Answers are grounded in this paper and cite their sources.
Warming is destabilizing the mountain cryosphere and triggering more frequent hazard events. These hazards mobilize large stores of carbon as well as rock, ice and water. Yet mountain carbon budgets largely overlook this flux, because routine monitoring rarely operates at the time, location and resolution that hazard events demand. We argue that cryosphere hazards should be treated as carbon events and set out how to close the gap. Here, cryosphere hazard events are mass movements in glacial and permafrost terrain whose onset is abrupt relative to gradual degradation, together with the cascades they trigger (Ding et al., 2021;Zhang et al., 2025). They range from rock-ice avalanches and glacial lake outburst floods (GLOFs) lasting hours to retrogressive thaw slumps that develop within days to years as a form of abrupt thaw (Turetsky et al., 2020).At high elevations, warming is often amplified (Pepin et al., 2015(Pepin et al., , 2025)). Global glacier mass-loss rates rose by 36 ± 10% between 2000-2011and 2012-2023(Zemp et al., 2025)), and mountain permafrost is warming (Wu and Zhang, 2008;Noetzli et al., 2024) and thawing abruptly (Turetsky et al., 2020). GLOF frequency in the Third Pole has risen since 1980 (Zhang et al., 2023), and global GLOFs from moraine-dammed lakes increased from 5.2 to 15.2 per year between 1981-1990and 2011-2020(Zhang et al., 2025)). Retrogressive thaw slumps are spreading rapidly in alpine permafrost (Xia et al., 2024), driving cascading damage locally and downstream (Ding et al., 2021;Shugar et al., 2021). Existing frameworks recognize erosion, lateral transport, and outgassing as carbon pathways (Chapin et al., 2006;Hilton and West, 2020;Regnier et al., 2022), but lack a way to quantify hazard-driven mobilization and trace its fate (Figure 1a).A complete net ecosystem carbon budget needs vertical exchange from eddy covariance and chambers, and lateral export and outgassing from river and lake sampling (Chapin et al., 2006;Keenan and Williams, 2018), yet these observations rarely capture brief events or hillslope-channel transport.Three monitoring problems limit how cryosphere hazards enter carbon budgets. Temporal aliasing means periodic monitoring misses an event when the sampling interval exceeds a hazard-driven concentration peak. Extrapolation error means that, once a hazard resets sediment and carbon availability, pre-event concentration-discharge relationships no longer predict post-event fluxes. Material selectivity means that sampling of dissolved and suspended loads excludes carbon moved as woody debris, coarse bedload, or buried in deposits. No single method can constrain these fluxes: atmospheric observations cannot resolve lateral transfers, and satellite erosion detection cannot quantify the concentration or reactivity of displaced material. Figure 1. (a) Conceptual diagram showing cryosphere hazards-induced mobilization, transport, and transformation of carbon along a hazard-to-fate cascade. Cryosphere hazards, including glacier detachment, rock-ice avalanche, glacial lake outburst flood (GLOF) and thaw slump, connect mountain carbon stores to downstream environments. Cascading flows entrain carbon from soils, rock debris and vegetation, transporting it through channels to valley floors and receiving waters. Along these pathways, decomposition can release carbon dioxide (CO2) and methane (CH4), whereas deposition and burial can retain carbon and vegetation recovery can restore CO2 uptake. Subsequent remobilization can expose stored material to further transport and decomposition, extending carbon effects beyond the initial disturbance. Whether a disturbance increases or decreases net carbon transfer to the atmosphere depends on carbon fate and ecosystem uptake, evaluated against a no-event trajectory over the same area and period. The diagram presents a conceptual accounting framework, not a quantified carbon budget, and unquantified pathways should not be read as zero fluxes. Over an accounting period, stock change within the boundary is ΔCstock = -Fatm -Fexport (Eq. 1), where Fatm is the cumulative net carbon flux to the atmosphere and Fexport is the net lateral export across the boundary, both positive when carbon leaves the system; burial within the boundary redistributes carbon within ΔCstock. The event term is □□!"!#$ = ∑ [□□ %$&,!"!#$($) -□□ %$&,*!+($) ] $ ! $ "(Eq. 2), relative to a no-event trajectory; Eevent > 0 indicates a source. POC, particulate organic carbon; DOC, dissolved organic carbon; DIC, dissolved inorganic carbon. (b) Gaps in representing cryosphere hazard events in mountain carbon budgets and ways forward.The consequence is systematic omission rather than random error. Hazard-driven fluxes are rare, brief and often occur in inaccessible terrain, so the disturbance-transfer term in the net ecosystem carbon balance usually lacks an empirical estimate and is underestimated. Known hazard fluxes come almost entirely from bespoke post-event surveys; after a GLOF, channelarmour disruption exposed sediment invisible to pre-event rating curves (Cook et al., 2018).The magnitude missed varies, so no single correction factor works across events. Earthquakeand storm-triggered landslides, used here as process analogues because their carbon export has been measured, show this variability. Landslide pulses increased biospheric carbon export (Hilton et al., 2008;Wang et al., 2016) and accounted for roughly 40% of organic carbon export in two New Zealand catchments (Howarth et al., 2025), whereas the Gorkha earthquake produced no detectable perturbation (Märki et al., 2021). Atmospheric inversions constrain aggregate CO2 exchange without attributing sources (Ciais et al., 2022), so a hazard signal may already sit within them unresolved.Cryosphere hazard cascades mobilize carbon spanning orders of magnitude in age and lability, from millennial permafrost organic matter and petrogenic rock carbon to recently fixed vegetation (Figure 1a). Source areas above treeline hold little biomass, but downstream cascades cross carbon-rich soils and forests. The 2023 South Lhonak disaster released about 50 million m 3 of water and eroded about 270 million m 3 of sediment (Sattar et al., 2025), yet its carbon was not estimated. A simple calculation shows the scale at stake. Assume a dry bulk density of 1.3-1.8 t/m 3 (Ji et al., 2020) and an organic carbon content of 0.43-0.79% for landslide debris from the Tibetan Plateau (Wang et al., 2023). The ~91 million m 3 of debris (excluding ice) detached at the China-Nepal border in 2026 (Li et al., 2026) would then hold roughly 0.51-1.29 Tg of organic carbon. That gross lateral mobilization is about 2.5-6.5 times the annual biospheric organic carbon export of the much larger Narayani catchment (Märki et al., 2021, their Fig. 4), mobilized in a few hours. This ratio is an upper bound, because the debris carbon also includes petrogenic carbon. Whether this carbon is flushed downstream, buried, or oxidized in the valley, it falls outside budgets built on routine river sampling.Mobilization changes the conditions that regulate decomposition. Aggregate disruption and exposure can increase microbial access, whereas mineral association and rapid burial may protect organic matter (Hilton and West, 2020). Carbon age alone does not determine lability.Old yet bioavailable dissolved organic matter occurs in Alaskan glacial runoff (Hood et al., 2009), and old carbon contributes substantially to river CO2 emissions (Dean et al., 2025). At permafrost thaw slumps, vegetation-free surfaces reduced respiration by about one-half but gross primary productivity by about four-fifths (Jiang et al., 2026a), and a plateau-wide synthesis found lower respiration in collapsed areas but increased CH4 release in alpine swamp meadows (Jia et al., 2026). Floods bury soil and biomass in valley fills, lakes and reservoirs, while channel reworking can later expose them. Drained thermokarst and proglacial lake beds and debris-flow surfaces are a particular blind spot, because dry inland waters can still emit carbon (Keller et al., 2020). The fate of mobilized carbon depends jointly on transport speed and reaction kinetics. Rapid flushing through cold, sediment-rich channels can export carbon before substantial mineralization, whereas retention in oxygenated sediments prolongs microbial breakdown.Hazards also mobilize inorganic carbon, exposing reactive minerals and changing waterrock interactions (Hawkings et al., 2026). Carbonic acid weathering and carbonation can consume CO2 (Styllas et al., 2026), whereas sulfide oxidation coupled to carbonate dissolution can release it (Hilton and West, 2020), and oxidation of petrogenic organic carbon is another CO2 source (Zondervan et al., 2023). In Tibetan Plateau rivers, net weathering is estimated to offset about 35% of river CO2 emissions (Zhang et al., 2026). Because hazard events mix inorganic and organic carbon, assessments based only on organic carbon decomposition may miss part of the response (Middelburg, 2024). The same mobilization couples carbon with other element cycles: permafrost thaw can increase N2O emissions (Voigt et al., 2020), glacial erosion exports reactive phosphorus (Hawkings et al., 2016), and thaw slumps release permafrost mercury, sharply raising methylmercury concentrations downstream (Schuster et al., 2018;St. Pierre et al., 2018).Whether a hazard event is a carbon source or sink depends on the quantity, boundary and processes assessed. Gross mobilization is the displacement of carbon from its original location, outlet export its transfer across a defined boundary, and net atmospheric exchange its transfer to or from the atmosphere. Redeposition within the same boundary redistributes carbon without creating a landscape sink, so these quantities should be accounted for separately and reconciled through mass balance (Eq. 1, Figure 1). We define the event term (Eq. 2, Figure 1) as cumulative net atmospheric carbon exchange with the event minus that under a specified no-event trajectory. Positive values indicate additional release or reduced uptake (a source), and negative values indicate additional uptake or avoided release (a sink). The reference site should share the affected site's warming and hydrology, differing only in the absence of the hazard. An event can thus increase atmospheric carbon relative to this reference while the landscape remains a net sink. Burial and ecological recovery can offset part of the disturbance, although forested-landslide evidence should be applied cautiously to alpine terrain. Post-Wenchuan sediment storage and revegetation raised carbon mass by about 10% (Liu et al., 2026), whereas greenness recovery after tundra thaw slumps took a decade at low-Arctic sites and more than 30 years at high-Arctic and highelevation sites (Xia et al., 2026). Greenness recovery does not imply recovered soil carbon stocks or net ecosystem exchange, so persistent productivity losses may matter even when deposited organic matter decomposes slowly.The atmospheric benefit of burial also depends on the reference trajectory. Relocating organic matter that would otherwise have remained protected adds little, whereas preserving material that would otherwise decompose avoids emissions (Galy et al., 2007). Only regrowth exceeding the no-event trajectory lowers the event term, and preserving petrogenic organic carbon averts a geological release without removing fixed atmospheric carbon. These distinctions keep sediment accumulation from supporting unwarranted sequestration claims.Field observations constrain exchange at some thaw slumps, geomorphic reconstructions constrain mobilization pathways, and downstream studies constrain delayed transport, yet few link all three within one event. Assessments should also report CO2-C and CH4-C separately, and converting CH4 emissions into a forcing metric requires an explicit metric and time horizon.We propose four priorities to close this gap (Figure 1b). First, establish a common accounting basis. Each assessment should specify its spatial boundary, accounting period, carbon forms and disturbance coverage, carry unsampled intervals into the uncertainty (Chapin et al., 2006;Ciais et al., 2022), and report mobilization, boundary export and the event term separately (Figure 1, Eqs. 1-2). This separates fluxes missing from an assessment from disturbance effects included but not attributed. Shared event identifiers should connect hazard inventories, water samples, deposit surveys and gas-flux records, distinguishing new erosion from remobilization of previously surveyed material (Jiang et al., 2026b).Second, trace carbon from source areas through transport pathways to deposits. Converting eroded volumes into carbon mass requires source material measurements that capture differences among lithologies, soil horizons, particle sizes and woody debris, with uncertainty in ice content and dry density propagated. Provenance measurements should distinguish biospheric from petrogenic carbon, because depleted radiocarbon alone cannot separate the two, and water chemistry should constrain weathering pathways and their CO2 uptake or release (Hilton and West, 2020). Post-event deposits record retained material but cannot reconstruct the dissolved and fine particulate carbon exported during the event (Märki et al., 2021).Third, connect short-lived transport pulses to longer-term carbon effects. Pre-positioned automatic samplers and redundant stream sensors can capture flood dynamics where deployment is safe (Cook et al., 2018). Later observations should track channel deposits, exposed lake beds, soil gas exchange and vegetation through successive seasons (Keller et al., 2020;Xia et al., 2026). Paired disturbed and reference sites separate event effects from regional change, and sediment cores can extend instrumental records if chronology and postdepositional carbon loss are treated explicitly (Howarth et al., 2025).Fourth, represent event diversity and overlapping legacies in regional models. Multiplying event frequency by a single mean carbon yield obscures differences among carbon-poor headwaters, vegetated valleys and efficient depositional traps. Models should link mobilization to carbon provenance, connectivity, transformation and ecological recovery, while allowing event recurrence to change and later floods to remobilize earlier deposits. Comparison with independent outlet fluxes or stock changes over a common period can show whether explicit event representation improves budget estimates.The approach generates testable expectations. Events crossing carbon-rich terrain with direct channel connections should mobilize more biospheric carbon than events confined to mineral debris (Hilton and West, 2020). Stable deposits should retain carbon longer than frequently reworked channels, and slow revegetation should prolong reductions in uptake.Testing these expectations would show where event accounting changes the inferred mountain carbon balance. Cryosphere hazards are becoming more frequent as warming continues (Ding et al., 2021;Zhang et al., 2025;Jiang et al., 2026b;Richter, 2026), so the carbon they mobilize is a growing budget term, and closing the observational gap is necessary to determine its climate feedback.
No takes yet. Share an insight, caveat, or question.
Song et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: