| 08 Dec 2025
Thaw slump erosion accelerates fluvial sediment transport after a heatwave on the Taymyr Peninsula, Russia
Abstract. Thaw slumps appear to be expanding across much of the Arctic, yet questions remain about the quantity and fate of sediment eroded from these mass-wasting features, its role in downstream material transport, and how erosion evolves after initial failure. Here, we document the watershed-scale consequences of the largest single-initiation thaw slump event to date, which covers more than 30,000 km² on the Taymyr Peninsula in northern Russia. Using automated satellite methods, we track the rapid failure of more than 1,700 individual thaw slumps and record their ongoing post-failure erosion. We use a combination of Landsat and Sentinel-2 data to show that suspended sediment concentrations (SSC) downstream of slump clusters spiked to 2–5x background levels immediately (1–2 days) after the acceleration of thaw slump failure during the 2020 Siberian heat wave. Elevated suspended sediment transport rates scale with the upstream density of slumps and have persisted as slumps continue to erode; sediment transport during the period 2020–2024 is thus unprecedented in the region during the 40-year Landsat record. Although elevated relative to pre-failure, sediment export to the ocean appears to be significantly less than what is transported by rivers into estuaries, suggesting that estuarine storage may account for much of the eroded lost material, potentially transforming estuarine physical processes and threatening aquatic habitat.
General comments
Dethier et al. investigate the geomorphic consequences of a major thaw slump event on the Taymyr Peninsula during the summer 2020 Siberian heat wave. Using automated analyses of Landsat and Sentinel-2 imagery, the authors track the initiation and evolution of ~1,700 thaw slumps and link this disturbance to abrupt, region-wide increases in suspended sediment concentration (SSC) in downstream fluvial networks.
The paper is excellent, thorough, and clearly written. It represents a valuable extension of Erikson et al. (2025) by scaling from the detailed analysis of a single thaw slump up to a regional synthesis. Importantly, this study moves beyond documenting the spatiotemporal distribution of thaw slumps--an emphasis of several recent studies (e.g., Bernhard et al., 2022; Dai et al., 2025)--to explicitly address the transport and fate of thaw-slump-derived sediment through fluvial systems, all the way out to the ocean. I was also impressed by the temporal resolution of the analysis: pinpointing the linkage between slump activity and increased sediment concentration in river networks at a timescale of just days. This kind of temporal resolution will prove valuable in constraining the relative importance of acute, short-lived events (e.g., heat waves, rainstorms) vs. long-term climatic drivers (e.g., decadal warming trends) in causing retrogressive thaw slumps and other types of permafrost degradation/collapse features.
Overall, I recommend this paper for acceptance following revision. My primary comments focus on the authors’ interpretation of the declining satellite-derived SSC near estuary mouths, and specifically their inference that a substantial fraction of thaw-slump-derived sediment is retained in estuarine sinks rather than exported to the ocean. As described below, I wonder to what extent the decline in SSC of surface waters in estuaries is a reflection of enhanced flocculation--and therefore a greater proportion of the sediment transport occurring deeper in the water column, invisible to satellites--rather than a signal of reduced sediment transport through the integrated water column, which is what the authors imply in the current manuscript.
Main conceptual comment:
The central question I kept returning to while reading this manuscript is whether the estuaries are truly trapping ~50% (or more) of the incoming sediment flux, thereby preventing that sediment from reaching the coastal ocean, or whether the observed downstream decline in satellite-derived SSC reflects enhanced flocculation and vertical redistribution of suspended sediment in estuarine waters.
To first order, the degree of flocculation should depend on the suspended sediment concentration, the organic carbon concentration, and the salinity (ion concentration), as well as the hydrodynamic controls (e.g., turbulence) (e.g., Lamb et al., 2020; Zhang et al., 2021; Abolfazli and Strom, 2023; Osborn et al., 2023; Kranck, 1973). In brackish environments, increases in ionic strength can promote rapid aggregation of mud-sized particles, increasing settling velocity and altering the vertical structure of the suspended sediment profile. As a result, a larger fraction of the sediment load may be transported deeper in the water column, below the depth to which satellite color inversion methods are sensitive.
This raises an important question for the interpretation of Figures 3c–d and related analyses. For example, in the pre-perturbation comparison (Fig. 3d), estuaries already exhibit lower estimated SSC than upstream reaches. Following the 2020 disturbance, could the observed decline in surface-water SSC toward estuary mouths partly reflect increased flocculation in response to elevated sediment supply interacting with brackish water, rather than a true decrease in depth-integrated sediment transport? In such a scenario, the estuaries may still be efficiently transmitting sediment seaward, but with a vertical concentration profile (i.e., Rouse profile) that shifts sediment transport to deeper parts of the water column that are effectively invisible to satellite detection.
This question does not affect the excellent analyses presented in the paper. Rather, I think the manuscript would be strengthened by explicitly acknowledging this uncertainty and by clarifying that satellite-derived SSC primarily reflects near-surface conditions. A short discussion of how flocculation and vertical sediment redistribution could complicate interpretations of sediment mass balance in estuaries would provide helpful context and nuance for the apparent estuarine storage signal.
Specific comments
Figure 1: I found the label of “SSC sampling site” (orange triangle) to be a bit misleading, since (at least to me), it implies taking physical water samples and measuring the SSC. Perhaps relabel to “SSC estimation site” or something similar? Likewise, line 159 says “…To take a sample for SSC analysis…” Perhaps rephrase to, “To make a satellite-derived SSC estimate…” or something similar.
Lines 28-30: “Erosion could decrease if vegetation growth is bolstered by warmer temperatures; resulting increases in root strength could increase sediment cohesion, leading to slower hillslope processes, fewer mass failure events, and slower rates of riverbank erosion (Ielpi et al., 2023)”. I know this is an idea that is popular in the literature, but it seems that, for this mechanism to be credible, there needs to be an explanation for how the ~1 meter rooting depth effectively stabilizes the ~10+ meter tall riverbanks of these major rivers.
Lines 277-278: “A tenfold increase in percent slump area corresponds to an average SSC increase of approximately 131 mg/L.” Can you reframe this numerical estimate so that it is dimensionally consistent? That is, the more dimensionally-consistent scaling should go like: SSC [mg/L] ~ rho_s*V_s/Q_w, where rho_s is the sediment density (mass per volume sediment), V_s is the mass of slump material per year, and Q_w is the volume of water discharge per year. Can you use other thaw-slump inventories that estimate retrogressive thaw slump volumes (i.e., from repeat DEMs), or your own DEM analysis, to estimate a simple empirical scaling between measured thaw slump area (reported in this manuscript – e.g., Fig. 1) and thaw slump volume? Similarly, can you estimate the total annual water discharge (Q_w)? This comparison will be interesting because it may reveal the approximate sensitivity of the suspended sediment flux to the input sediment flux (albeit with some uncertainty from the approximations required to estimate river discharge and thaw slump volume). Is this proportionally between input and transmitted sediment flux close to 1, or is it much smaller than 1?
Figure 3: Does the non-linear relationship between thaw slump area and SSC (fitted here as a log-linear relationship between the SSC anomaly and the fractional thaw slump area) indicate that the sediment input from thaw slumps is maxing-out the river’s sediment transport capacity? On a related note, do you see any key morphological changes to the river networks in the last 5 years that reflect the increased sediment influx?
Figure 3a-b: It would be useful to remind readers in this figure what the time-averaged (or pre-2020) average SSC is in each river reach, so that we know how the plotted SSC anomalies in mg/L compare to the pre-perturbation average sediment concentrations.
Lines 421-422: Add references to this sentence?
Lines 425-426: “Still, we note that the estuaries appear to have increased annual storage to accommodate increased sediment influx, aggrading rather than simply transporting these additional loads.” Do you see any direct evidence for aggradation here? I understand you don’t have repeat bathymetric surveys, but how about evidence for sediment infilling based on the accommodation space on the shallow edges of the estuaries being filled with sediment? See my comments above about being worried that the estuary signal could be, in part, a changing vertical distribution of the sediment load in the water column rather than a reduced depth-integrated sediment flux.
Technical corrections and questions
Lines 157-158: “Areas with slopes exceeding 8 m/m (derived from ArcticDEM) were excluded to avoid non-water surfaces.” This slope seems extraordinarily high to me. Is there any rationale for choosing such a high slope cutoff for this processing step of helping to distinguish water vs. land? For example, why not spatially filter (average) the ArcticDEM to remove noise and then use a much smaller slope threshold? No need to re-do the analysis, but some justification might be warranted since 8 m/m is a non-physical water surface slope (other than for waterfalls)!
Lines 190-192: This sentence feels repetitive with line 165.
References
Abolfazli, E. and Strom, K., 2023. Salinity impacts on floc size and growth rate with and without natural organic matter. Journal of Geophysical Research: Oceans, 128(7), p.e2022JC019255.
Bernhard, P., Zwieback, S., Bergner, N., and Hajnsek, I.: Assessing volumetric change distributions and scaling relations of retrogressive thaw slumps across the Arctic, The Cryosphere, 16, 1–15, 2022.
Dai, C., Ward Jones, M. K., van der Sluijs, J., Nesterova, N., Howat, I. M., Liljedahl, A. K., Higman, B., Freymueller, J. T., Kokelj, S. V., and Sriram, S.: Volumetric quantifications and dynamics of areas undergoing retrogressive thaw slumping in the Northern Hemisphere, Nat. Commun., 16, 6795, 2025.
Erikson, C. M., Dethier, E. N., and Renshaw, C. E.: Seasonal dynamics of a coupled hillslope — river system in the Arctic revealed by semi-automated satellite image analysis, Remote Sens. Environ., 328, 114883, 2025.
Kranck, K., 1973. Flocculation of suspended sediment in the sea. Nature, 246(5432), pp.348-350.
Lamb, M. P., de Leeuw, J., Fischer, W. W., Moodie, A. J., Venditti, J. G., Nittrouer, J. A., et al. (2020). Mud in rivers transported as flocculated and suspended bed material. Nature Geoscience, 13(8), 566–570.
Osborn, R., Dunne, K.B., Ashley, T., Nittrouer, J.A. and Strom, K., 2023. The flocculation state of mud in the lowermost freshwater reaches of the Mississippi River: Spatial distribution of sizes, seasonal changes, and their impact on vertical concentration profiles. Journal of Geophysical Research: Earth Surface, 128(7), p.e2022JF006975.
Zhang, Y., Ren, J., Zhang, W. and Wu, J., 2021. Importance of salinity-induced stratification on flocculation in tidal estuaries. Journal of Hydrology, 596, p.126063.