Mountain glaciers recouple to atmospheric warming over the twenty-first century

nature climate change Article https://doi.org/10.1038/s41558-025-02449-0 Mountain glaciers recouple to atmospheric warming over the twenty-first century Received: 12 November 2024 Accepted: 1 September 2025 Thomas E. Shaw 1 , Evan S. Miles 2,3,4, Michael McCarthy 1,2, Pascal Buri 5, Nicolas Guyennon 6, Franco Salerno 7,8, Luca Carturan Benjamin Brock 10 & Francesca Pellicciotti 1 , 9 Published online: xx xx xxxx Check for updates Recent studies have argued that air temperatures over many mountain glaciers are decoupled from their surroundings, leading to a local cooling which could slow down melting. Here we use a compilation of on-glacier meteorological observations to assess the extent to which this relationship changes under warming. Statistical modelling of the potential temperature decoupling of the world’s mountain glaciers indicates that currently glacier boundary layers warm ~0.83 °C on average for every degree of ambient temperature rise. Future projections under shared socioeconomic pathway (SSP) climate scenarios SSP 2-4.5 and SSP 5-8.5 indicate that decoupling, and thus relative cooling over glaciers, is maximized during the 2020s and 2030s, before widespread glacier retreat acts to recouple above-glacier air temperatures with its surroundings. This nonlinear feedback will lead to an increased sensitivity to warming from midcentury, with glaciers losing their capacity to affect the local climate and cool themselves. Mountain glaciers play a key role in climate–land interactions, and shape the mountain water cycle1, influence mountain hazards2 and form an essential part of the world’s water towers that meet the needs of billions of people globally3. The decline of glacier health in recent decades4–6 has highlighted their sensitivity to ongoing climate change, and sparked a large amount of research on their future trajectories and the mountain water resources of the twenty-first century7,8. The magnitude and rate of warming is central to understanding and modelling future glacier health, as near-surface air temperature governs both accumulation and ablation and is thus a cornerstone variable in all glaciohydrological modelling frameworks. However, air temperature measured above glacier surfaces (TaGla) often diverges from air temperature of the surrounding non-glacierized terrain (TaAmb), due to the presence of a glacier microclimate which arises from near-surface cooling and the generation of a shallow glacier katabatic wind under warm atmospheric conditions9,10. This ‘temperature decoupling’ (defined here as the ratio of TaGla to TaAmb) can introduce nonlinearities in the response of glaciers to climate warming which translate into an overestimation of snow and icemelt, as well as a distinct pattern of elevation-dependent warming in mountain regions11,12. However, such nonlinearities are a feature largely overlooked by the downscaling and bias correction of meteorological data in current temperature index modelling studies of glaciers. Observations from glaciers of distinct sizes in different climates suggest that summer air temperature over ice is ‘decoupled’ from ambient temperature changes in the glacier’s surroundings, such that TaGla does not experience a 1:1 variability with TaAmb under warm conditions13,14 (Methods, equation (3)). Such analyses of glacier microclimates have, however, typically been restricted to a small number of individual glaciers for given years13,15–17, raising questions about how strongly decoupled TaGla might be on glaciers around the world, what factors control the magnitude of decoupling and how this may evolve with future warming. Institute of Science and Technology Austria, Klosterneuburg, Austria. 2Swiss Federal Institute WSL, Birmensdorf, Switzerland. 3Department of Geography: Glaciology and Geomorphodynamics, University of Zürich, Zurich, Switzerland. 4Department of Geosciences, University of Fribourg, Fribourg, Switzerland. 5Geophysical Institute, University of Alaska Fairbanks, Fairbanks, AK, USA. 6National Research Council, Water Research Institute (IRSA-CNR), Rome, Italy. 7National Research Council, Institute of Polar Sciences (ISP-CNR), Milan, Italy. 8National Research Council, Water Research Institute (IRSA-CNR), Brugherio, Italy. 9Department of Land, Environment, Agriculture and Forestry, University of Padova, Padova, Italy. 10Department of Geography and Environmental Sciences, Northumbria University, Newcastle, UK. e-mail: 1 Nature Climate Change Article https://doi.org/10.1038/s41558-025-02449-0 a b Elevation [m above sea level] 500 3,750 6,000 n 0 ×10 5 2 Latitude (°) Clean Debris –4 –2 0 2 Cooling (°C) c Up k (–) 0.8 Mixed 0.6 0.4 500 m Glacier wind direction (–) 1.0 12,000 m 0.2 0 5 10 15 20 25 30 Down Te (°C) Fig. 1 | Observed patterns of cooling and decoupling. Observational data of on-glacier summer temperatures over debris-free (circles) and debris-covered (triangles) ice surfaces as measured at AWSs. a,b, The location and total data distributions (number of hours) by latitude band are shown (a) with the mean bias of TaGla compared with the respective, off-glacier TaAmb at the same elevation given by latitude band (b) (n = 350). Horizontal dashed lines provide context for the latitudinal bands. c, The temperature decoupling factor k per AWS location (y axis) versus the ambient equivalent temperature (Te) at that AWS (n = 350), where the symbol sizes are scaled by the distance along the glacier flowline of the AWS location and coloured by an index (Methods) expressing the prevalence of observed down-glacier katabatic winds (blue/purple) or up-glacier valley winds (red/orange). Hollow circles indicate where no wind data are observed and horizontal (vertical) error bars indicate the uncertainty (range) of the off-glacier Te estimate (k calculation) from ±25% of the local lapse rate. Arrows highlight the trajectory towards continued decoupling (blue), decoupling and ‘recoupling’ (red) and the trajectory for debris-covered ice (grey). The horizontal solid line gives the 1:1 k value with a mean uncertainty of that line taken from the observations (horizontal dashed lines). Here we compile an inventory of data from 350 on-glacier automatic weather stations (AWS) spanning 62 glaciers in 169 individual summer seasons, with a specific focus on hourly 2-m TaGla compared with its local TaAmb. The ratio of TaGla to TaAmb during the summer (Supplementary Fig. 3) is used to express the magnitudes of ‘temperature decoupling’ (or ‘k’) over different mountain glaciers in space (AWS location) and time (year of observation). We identify the main topographic and hydrometeorological drivers that regulate this decoupling and explore their predictive capability in a statistical model that accounts for its spatiotemporal variability (Methods). After optimizing the model with the dominant predictors, we extend the estimate of decoupling to the world’s mountain glaciers for the me (...truncated)