1.2.2.3 Glaciers

Glaciers outside the Greenland and Antarctic ice sheets (here termed mountain glaciers) are spread over high altitudes and high latitudes. A range of processes contribute to their individual mass balances, most notably solid precipitation (mainly snow) and surface melt, but also, among others, calving into lakes or ocean (Hock et al., 2019; Meredith et al., 2019). Mass balance thresholds and feedbacks may impact individual glaciers, but when aggregated to the global scale glacier changes are projected to respond relatively linearly this century (Rounce et al., 2023). At longer timescales and higher warming levels, nonlinear characteristics are projected as glaciers disappear (Marzeion et al., 2018). 

Evidence for tipping dynamics

In glaciers close to the melting point, the physical nature of ice inherently involves nonlinear feedbacks, in particular related to interactions between ice and water such as enhanced subaqueous ice melt, heat transport into the ice, or lubrication at the glacier bed. Such feedbacks act typically on the spatial scale of individual glaciers (Figure 1.2.10).

Dynamic instabilities of glaciers such as surges or even catastrophic detachments, but also less pronounced ice velocity fluctuations, can be related to increased melt-water production through positive/amplifying feedback mechanisms (Truffer et al., 2021; Kääb et al., 2021). However, these processes are still not very well understood and there is little evidence so far indicating that such processes could act synchronously over entire glacier regions (Kääb et al., 2023). On a regional scale, loss of ice thickness appears to rather reduce glacier flow speeds (Dehecq et al., 2019). Significantly increased ice flux, such as through surges, transports ice from high-elevation zones characterised by low rates of ice melt (ablation) to low-elevation zones with high ablation rates. 

In contrast, retreat rates of calving glaciers, most of them found in polar regions, are understood to be governed by a feedback where a thinning of the glacier tongue (the narrow floating part of a glacier extending into the sea or a lake) leads to loss of glacier grounding at a topographic pinning point (places where a ridge or valley narrowness slows down glacier flow). This loss of pinning leads to accelerated glacier retreat, associated with increased ice flow velocities, calving rates and further thinning of the tongue, until they stabilise again at a new pinning point or retreat out of the water (Strozzi et al., 2017; Kochtitzky et al., 2022a). Once a destabilisation threshold is passed through processes at the ice-ocean or ice-atmosphere interface, the retreat phase is largely self-perpetuating, independent of climatic conditions or their changes (Pfeffer, 2007). In turn, calving glaciers need typically substantial positive mass balances in order to advance through deep water to a new pinning point. Nonlinear enhanced retreats of calving fronts can be roughly synchronised on regional levels and are in fact a significant component of the current mass loss of polar glaciers, roughly 20-25 per cent (Kochtitzky et al., 2022b).

Glaciers impact atmospheric conditions at their surface by increasing local surface altitude, enabling a feedback between surface elevation and mass balance. Ice thinning can drop glaciers into higher melt (‘ablation’) zones, while a rise in the equilibrium line altitude (ELA – the elevation where local mass balance, i.e. snow input versus melt output, is zero) can shift glaciers into lower snow accumulation zones, with both potentially leading to disproportionately large shifts when large areas of glacier are concentrated in narrow elevation bands. These elevation feedbacks could possibly be regionally synchronised at similar global warming levels, for instance for Arctic ice caps. These effects are typically included in regional and global glacier mass balance models and thus in projections (Rounce et al., 2023; Marzeion et al., 2020).

Reduced glacier albedo, for instance from deposition of dust, black carbon or thin debris, but also through reduced snow cover, significantly increases glacier mass loss (Cook et al., 2017; Naegeli and Huss 2017). Related mass balance feedbacks can happen when years with particularly negative mass balance lead to enhanced accumulation of albedo-reducing matter on the glacier surface, enhancing in turn glacier ablation (Gabbi et al., 2015). Another type of positive/amplifying feedback is deposition of wind-driven dust originating from adjacent mountain areas, a process that is believed to increase with continued uncovering of glacial sediments from ice and snow. Such feedbacks involving albedo can be assumed to affect nearby glaciers in similar ways, and thus represent potential regional effects that are not included in large-scale models yet.              

On local scales, abrupt permafrost thaw processes creating ‘thermokarst’ features (see 1.2.2.4) can be self-perpetuating by enhancing the ice melt in particular of low-angle glacier tongues with low ice flow speeds. Such processes particularly impact debris-covered glaciers, most prominently through the growth of supraglacial ponds on them. There is evidence that such thermokarst processes can enhance glacier ablation on regional scales (Kääb et al., 2012; Buri et al., 2016, Compagno et al., 2019).      

Glacier shrinkage has a range of local to global effects. Several types of glacier hazards can increase in frequency and magnitude as a consequence of glacier retreat, such as debris flows or rock slides and rock avalanches (Hock et al., 2019). Slope instabilities and the uncovering of formerly ice-covered areas leads to increased mobilisation of sediments with both negative (e.g. sedimentation of river infrastructure) and positive (e.g. release of nutrients) downstream impacts. Also the formation of glacier lakes, and thus the potential for glacier lake outbursts, is associated with glacier retreat (Carrivick and Tweed 2016; Linsbauer et al., 2016). 

Changes in glacier river runoff can have impacts on ecosystems (Bosson et al., 2023) and humans, in particular where dry-season water supply is to a large extent depending on glacier ablation. Whereas peak water – the shift from increased runoff from enhanced glacier melt to reduced runoff under continued shrinking of glacier ares – constitutes on regional scales a soft decadal-scale transition rather than a threshold (Huss and Hock 2018), drastic declines of dry-season glacier melt runoff can exert strong pressure on ecosystems, hydropower production and irrigation, for example (Hock et al., 2019). It is important to note that the significance of glacier runoff for downstream areas depends on the seasonally variable percentage of glacier runoff in comparison to other sources of runoff, such as liquid precipitation or snow melt (Kaser et al., 2010). Measurements and projections of glacier mass loss alone are thus only meaningful in relation to potential impacts as part of a seasonally resolved hydrological balance. On longer time-scales and regional spatial scales, pronounced regional glacier shrinkage (or even partial disappearance of glaciers) leads to a transition from glacier-dominated to paraglacial landscape systems, with fundamental changes in all abiotic and biotic processes in the region and its downstream areas (Knight and Harrison, 2016).   

Such a transition to a paraglacial landscape system may exhibit threshold-like behaviour, if climate change is happening rapidly relative to glacier response times, which can span from decades to centuries (Jóhannesson et al., 1989, Haeberli and Hoelzle, 1995). The lagged response of glaciers can lead to a substantial disequilibrium between glacier extent and concurrent climate conditions, such that a large part of a glacier’s mass is committed to be lost, even though this loss has not yet been realised. On the global scale, the committed mass loss for present-day glaciers is estimated around 30 per cent (Bahr et al., 2009, Mernhild et al., 2013, Marzeion et al., 2018), but regionally it can be substantially higher (~60 per cent in central/northern Europe and ~50 per cent in western Canada/US). 

Sea level contribution represents the most global but also most integrating consequence of global glacier mass loss and does not show threshold behaviour because any positive/amplifying feedbacks acting at the glacier or regional scale are averaged out in the huge ensemble of individual glaciers (c. 200,000) (Hock et al., 2019, Marzeion et al., 2020, Hugonnet et al., 2021).

Assessment and knowledge gaps

Glacier shrinkage involves a number of nonlinear, self-perpetuating processes that mostly act on local scales. Few of these feedbacks seem to be able to reach magnitudes and regional synchronisations substantial enough to enhance regional glacier shrinkage in a nonlinear way. However, the potentially large disequilibrium between glacier extent and concurrent climate implies that, regionally, glaciers may be synchronously transitioning from one state to another, even if the individual glaciers’ tipping points are distributed over a broad temperature range. Such effects might explain the almost synchronous retreat of Arctic tidewater glaciers (Kochtitzky et al., 2022a, Malles et al., 2023). Elsewhere, glacier shrinkage is mostly a reversible response to climatic change, despite the irreversible changes that may happen on local scales, such as glacier-related slope failures. 

Glaciers can recover from mass loss, but may need much more time for recovery than for melt. Reversibility of biophysical or social downstream effects of glacier shrinkage also requires long timescales (Hock et al., 2019). It is also important to note that a number of negative damping feedbacks are involved in glacier response to atmospheric warming – most importantly the retreat of glaciers to higher elevations, where they experience lower melt rates, or the thickening of insulating debris covers related to increased production of debris associated with reduced ice cover and permafrost on adjacent mountain flanks (e.g., Compagno et al., 2022). We assess with medium confidence that, while glaciers are not tipping points on a global scale, at a regional scale they may be subject to self-sustained retreat tipping points. A number of the aforementioned glacier feedback processes are not, or not adequately, represented in numerical models. This limitation of models is motivated by the complexity of the processes and the lack of ability to resolve the relevant local scale in the atmosphere and ocean models providing the boundary conditions for the glacier models. The current regional or global glacier projections are struggling to predict the integrated behaviour of local feedbacks and their interactions accurately, and the thresholds and timescales at which slow but nonlinear associated responses of glaciers might emerge are not well known. First results from recent advances in the representation of local feedbacks indicate so far that also in the future the positive/amplifying feedbacks are mostly relevant at the local scale, hardly affecting regional and global scale projections (Compagno et al., 2022, Malles et al., 2023).