From flooding to ‘greening’ – how ocean waves contribute to the seasonal melting of Antarctic sea ice
June 9, 2026: Australian Antarctic Division
A possible “missing link” in what drives the rapid melt back of Southern Ocean sea ice each summer has been identified in a new study by an international team, led by Australian Antarctic Program* scientists.
Study leader Dr Rob Massom said ocean waves contribute to the seasonal melting of Antarctic sea ice in ways that have been overlooked, until now.
“Ocean waves promote melting at the base and sides of sea-ice floes, by breaking them up and exposing more of their edges to ocean waters that are warmed by sunlight each summer. But this is not the full story,” he said.
“Our new study shows how waves can also cause surface melting of sea-ice floes, by washing over and flooding them, removing their snow cover, and grinding them into slush.
“This ice floe–wave interaction also creates conditions for the rapid growth of algae in both the seawater ponds on the floe surface, and within the ice floe.
“This then enhances melting in a beautiful interplay of physical and biological processes.”
When sea-ice algae proliferate they turn the sea ice green. This darkening of the ice reduces its ability to reflect sunlight, amplifying sea-ice melting in summer. (Photo: Rob Massom)
Australian Antarctic Division sea-ice scientist, Dr Rob Massom, led the study published in The Cryosphere, that identifies previously unconsidered roles for ocean waves in the melting of sea ice. Photo: AAD
Detecting change in planet’s icy pulse
Each year, the annual cycle of sea-ice growth and retreat around Antarctica fluctuates from 18–19 million square kilometres in winter, to 2–3 million square kilometres in summer – one of the largest seasonal changes on Earth.
This “heart beat” of the planet’s climate system moderates global temperatures and drives ocean circulation, and is vital to the survival of biodiverse Southern Ocean marine ecosystems.
“Antarctic sea ice and the snow that accumulates on its surface help keep our planet cool by reflecting sunlight back into space,” Dr Massom said.
“Much of this reflectance is due to the snow cover, which forms one of the brightest natural surfaces on Earth.
“The snow also acts like an insulating blanket, shielding the ice from rising air temperatures in summer.
“Snow-covered ice floes reflect more than 85 per cent of the sunlight hitting them, compared to about 60 per cent for bare, thick sea ice. The darker ocean absorbs about 93 per cent of the sunlight, causing the open water around the floes to seasonally warm.”
Dr Massom said that wave-driven surface melting could help account for the large differences in the timing and magnitude of Summer sea-ice retreat observed by satellites, compared to those simulated by climate models.
“Climate models largely underestimate the average rate of sea-ice retreat observed by satellites each Summer,” he said.
“This suggests incomplete knowledge and representation of important interactions and processes involving the ice, ocean, atmosphere and biota.
“This study provides a possible solution towards addressing a major gap in our ability to accurately model Antarctic sea ice and its seasonal cycle, as we strive to understand the causes of recent dramatic sea-ice losses around Antarctica and reduce uncertainty in model projections of future sea-ice conditions and climate.”
A perfect storm of ‘wave melting’
Using observations, modelling and theoretical insights, the study shows that waves in the stormiest ocean on Earth can wash snow off sea-ice floes, and cause seawater to pool on their surface.
This substantially reduces the ability of the ice to reflect sunlight, causing it to absorb more heat from the sun and melt from the surface down (in addition to the bottom and sideways melting) as Summer progresses.
Wave action also acts like a blender, grinding the ice floes together and creating a ‘wave slush’ that exposes more of the ice to sunlight and the warming ocean.
“The outer sea-ice zone is more than just a collection of snow-covered floes broken up by waves, as is generally accepted to be the case,” Dr Massom said.
“The presence of wave slush creates perfect conditions for sea-ice algae to grow, turning the ice green, reducing its albedo further, and amplifying melting.”
The research team found that these three linked processes – called ‘wave flooding’, ‘wave pulverisation’ and ‘wave greening’ – could enhance the speed of summer melting by between 5.2 cm and 6.1 cm per day in Summer.
“Our calculations suggest that wave melting alone could melt a one metre-thick slab of sea ice in just 20 days, and in about 16 days if amplified by greening,” Dr Massom said.
“These enhanced melt rates are likely to be underestimates, due to potential acceleration and amplification of surface melting by a suite of previously unconsidered positive feedback mechanisms, driven by the same wave processes.”
Seawater ponding on ice floes, and snow loss, reduces the ability of the ice to reflect sunlight, leading to surface melting in summer. (Photo: Rob Massom)
Zoning in on sea-ice melt
These wave melting processes and feedbacks occur in the ‘marginal ice zone’ – the outer part of the sea-ice zone affected by incoming waves from the adjacent, stormy open ocean.
Wave melting may also be important within the interior sea-ice zone, due to the periodic penetration of ocean swells, and wind-driven waves in areas of open water within the sea ice – although more observations are needed to confirm this.
Dr Massom said the research is also relevant to the changing Arctic, where declining sea-ice coverage is opening larger areas of the central Arctic Ocean to wind-generated waves.
He said predicted increases in storminess and waviness over the Southern Ocean are likely to intensify wave melting, greening and associated feedbacks over coming decades, to potentially disrupt the annual sea-ice cycle and cause further sea-ice loss.
“This underlines a need for modelling research, and observations of these previously neglected wave processes and feedbacks, to fully understand and quantify their overall contribution to seasonal sea-ice melting around Antarctica and in the Arctic.
“Sophisticated technologies such as autonomous camera systems on icebreakers are crucial to assisting with the new observations.
“Ultimately, we should encourage and enable the inclusion of wave-melting processes and feedbacks in next-generation climate and Earth-system models.
“This would help improve understanding of recent sea-ice change, and more accurate prediction of the likely fate of polar sea-ice systems and the wider Earth system as the planet continues to warm.”
Further research is needed to determine how wave processes affect the production of sea-ice algae, and their key role in removing carbon dioxide from the atmosphere (to help slow atmospheric warming), supporting krill and polar-marine ecosystems, and driving biogeochemical processes that influence cloud formation and climate.
The study is published today in The Cryosphere and The Conversation.
*Australian Antarctic Program researchers involved in this study are affiliated with the Australian Antarctic Division, Australian Antarctic Program Partnership, and the Australian Centre for Excellence in Antarctic Science.
Learn more about sea ice in the AAD's feature Antarctic sea ice in crisis.
Massom, R. A., Reid, P. A., Warren, S. G., Light, B., Perovich, D. K., Bennetts, L. G., Uotila, P., O'Farrell, S. P., Meylan, M. H., Meiners, K. M., Wongpan, P., Fraser, A. D., Toffoli, A., Passerotti, G., Strutton, P. G., Chua, S. M. T., and Fedrigo, M.: The influence of ocean waves on Antarctic sea-ice albedo and seasonal melting, and potential coupled physical and biological feedbacks, The Cryosphere, 20, 3271–3298, https://doi.org/10.5194/tc-20-3271-2026, 2026.
Abstract
Identifying the full suite of processes that drive the melting of Antarctic sea ice each summer is crucial to improving the currently-poor ability of contemporary models to accurately simulate the climatological retreat phase of the annual sea-ice cycle. This is critical to (1) understanding and attributing observed trends and recent abrupt changes in sea-ice coverage and (2) the more robust prediction of future sea-ice conditions and impacts. This paper identifies wave-driven processes that can accelerate the seasonal melting of sea ice both in the marginal ice zone (MIZ) and in open-water areas within the interior sea-ice zone (SIZ). It builds on the long-held view that seasonal Antarctic sea-ice ablation is primarily driven by ice-floe lateral and basal melting enhanced in the MIZ by wave breakup of ice floes, by demonstrating that ocean waves play important additional roles in generating surface and interior melting (termed “wave melting”) via three sets of processes: “wave flooding”, “wave pulverisation”, and “wave greening” (involving algal proliferation in wave-modified ice). Based on existing observations and simple one-dimensional modelling, these wave processes are estimated to reduce ice albedo by 0.38–0.64 compared to snow-covered ice, resulting in vertical melt-rate enhancements of 0.9–5.2 cm d−1 amplified by wave greening to 1.1–6.1 cm d−1. The study also identifies five positive feedback and sub-feedback mechanisms that likely accelerate the ice melting further. It addresses a gap in current climate and Earth system models, which account for wave effects on floe-size distributions but overlook these coupled wave-driven dynamic, thermodynamic and biological processes that may contribute to explaining why and how Antarctic sea ice can melt back so rapidly each summer. An intention of this foundational study is to stimulate further targeted investigation aimed at quantifying the role of wave melting in the annual sea-ice cycle – as well as the contribution of wave greening to primary production in the sea-ice zone and its role in key biogeochemical processes that feed back to climate. The work has implications for planetary albedo, global climate feedbacks, marine ecosystems, and the accuracy of future sea-ice and climate projections in an increasingly-stormy Southern Ocean, as well as in a changing Arctic.
A schematic showing the multiple ways ocean waves can reduce the reflectivity of sea ice and its snow cover – by washing over and pooling seawater on the surface, deforming the ice to create surface seawater ponds, grinding ice floes together to create a slush, and by ‘greening’ and further darkening the ice. (Photo: From Massom et. al. The Cryosphere, 20, 3271–3298, https://tc.copernicus.org/articles/20/3271/2026/)
Multidecadal Atlantic “Warming Hole” Heat Content Variations Are Caused by Ocean Heat Transport, Not by Surface Fluxes
The Atlantic's cold blob is a sign of weakening of the Atlantic Meridional Overturning Circulation (AMOC), key ocean currents, and a devastating climate tipping point, researchers conclude.
Abstract
The northern Atlantic south of Greenland and Iceland is the only part of the world which has cooled significantly since the 19th Century both in the atmosphere and ocean. The oceanic cooling is widely assumed to be a result of reduced ocean heat transport into this region. However, some studies have suggested it could be due to increased net heat loss at the sea surface. Here we use observation-based reanalysis data of ocean heat content and surface flux changes in this region to show that the observed cooling trend cannot be explained by surface heat flux changes, and that multidecadal heat content variations are generally larger and more tightly correlated with ocean heat transport than with surface heat flux variability.
Plain Language Summary
A region of the northern Atlantic–sometimes called the “cold blob”–has cooled since the 19th Century while the rest of the world has warmed. It is particularly the ocean which has cooled there. Scientists have been discussing whether this is because ocean currents bring less heat into this region, or because more heat is being lost through the sea surface there. An analysis of temperature data sets based on measurements show it is the former–changing ocean heat transport–which dominates heat content changes in the “cold blob.” This is of concern because a further weakening of Atlantic heat transport in future climate change could lead to serious impacts on climate and weather conditions in Europe and other parts of the world.
Introduction
One of the most remarkable features of climate change since the 19th Century is this: while otherwise surface temperature of the entire planet has been warming, a region in the subpolar North Atlantic has not only defied this warming trend but has significantly cooled. This region (shown in Figure 1) is located to the south of Greenland and Iceland and to the west of the British Isles, and it has been dubbed the Atlantic “warming hole” or “cold blob.” This pattern is also seen in surface air temperature trends, see for example figure TS.3 of the IPCC 6th Assessment report (IPCC, 2021).
Figure 1: Sea-surface temperature linear trend (°C) from 1880 to 2025, NASA GISTEMP data (Lenssen et al., 2024). Gray areas indicate missing data. Interactive map generated at https://data.giss.nasa.gov/gistemp/maps/ on 14.1.2026.
(Dima & Lohmann, 2010) analysed patterns of SST variability since 1870 using the empirical orthogonal function method and found that the northern Atlantic cooling is part of a pattern anticorrelated with the South Atlantic. They concluded that this pattern is linked to Atlantic meridional overturning circulation (AMOC) variations, and that the AMOC has been weakening since the 1930s. That is plausible, since the “cold blob” region is where the AMOC delivers its heat and passes it to the atmosphere, and much of this heat is drawn from the South Atlantic and transported northward across the equator (Trenberth & Fasullo, 2017). In fact, that is the main reason why the Northern Hemisphere is 1°–2°C warmer than the Southern Hemisphere (Feulner et al., 2013).
Subsequently, (Drijfhout et al., 2012) analysed temperature trend patterns in observations and historic-forcing simulations of CMIP5 climate models. Using bivariate regression, they demonstrate that the warming hole is associated with the AMOC. This result was supported by (Caesar et al., 2018), who find a strong correlation of AMOC weakening with the cold blob temperature in future global warming simulations of CMIP5 models. Several further studies have likewise concluded that the “cold blob” is of anthropogenic origin (Chemke et al., 2020) and due to AMOC slowdown (Lee et al., 2026; K. Y. Li and Liu, 2025).
Other studies have used observed sea surface salinity changes (Zhu & Liu, 2020) or various types of paleoclimatic proxy data (Caesar et al., 2021; Rahmstorf et al., 2015) to conclude that the AMOC has slowed down since preindustrial time, or further indications of a weakening over more recent decades (Biló et al., 2024; Michel et al., 2025; Pontes & Menviel, 2024; Ren, Li, et al., 2025, Ren, Xie, et al., 2025; Zhu et al., 2023). A further weakening of the AMOC could have major repercussions for future climate for millennia, given that the AMOC is known to have a tipping point beyond which it is likely to shut down, as reviewed in (Rahmstorf, 2024).
Note that the linkage of the “cold blob” to the AMOC refers to the longer-term, multi-decadal evolution. This connection is stronger in the winter half of the year (Caesar et al., 2018) and is expected to involve a time lag of a few years (Caesar et al., 2022). Shorter-term SST variability is likely dominated by weather and thus surface forcing (Fox et al., 2022). That is expected particularly in summer when a shallow and warm surface mixed layer develops, which is more susceptible to surface forcing than to horizontal ocean heat transport. A striking example of this was the summer 2023 with record-breaking sea-surface temperatures in the North Atlantic including in the “cold blob” region, as an exceptionally shallow surface mixed layer–in some areas only 10 m deep–heated up in the summer sun (England et al., 2025). The “cold blob” subsequently reappeared after deep winter mixing.
Nevertheless, it has been proposed based on modeling that at least a part of the long-term cooling trend in the subpolar Atlantic could be due to surface forcing (Fan et al., 2023; He et al., 2022; L. Li et al., 2021). However, climate models disagree on the cause of the “cold blob” (Fan et al., 2024), so that an analysis of observational data is needed. Here, we present such a data analysis to examine the causes of the “cold blob” further.
Results
The normalised SST trends in the Atlantic since 1993 from the Copernicus satellite data are shown in Figure 2. The cold blob is clearly visible also for this time period where high resolution satellite data are available. In addition, we see a strip of strong warming along the American coast north of Cape Hatteras, a feature known to be an “AMOC fingerprint” dynamically linked to an AMOC weakening via a northward shift of the Gulf Stream (Zhang, 2008). Such a shift is also suggested by the below-average warming to the south of the strong warming strip, with some blue patches.
Figure 2: Local sea surface temperature trend divided by the global SST trend for 1993–2021. This normalization is useful for comparing different time intervals and for removing the average global warming trend. Data source: (Copernicus Climate Change Service, 2023).
Another important diagnostic is the heat content change in the water column (Cheng et al., 2022). While the global ocean has generally accumulated heat at a rate of the order of 1 W/m2 during this period, the cold blob region has lost heat (Figure 3). It is thus clear that the cold blob is not merely a surface layer phenomenon. The heat content trend during 1955–2024 over this region is −4.9 ± 2.8 × 1011 W, or on average −0.15 ± 0.09 W/m2.
Figure 3: Trend of ocean heat content in W/m2 in the full-depth water column for 1955–2024, the interval over which these data are considered sufficiently reliable (Cheng et al., 2024). Non-significant trends (90% level) are indicated by stippling. Note that even no trend in the “cold blob” region would be highly relevant when almost the whole globe is warming.
To analyze the role of surface fluxes we use the ERA5 reanalysis data, which combine data and model physics by way of data assimilation as used in weather forecasting. They thus provide the most comprehensive available data set following actual weather systems with hourly output and 31 km horizontal resolution (Hersbach et al., 2020). With surface heat flux we mean the net flux from all contributions: shortwave and longwave radiative fluxes, sensible and latent heat fluxes (positive is downward into the ocean).
Figure 4 maps the trends in SST and in surface heat flux through the ocean surface since 1955 (the period covered by quality reanalysis data) and 1993 (the satellite data period).
Figure 4: Trend in sea surface temperature over 1955–2022 and 1993–2022 (top panels), and trend in surface ocean heat flux over the same periods (bottom), all from ERA5 reanalysis (Hersbach et al., 2020). The contour shows the “cold blob region” as used in the subsequent figures; its exact location depends somewhat on time period but does not affect the results. The contour encircles the region without significant SST warming during 1955–2022 (90% level). While the key feature here is lack of warming, the center of the “cold blob” has cooled significantly.
The ERA5 SST trend shows similar features as the satellite SST trend (Figure 2), including the cold blob and the warm strip along the American coast north of Cape Hatteras, here shown without normalization; hence the orange background shows the general global warming trend.
To explain a cooling trend in the cold blob region by surface heat loss while the AMOC is steady, this heat loss would need to increase to outcompete the AMOC's heat supply. The opposite is seen in the ERA5 data: surface heat loss has in fact decreased (since 1993 significantly, since 1955 slightly) over the cold blob. The latter is to be expected when the AMOC supplies less heat to the region and thus less is released to the atmosphere.
It must be mentioned that reanalysis data for surface fluxes have substantial uncertainties, since unlike SST these fluxes are not measured but modelled using bulk formulas based on atmospheric parameters like differences between air and skin temperature, air and surface humidity, and wind speeds. We have therefore repeated the analysis with the US NCEP/NCAR reanalysis (Kalnay et al., 1996) as well as the Japanese JRA-3Q reanalysis (Kosaka et al., 2024) products, both available from 1955. The flux time series over the “cold blob” for all three reanalysis products are shown in Figure S1. They deviate mainly after 2015; the ERA5 reanalysis used here is very close to the average of all three.
The same argument as for the “cold blob”–with reversed sign–applies to the warm part of the AMOC fingerprint along the American coast. Such a warming could in principle result from decreasing heat loss at the surface–but the data show the opposite trend, namely increasing heat loss. This is to be expected if the Gulf Stream has shifted north and brings more heat to this area.
ARGO data show that the Gulf Stream has in fact shifted north since 2001 (Todd & Ren, 2023). That time interval largely overlaps with the period of direct AMOC observations starting in 2004, in which the AMOC has decreased (McCarthy et al., 2025). That is consistent with the mentioned dynamic link between a weakening AMOC and a northward shift of the Gulf Stream (Zhang, 2008).
The availability of full-depth ocean reanalysis data further allows us to perform a heat-budget analysis for the “cold blob” region. For this purpose, we focus on the ’cold blob’ region shown in Figure 4, where the water column has been losing heat since 1955. Consider the heat budget of the ocean volume under this area:
dHC/dt = OHT + SHF
where HC = heat content, OHT = ocean heat transport into the region and SHF = heat gain from the ocean surface. From the heat content and surface flux data, we can thus calculate ocean heat transport as a residual as:
OHT = dHC/dt - SHF
Figure 5 shows the multidecadal time evolution of these three metrics.
Figure 5: Heat content change, surface heat flux anomaly and implied heat transport anomaly (all given in Watt). The data are 10-year running averages over monthly data to highlight the decadal and longer time scale. To obtain absolute heat loss and transport subtract 1.21 × 1014 W, which is the average heat loss over the full data period. Data sources: heat content IAPv4 (Cheng et al., 2024), surface heat loss ERA5 (Hersbach et al., 2020).
The time-averaged surface heat loss from this area and the corresponding heat transport into the area are around 0.121 Petawatt (in equilibrium, with dHC/dt = 0, these two would balance). The multidecadal variability in OHC cannot be explained by surface heat flux variability and this implies that OHT changes are the main driver of the multidecadal OHC variability. The multidecadal changes of heat content are coherent and largest over the top ∼1,000 m of the water column, which coincides with the thickness of the northward flowing AMOC layer. The anomalies appear to penetrate down from there over a 10-year time scale; below 2,500 m depth we see very little change (Figure 6).
Figure 6: Temperature anomalies in the cold blob region of Figure 4 as function of depth and time. Data source: (Cheng et al., 2024).
The heat content changes are generally larger and more tightly correlated with ocean heat transport than the surface heat loss variations. This is physically expected, since any transport changes affect heat content directly, while they affect surface heat loss only indirectly with delay after the sea surface has warmed. It is also clear that phases of heat content increase (e.g., the one peaking in year 2000) coincide with phases of anomalously large surface heat loss, so that the surface heat flux does not drive heat content change, but rather responds to surface warming. This is a robust result across different reanalysis products despite the substantial uncertainty in reconstructed surface fluxes. The qualitative time evolution of ocean heat transport and heat content change corresponds to the 30-year AMOC reconstruction by (Worthington et al., 2021) based on hydrographic data: low in the 1980s, rising to a peak around 2000, declining until 2010 and then recovering. A similar evolution is found in long-term paleoclimate reconstructions based on sediment data (Caesar et al., 2022).
For the period since 1955, neither of these three curves show a statistically significant trend given the large multidecadal variability. The exact numbers should in any case be treated with caution given their uncertainty, and such a trend calculation would be the trend of the trend for ocean heat content, that is the second derivative, and thus not very robust. The lack of statistical significance over this time period is consistent with the “AMOC index” of (Caesar et al., 2018): this also does not show a statistically significant downward trend from 1955 onwards; the statistical significance of that index essentially derives from it starting in 1870 and the reconstructed AMOC being stronger in the first half of the data series than in the second. In the presence of multidecadal variations, long data series are needed to establish significant trends, and hence a limitation of our study is the lack of below-surface data going back further in time.
Conclusions
The subpolar Atlantic is the only region of the world ocean which has been cooling significantly (Cheng et al., 2024; IPCC, 2021). Our analysis of this “cold blob” and of ERA5 reanalysis data strongly suggest that this is not just a surface phenomenon but a deep-reaching loss of ocean heat content, and that it cannot be explained by increasing surface heat loss but requires declining or weakened lateral heat transport. Surface heat loss appears to respond as a negative feedback to heat content changes: periods of increasing heat content coincide with periods of large surface heat loss. Thus from observational data, we reach the same conclusion as (K. Y. Li and Liu, 2025) did based on the analysis of model results.
Of course, we cannot rule out some contribution to surface heat loss for example from increasing cold winds linked to the positive phase of the North Atlantic Oscillation (Fan et al., 2023). However, to evaluate such a contribution, it is critical to include the effect of changing heat content and lateral heat transport, as these terms show even larger multidecadal variations than the surface flux. Also, the variations in North Atlantic Oscillation may be a delayed response to AMOC variations (Haarsma & Drijfhout, 2025).
Our analysis supports the interpretation of the observed “cold blob” as a sign of a weakening AMOC, which is a major component of the lateral heat transport into that subpolar gyre region. A contribution from increasing lateral heat transport out of the subpolar gyre toward the Nordic Seas has also been suggested (Keil, 2020), and both may well be dynamically linked (Roewer et al., 2026).
There is substantial evidence for a weakening AMOC independent of the “cold blob.” On long time scales this includes paleoclimatic proxy data suggesting the AMOC is at its weakest in a millennium (Caesar et al., 2021, 2022). Also, salinity in the “cold blob” region is at its lowest in 120 years of data, consistent with reduced AMOC salt transport from the subtropical net-evaporation region (Holliday et al., 2020).
On shorter time scales this includes a robust observed weakening of the Gulf Stream over the past 4 decades (Piecuch & Beal, 2023), consistent in magnitude with the 15% AMOC weakening inferred from the subpolar SST data (Caesar et al., 2018), and an ocean density reduction in the subpolar gyre since 1950 which “is suggestive of a long-term AMOC weakening of 2.2 Sv or 13%” (Chafik et al., 2022).
Global warming scenarios of the current climate model generation CMIP6 tend to show an AMOC weakening starting only late in the 20th Century (McCarthy & Caesar, 2023), later than the “cold blob” history suggests. This could be related to radiative forcing issues (aerosol forcing, (Robson et al., 2022)), a generally too stable AMOC in models (in the monostable rather than bistable regime (Arumí-Planas et al., 2024)), or the neglect of increasing Greenland meltwater influx (Pontes & Menviel, 2024).
Given the well-established existence of a tipping point of the AMOC, as well as recent studies finding a range of different “early warning signals” of the ocean circulation approaching such a tipping point (Boers, 2021; Ditlevsen & Ditlevsen, 2023; Michel et al., 2022; van Westen et al., 2024), the strong evidence for a weakening AMOC is a serious concern for society and policy. While large uncertainty remains over how close the Earth is to this tipping point, standard CMIP6 simulations of future global warming scenarios suggest it is crossed in a substantial subset of these model simulations around the middle of this century (Drijfhout et al., 2025; van Westen et al., 2025). From a risk management perspective (Rahmstorf & Zickfeld, 2005), this risk requires urgent attention by policy makers.
Acknowledgments
The authors benefited from stimulating discussions with H. Van den Budenmeyer and helpful reviewer comments. Open Access funding enabled and organized by Projekt DEAL.
Conflict of Interest
The authors declare no conflicts of interest relevant to this study.
Availability Statement
All data used are publicly available. These are:
NASA GISTEMP global temperature data, https://data.giss.nasa.gov/gistemp/maps/.
Copernicus satellite data (Copernicus Climate Change Service, 2023).
ERA5 reanalysis data (Hersbach et al., 2020).
IAPv4 ocean temperature and ocean heat content gridded data set (Cheng et al., 2024).
Rahmstorf, S., Jendrkowiak, J., Gou, R., Cheng, L., Ruiz-Angulo, A., & Björnsson, H. (2026). Multidecadal Atlantic “warming hole” heat content variations are caused by ocean heat transport, not by surface fluxes. Geophysical Research Letters, 53, e2025GL118383. https://doi.org/10.1029/2025GL118383
