The submitted manuscript investigates processes driving arctic/permafrost-affected rock slope failures. By analyzing monitoring data (6 to 12 years collection periods) of two well-researched landslides in Norway, the authors discuss the role of climatic influences (rain, snow, permafrost degradation) on failure kinematics, and provide insights into seasonal controls and deformation behaviour of rock slope beyond millions of cubic meters. The strength of the manuscript lies in the fusion of datasets (meteorological, seismic, surface and subsurface displacement & velocities, hydrogeological data) and their combined interpretation, and the data itself (scarcity of in-situ data). Moreover, the authors demonstrate the use of an Auto Regressive Distributed Lags model in predicting sliding velocity on the basis of past sliding velocity and exogenous parameters (precipitation, temperature, relative seismic velocity change). The model was used to backcalculate the individual share of the parameters in contributing to overall deformation. With a detailed analysis of in-situ records, this manuscript substantially contributes to deciphering the complex driving mechanism of arctic rock slope deformations.

The data, data processing, and methodology are precisely described and sound. Overall manuscript quality is good. I have minor remarks on the figures and the discussion section. Literature is adequate.

I suggest publication after minor revisions:

Discussion: A “synthesis” section or “synoptic discussion” would strengthen understanding of the interwoven process. An illustrated conceptual model (i.e., a figure showing two cross-sections) of both landslides could be really nice to highlight the differences and similarities. I.e., permafrost distribution, geological structures, shear planes/zones, and location/degree of fractured rock mass, water availability/height of pressure head, differential kinematics, and reaction time lag. You demonstrated multiple relationships effectively, but a clear synthesis of the core insights would strengthen the overall message. The writing can be somewhat technical at times (many numbers), making some aspects difficult to read, though it is generally acceptable given the type of analysis presented.

Strong discussion in general. Focus is yet specific on the studied sites. A designated section on transferability, intercomparability to other arctic/permafrost rock slope studies would be appreciated. Were the same seasonal shifts in hydrogeology or in slope velocities and their magnitudes reported elsewhere in comparable settings?

Maybe compare your results to Scondroglio et al. (2025) – long-term joint seepage record in sporadic permafrost terrain; Offer et al. also (2025) show seasonal permeability changes through ERT, and piezometric records in mountain permafrost.

Scandroglio, R., Weber, S., Rehm, T., and Krautblatter, M.: Decadal in Situ Hydrological Observations and Empirical Modeling of Pressure Head in a High-Alpine, Fractured Calcareous Rock Slope, Earth Surface Dynamics, 13, 295–314, https://doi.org/10.5194/esurf-13-2952025, 2025.

Offer, M., Weber, S., Krautblatter, M., Hartmeyer, I., and Keuschnig, M.: Pressurised water flow in fractured permafrost rocks revealed by borehole temperature, electrical resistivity tomography, and piezometric pressure, The Cryosphere, 19, 485–506, https://doi.org/10.5194/tc-19-485-2025, 2025.

Concerning hydrogeological flow paths: It would be interesting to show how piezometric measurements in boreholes correlate with snowmelt and rainfall. Maybe add a paragraph or figure on the connection between surface water availability and subsurface water or infiltration times. As I understood, slope kinematics are mainly explained by surface water (rain/snow) rather than recorded borehole hydraulic heads (except BH-2 upper; Figure A2). Typically, transient hydrostatic pressure buildup in the rock mass may explain kinematics. I am aware you noted “…because open boreholes can misrepresent subsurface water flow (Aspaas et al., 2024),..“ in line 460. Still, it might be interesting to see!

% Minor Remarks

“Catastrophic landslide” is used throughout the text. Please clarify what you mean. You study slow-moving gravitational rock slope deformations. I guess you mean rapid rock detachment evolving into highly mobile mass movements? Catastrophic is related to damage to humans/infrastructure and not to the process itself.

Ln21: How are they highlighted? More specifically.

Ln27: Not only permafrost thaw, but also warming permafrost temperatures may be relevant in the final detachment phase (Mamot et al. 2018).

Mamot, P., Weber, S., Schröder, T., and Krautblatter, M.: A Temperature- and Stress-Controlled Failure Criterion for Ice-Filled Permafrost Rock Joints, The Cryosphere, 12, 3333–3353, https://doi.org/10.5194/tc-12-3333-2018, 2018.

Ln27ff: Rather than singular individual drivers, the coupling of processes typically leads to enhanced progressive failure (Grämiger et al., 2020) or may peak in final detachment (Pfluger et al., 2025): Polythermal shifts in glacier or rock slope regimes impact hydrogeology – previous impermeable areas are fed with infiltraing meltwater. Hydrogeology impacts deformation and thermal conditions in turn (hydro-thermo-mechanical feedbacks). Often impossible to decipher individual factors for destabilization.

Grämiger, L. M., Moore, J. R., Gischig, V. S., Loew, S., Funk, M., and Limpach, P.: Hydromechanical Rock Slope Damage During Late Pleistocene and Holocene Glacial Cycles in an Alpine Valley, Journal of Geophysical Research: Earth Surface, 125, https://doi.org/10.1029/2019JF005494, 2020.

Pfluger, F., Weber, S., Steinhauser, J., Zangerl, C., Fey, C., Fürst, J., and Krautblatter, M.: Massive Permafrost Rock Slide under a Warming Polythermal Glacier Deciphered through Mechanical Modeling (Bliggspitze, Austria), Earth Surface Dynamics, 13, 41–70, https://doi.org/10.5194/esurf-13-41-2025, 2025.

Table 1: *Sloping Local Base Level (SLBL) -> Reference to calculation, publication, or the model is missing.

Figure 2: Figures and subfigures should be labeled uniformly throughout the manuscript (a),(b),…

I would recommend each subfigure with its own legend for clarity.

Highlight location for displayed monitoring records (b), (c) in the map (a).

Figure 2. Caption. Borehole displacement given for what depth/location? Specify. Please give a reference altitude for the temperature. Was the temperature/precipitation measured at the local site or from seNorge? (same for Figure 3).

Figure 4. The figure appears before the notion in the text. For what period is the displacement shown, add in the figure or caption.

Figure 5. Both temperatures look almost the same; It would be good to have a reference altitude for the temperature. i.e., use the location of the head scarp, as the two landslides vary distinctly in elevation range.

Ln310: If a central driver is water pressure, show the plot also in the main manuscript rather than the appendix. Best add it into one and the same with snow/rainfall data to show if there is a link in the records.

Section 3.4.2: Correlations: Please use a table to show the correlations of tested pairs. Visually, it is easier to compare in a table than written in text form.

Ln366: This first paragraph is a bit generic. Rather than using it as an introduction (as it was anyway in section 1), I would skip it here, and emphasize your own subsections on comparability and transferability of your results.

Ln498: Please refer to the Figures accordingly.

ff. For me, it is unclear what dv/v decrease is associated with shear plane wetting and what is associated with localized/volumetric deformation. The impact of deformation is likely higher, but both processes affect the measured data at the same time. Can you estimate orders of magnitude for the dv/v reduction impact of deformation and of water availability?

Ln519: Conclusion: First paragraph is repetitive; I would not cite literature in the conclusion; it is about your work. For my taste this paragraph can be skipped. I.e. merge the last sentence with the first of line 525.

Ln537: Be specific. How does it improve hazard assessment and monitoring strategies? It is easy to say through more data (i.e., surface and subsurface observations), but how many boreholes do we have at such locations? I would put the emphasis more on the processes.

I was happy reviewing the manuscript. Detailed insights into spatio-temporal deformation processes of permafrost/arctic rock slopes. Great contribution!

This paper investigates the kinematics, mechanisms and controls of two rockslides in the arctic region of Norway (Jettan and Gàmanjunni), leveraging time series (2015–2025 for Jettan; 2021–2025 for Gàmanjunni) of surface displacements and hydro-meteorological data (both landslides), as well as borehole displacements, temperature, pore pressure and seismic data (Jettan only). The authors perform statistical analyses aimed at establishing correlations between seasonal and interannual displacements and different variables, representing possible landslide triggers/controls (rainfall, snowmelt, pore pressures, temperature) or landslide internal damage (seismic dv/v). The authors also attempt to understand the temporal evolution of the landslide mechanisms, particularly for Jettan, where high-quality borehole data are available.

Improving the understanding of landslide mechanisms and controls in high-mountain environments is important for enhancing landslide forecasting capacities in the context of climate change. The topic is relevant and suitable for the journal. The authors benefit from a wealth of high-quality data collected over many years at two well-characterised landslides and apply sound statistical analysis methods. Nevertheless, the quality of the data and analyses is not fully reflected in the paper's narrative and discussion of results. In particular (see also the detailed comments below):

• The paper appears to pursue two distinct objectives: a) understanding the controls of different environmental factors (including permafrost) on the dynamics of two landslides located above and below the permafrost limit; and b) describing the mechanisms of the Jettan landslide specifically, as supported by borehole data. For this reason, the discussion is not straightforward and the take-home messages are somewhat unclear.
• The authors performed sophisticated analyses of excellent data, but the results are presented in an extremely descriptive — albeit very detailed — manner, a bit disconnected from their interpretation, which in turn is not framed within robust geological models of the studied landslides. As a result, the conclusions appear rather generic and do not seem to contribute substantially to current landslide forecasting capabilities. I suggest that more details on the geological-structural model of the studied landslides would support a more convincing interpretation and discussion of the results. See, for example, the paper by Etzelmuller et al. (2022) in ESurf, of which some of the authors of the manuscript under review are co-authors.
• Still regarding the geological reference model: the simplified landslide cross-sections of Jettan suggest strong lateral variations in landslide thickness. Could the authors elaborate on the possible effects of this complex geometry on the observed landslide behaviour? Could these characteristics account for the differences between the behaviours of Jettan and Gàmanjunni without invoking permafrost controls? This represents another major issue in the manuscript: the authors present very interesting data and a thorough description of the kinematics of these landslides, yet the interpretation of the controls on the observed behaviour does not appear to be fully supported by the data analysis.
• A major issue: what criterion was used to identify the outlined "acceleration periods" (e.g. creep bursts)? In attempting to correlate acceleration events with the measured (or calculated) environmental variables, an objective criterion is required to define such events. In Figure 6, while the seasonal accelerations are clearly identifiable, creep bursts occurring in the shear zones in boreholes are less obvious and resemble other similar signals that were not outlined as acceleration periods. How were these identified? Clarifying this point is essential to make all subsequent correlations credible to the reader.

DETAILED COMMENTS

Abstract: A few additional words on the basic geological features of the two studied landslides (including their differences) would be useful here to properly frame the proposed results.

Page 2, line 23: "millimetres to metres per year": is there a reference for this definition of "slow-moving landslides"? Does this definition apply to landslides in all materials (soil and rock) or specifically to rockslides?

Page 2, lines 33–36: Aside from the indispensable work by Iverson, the references here are limited to work by the authors and some US researchers collaborating with them. Several other citable references on this topic exist from different landslide research groups in Europe and beyond.

Page 2, lines 41–42: "the objective of this study": the objectives appear twofold and somewhat unclear. See the General Comments above and further comments below.

Page 3, Table 1: "Sloping Local Base Level (SLBL)": this concept may be unfamiliar to some readers, yet it is important. Please include adequate references and an explanation in the caption.

Figure 1: I assume the authors have access to a permafrost map or an APIM-like dataset. Would it be possible to overlay the likely permafrost extent to show the relative positions of the two landslides?

Figures 2 and 3: Delineating landslide boundaries and their main features (e,g, back-cracks, scarps) would greatly help the reader to understand the processes below the monitoring data.

Page 6, line 106: "Gàmanjunni landslide dataset is limited to surface displacement and meteorological data": this is an important issue. According to the paper's title and abstract, the manuscript focuses on comparing the seasonal and multi-annual behaviour of two nearby landslides located below and above the permafrost limit. However, the two landslides differ considerably in their characteristics and available data, limiting their comparison. At the same time, the authors present a more in-depth analysis of the mechanisms of the Jettan landslide, supported by the availability of both surface and borehole data. While this analysis is very interesting, it raises the question: what is the actual scope of the paper?

Page 6, line 124: "Geoengineering Service Center": the company is Italian and named "Centro Servizi di Geoingegneria (CSG)". Translating a company name should be avoided; I suggest referring to it as "CSG".

Figures 2 and 3 (time series): What averaging time windows were used to calculate the velocities of the two landslides? Are the same windows applied to both case studies? How were they selected? This is important for understanding how correlations between triggering factors and landslide response are evaluated.

Page 7, bottom of page (line numbers missing): "a piezometer was placed near the bottom of each borehole": how was this setup decided? Is there a single, fully interconnected aquifer within the landslide? Can the authors confirm that the pore pressures measured here (below the main sliding surfaces) are representative of conditions at or above the sliding surface? The presentation and interpretation of the data would benefit from a clear explanation of this.

Page 8, Figure 4: The landslide cross-sections suggest strong lateral variations in landslide thickness. Could the authors elaborate on the possible effects of this complex (3D) geometry on the measured kinematics? Could these characteristics account for the differences in behaviour between Jettan and Gàmanjunni without invoking permafrost controls? As noted above, this represents a major issue in the manuscript: the authors present very interesting data and a thorough description of the kinematics of these landslides, yet the interpretation of the controls on the observed behaviour does not appear to be fully supported by the data analysis.

Page 10, line 195: "W": how many days does W represent? The averaging time window used to calculate velocity is generally critical when attempting to correlate changes (accelerations/decelerations, i.e. "bursts") with environmental variables and should therefore be clearly stated.

Page 11, lines 199–202: This is not entirely clear: what do the authors mean by "temporal aggregation of the time series"?

Page 11, lines 203–209: This is a valuable component of the analysis, technically described in the appendices, but not fully exploited to provide an integrated understanding of the relationships between rockslide mechanisms, kinematics, and environmental controls. As noted in the general comments, this seems to be a limitation of the study: the authors performed sophisticated analyses of excellent data, but the interpretation is not framed within robust geological/technical models of the studied landslides, resulting in rather speculative conclusions that do not appear to contribute substantially to current landslide forecasting capabilities.

Page 11, line 219: "the model allows predicting": if I understand correctly, the aim of the ARDL regression model is to quantify the relative contributions of different controlling variables (e.g. rainfall, snowmelt, temperature) to the measured displacements, rather than to make predictions about future landslide behaviour. If so, I would suggest the authors state this clearly to avoid confusion. Could the phrasing be changed to "the model allows explaining a time series…" instead of "predicting"?

Page 12, lines 236–238: "since the process is computationally intensive, we down-sampled the time series to weekly intervals…": I understand it. However, did the authors perform an a priori verification of the typical response time of the landslides to the triggers? Is a weekly sampling rate sufficient to capture the observed processes? There is a risk of feeding a sophisticated model with data at an inadequate temporal resolution. Could the authors elaborate on this point?

Page 12, line 262: "7.8 mm/yr": what is the accuracy of this velocity estimate? Lines 287–289 on page 14 suggest an error of 1 mm/yr, in which case it would be more appropriate to round the value to, e.g., "8 mm/yr". Please consider this here and elsewhere throughout the manuscript.

Page 15, Figure 6: What criterion was used to identify the outlined "acceleration periods"? As noted above, an objective criterion is required to define such events when attempting to correlate them with measured (or simulated) environmental variables. In this figure, while seasonal accelerations at shallower levels are clearly visible, those in borehole shear zones are less distinct and resemble other similar signals that were not identified as acceleration periods. How were these selected? Clarifying this point is essential for the correlations proposed in Section 3.4 to be credible to the reader.

Page 17, lines 344–345: "how shear zone slip velocity responds to borehole measurements" — what does this mean? Is not shear zone slip velocity itself the quantity measured in the borehole? I suggest the authors clarify this. As already noted, the presentation of results is somehow difficult to follow, being descriptive and lacking a clear framing within a geological landslide reference model. Furthermore, the same analyses are not consistently available for both Jettan and Gàmanjunni (no borehole measurements at the latter), which hampers a complete comparison.

Section 4.1, pages 19–21: As discussed above, I am not convinced that comparing the displacement patterns of the two landslides and their correlations with the considered environmental variables can yield conclusive evidence regarding the influence of permafrost on landslide behaviour. Despite their proximity and generally similar geological settings, the two landslides differ significantly in geometry and structure (as the authors themselves acknowledge on page 20, lines 396–413), and these differences could account for the observed contrasts in behaviour. Moreover, permafrost occurrence at Gàmanjunni is inferred rather than measured. In this context, the interpretation that "a combination of a well-developed shear plane, coherent rock mass movement, and thawing permafrost inside the landslide could be the key factors driving Gàmanjunni’s faster displacements" is reasonable but remains rather generic and poorly conclusive. I suggest the authors elaborate carefully on this and better support their conclusions with data.

Sections 4.2, 4.3 and 4.4: I found these sections very interesting. They provide a rare example of a description of rockslide deformation mechanisms in space and time, based on high-quality multi-parametric monitoring data, and are well supported by data — except for a few statements that appear somewhat speculative (e.g. page 22, lines 475–476). Nevertheless, this part could serve as the subject of a separate paper, as it is dedicated exclusively to the Jettan landslide (where borehole data are available) and diverges from the main narrative and scope of the paper as established in the preceding sections. I encourage the authors to try to reconcile and harmonise the different parts of the manuscript within a single, coherent narrative framework.