should not be extrapolated, but the combination of a maturing shear zone, growing snowmelt sensitivity, and softening rock mass makes Jettan a slope that hazard managers will be watching closely.

Beyond the immediate hazard question, the study carries a broader message for a warming Arctic, a region that has warmed nearly four times faster than the globe since 1979. Distinguishing between hydrologically driven seasonal accelerations and deformation tied to permafrost thaw is essential for assessing which slopes face elevated long-term risk. The comparison between Jettan and Gámanjunni shows that permafrost-influenced slides can creep faster, possibly through temperature-dependent creep along ice-filled joints, without that necessarily implying greater danger of sudden failure. It also demonstrates that seismic ambient noise monitoring, which is far cheaper than drilling, can serve as a cost-effective complement to boreholes, tracking water infiltration and mechanical stiffness changes across an entire landslide rather than at isolated points. As snowpacks deepen, melt intensifies, and frozen ground retreats across the circumpolar North, the fjord-side laboratories of Troms County are showing how to listen to what the mountains are trying to say.

Subject of Research: Hydrological and permafrost controls on the seasonal and long-term deformation of two Arctic landslides in northern Norway

Article Title: Seasonal and inter-annual evolution of the deformation of two Arctic landslides

Article References: Seasonal and inter-annual evolution of the deformation of two Arctic landslides. (n.d.). https://doi.org/10.5194/esurf-14-729-2026

Image Credits: AI Generated

DOI: 10.5194/esurf-14-729-2026

Keywords: Arctic landslides, permafrost, snowmelt, groundwater pressure, seismic ambient noise, GNSS monitoring, borehole inclinometers, Norway, climate change, rockslide hazard, shear zones, landslide monitoring