1. Introduction
Despite anthropogenic global warming, the subpolar North Atlantic has cooled (or warmed less than the global average) in both observations and climate models. This is commonly referred to as the North Atlantic warming hole (NAWH). Understanding the processes responsible for the formation and projected future of the NAWH is important because the NAWH has significant impacts on the North Atlantic jet, Eurasian surface temperatures, and European precipitation (Gervais et al. 2019, 2020). The observed NAWH over 1982–2017 cannot be simulated from internal variability alone, implying a role for external forcing (Chemke et al. 2020). Time-varying aerosol forcing may play a role in the future evolution of the warming hole (Dagan et al. 2020; Fiedler and Putrasahan 2021; Kusakabe and Takemura 2023), although climate models without aerosol forcing can still simulate a NAWH (Dagan et al. 2020). In long climate model simulations under continued greenhouse gas forcing, the NAWH is a transient feature (Huang et al. 2020), suggesting that it will decay on some time scale under continued forcing. Because of the multiple time scales of responses, we distinguish between processes likely relevant over the historical period (here defined as 1850–2015) and those that may become important into the future (here defined using climate projections that extend from 2015 to 2100).
In observations and climate models over the historical period, several mechanisms have been shown to be possible causes of the NAWH. These include ocean dynamical changes, atmospheric drivers, and coupled ocean–atmospheric mechanisms. The warming hole has been connected to a decline in the Atlantic meridional overturning circulation (AMOC) (Caesar et al. 2018) and an accompanying reduction in poleward ocean heat transport (Chemke et al. 2020; Hu et al. 2007). However, it is unclear whether a weakening of the AMOC has occurred in the observational record (Fu et al. 2020; Kilbourne et al. 2022) and over what time scale such a change might occur (Lobelle et al. 2020; Ferster et al. 2022). Detection of an AMOC decline over the historical period is further complicated by model biases which are common at the model resolutions often used in climate change research (e.g., IPCC-class models) (Roberts et al. 2020; Cheng et al. 2013). For example, in the model analyzed in this study, the Community Earth System Model, version 2 (CESM2), the AMOC is 2–3 Sv (1 Sv ≡ 106 m3 s−1) stronger than in observations (Danabasoglu et al. 2020). Additionally, in models, the warming hole cannot be fully attributed to an AMOC decline over the historical period, over which AMOC changes are modest (Drijfhout et al. 2012; Keil et al. 2020; Menary et al. 2020). Some studies suggest that the NAWH emergence can precede an AMOC weakening, as a reduction in deep convection can lead to both the NAWH and the AMOC weakening (Gervais et al. 2018). Ocean dynamical changes besides AMOC changes can also induce a warming hole. Enhanced poleward heat transport out of the warming hole region, toward higher latitudes, is a more robust change in a historical ensemble than AMOC changes (Keil et al. 2020).
Atmospheric drivers are another possible cause of the NAWH over the historical period. Increased surface westerlies over the subpolar North Atlantic lead to increased turbulent heat fluxes out of the ocean, inducing a relative cooling (He et al. 2022; Fan et al. 2023). Additionally, an increase in storminess and associated increases in surface eddy kinetic energy may increase surface heat loss from the ocean (Li et al. 2022). Hu and Fedorov (2020) suggest, instead, that turbulent heat fluxes mostly damp the warming hole, and the warming hole results from a decrease in Ekman heat transport convergence in the subpolar North Atlantic. A role for Ekman heat transport changes would implicate both atmospheric drivers, which alter the surface winds, and the dynamic ocean response to surface wind changes. Another modeling study argues that changes to North Atlantic sea surface temperatures (SSTs) lead to a southward shift of the subtropical gyre and a reduction in the distance that heat is transported poleward into the subpolar gyre (Karnauskas et al. 2021). This could induce a cooling feature similar to the NAWH.
The cause of increased westerlies over the subpolar North Atlantic over the historical period is not well understood. Hu and Fedorov (2020) suggest that the increase in the westerlies is related to warming in the tropical Indian Ocean. He et al. (2022) show that increased wind speeds occur in models without a dynamical ocean, which suggests the wind changes are not a response to SST anomalies with oceanic origin. Regardless of the cause, the ocean response to changes to westerlies will emerge on several time scales. Changes to Ekman heat transport would be near instantaneous, while changes to the geostrophic gyre circulation would emerge over a decadelong adjustment period.
On multidecadal time scales under prolonged anthropogenic forcing, modeling studies document the importance of changes in ocean currents in redistributing heat and leading to delayed warming in the subpolar North Atlantic (Marshall et al. 2015; Winton et al. 2013). The warming hole has been related to a decline in the AMOC (IPCC 2023; Weijer et al. 2020; Roberts et al. 2020; Bellomo et al. 2021) and an accompanying decline in poleward ocean heat transport in future scenarios (Drijfhout et al. 2012; Caesar et al. 2018; Rahmstorf et al. 2015; Menary and Wood 2018; Liu et al. 2020). At the same time, SST anomalies are expected to strengthen under prolonged forcing and may have significant impacts onto the overlying atmosphere. Tropical convection increases latent heat release, enhancing the temperature gradient aloft, while Arctic amplification decreases the temperature gradient at lower levels, with opposing effects on the North Atlantic jet (Shaw et al. 2016). Any change to the North Atlantic jet could alter surface winds and thus change both turbulent heat fluxes and wind-driven ocean circulation. Additionally, the warming hole itself can induce significant changes to the overlying atmosphere, including enhancing the North Atlantic jet and intensifying the surface westerlies (Gervais et al. 2019). These atmospheric changes can then further alter North Atlantic SSTs.
A recent study on SSTs in the subpolar North Atlantic proposes a positive feedback loop that can lead to enhanced internal variability in the warming hole development (Gu et al. 2024). This feedback loop can only operate once the Labrador Sea vertical salinity gradient increases, due to melting sea ice (Jahn and Holland 2013; Gervais et al. 2018) and enhanced river runoff (Nummelin et al. 2016). Once this occurs, stochastic atmospheric variability linked to the North Atlantic Oscillation (NAO) modifies the vertical mixing in the Labrador Sea. A negative NAO leads to less heat being mixed upward in the Labrador Sea, resulting in cooling that spreads via the mean ocean circulation and a deceleration of sea ice melt. This feedback loop depends on wind-driven mechanical stirring.
Thus, modeling studies have established that a warming hole can be caused by changes to buoyancy-driven ocean dynamics, wind-driven ocean dynamics, or atmospheric processes. Regardless of the initial cause of the NAWH, other processes may alter the warming hole as it evolves through time under continued external forcing. There is evidence that the NAWH is fundamentally a coupled ocean–atmosphere process (Gervais et al. 2019; Karnauskas et al. 2021). Under this hypothesis, the NAWH impacts the overlying atmosphere through linear and nonlinear responses to altered SSTs, and these atmospheric circulation changes further impact the ocean through changes in the overlying wind stress. This hypothesis complicates attribution of ocean versus atmosphere in the formation and development of the NAWH, as correlations may exist without establishing causality. The total effect of wind-driven ocean circulation changes onto the warming hole is not well quantified. To address this, we utilize a hierarchy of coupled model ensemble experiments designed to isolate the impact of externally forced wind-driven ocean circulation changes on climate.
2. Methods
a. Model experiments
We compare ensembles of two different coupled model experiments within CESM2 (Danabasoglu et al. 2020), which has a nominally 1° degree horizontal resolution in the ocean and atmosphere. Both ensembles are forced by historical forcing for 1850–2014, followed by shared socioeconomic pathway (SSP) 3–7.0 for 2015–2100. SSP3–7.0 is a medium-to-high emission scenario (see Riahi et al. 2017; O’Neill et al. 2016). Both ensembles use smoothed biomass forcing to avoid spurious warming in the late historical era (Fasullo et al. 2022).
The fully coupled model (FCM) is the typical set up of CESM2, similar to the set up submitted to phase 6 of the Coupled Model Intercomparison Project (CMIP6). It consists of atmosphere, ocean, sea ice, and land models, which exchange fluxes through the Common Infrastructure for Modeling the Earth (CIME) coupler. Coupling between the Greenland Ice Sheet and the atmosphere and land models are supported within CESM2, but land ice was fixed in the ensemble experiments used here, due to the high computational cost of land ice coupling. The time-varying wind stress (momentum flux) is passed from the atmosphere to the ocean at each hourly time step. The time-varying buoyancy fluxes are calculated using the time-varying wind speed and ocean and atmosphere temperature and moisture differences at each hourly time step. These buoyancy fluxes are then exchanged between the ocean and atmosphere models through the model coupler. There are 50 FCM ensemble members. The ensemble members used are part of the CESM2 large ensemble (Rodgers et al. 2021).
The mechanically decoupled model (MDM) includes the exact same model components as the FCM, except that the wind stress (momentum) forcing on the ocean is fixed to a repeating 6-hourly climatology calculated from 50 years of a preindustrial version of the FCM (see McMonigal et al. 2023). The MDM atmosphere remains freely evolving, producing fully time-varying wind stress. In the typical set up of CESM2, such as the FCM, the atmosphere passes a time-varying wind stress to the model flux coupler, and the model coupler then passes this value to the ocean model. However, in the MDM, the wind stress passed to the ocean model from the flux coupler is overwritten with the FCM climatology. Therefore, the MDM ocean feels a climatological wind stress forcing which cannot be altered by internal variability or by external forcing. All other components in the MDM are fully coupled and time varying. Importantly, the buoyancy fluxes in the MDM are fully time varying, using the time-varying wind speed simulated by the atmosphere in the bulk formulas. Oceanic processes in the MDM can still impact the atmosphere; for example, variance in SSTs due to buoyancy-driven ocean circulation changes can alter overlying winds, heat fluxes, and wind stresses. However, wind stress changes in the atmosphere are unable to alter the ocean. The mean state of the MDM AMOC is similar to the FCM AMOC (Larson et al. 2020); however, AMOC variations in the MDM are solely buoyancy driven. For the MDM ensemble, five ensemble members are used for 1850–2014 and 10 ensemble members are used for 2015–2100. The MDM requires fewer ensemble members to average out internal variability because the MDM does not contain El Niño–Southern Oscillation (ENSO; McMonigal et al. 2023) or other internal variability where wind-driven ocean circulation plays a role, such as Atlantic Niño. In the NAWH region, SSTs in the MDM have 17% lower ensemble standard deviation than in the FCM. The impact of differing ensemble sizes for the FCM and the MDM on our results is assessed in section 3a and is found to be small.
The mean state of preindustrial versions of the FCM and MDM is similar (McMonigal et al. 2023; Larson et al. 2024). This is expected because in the preindustrial control run, the mean momentum and buoyancy forcing are nearly identical between the FCM and the MDM. Despite this, the MDM has slightly shallower mixed layer depths and slightly warmer SSTs than the FCM in most locations (Larson et al. 2024), likely due to the lack of high-frequency wind variability (Luongo et al. 2024). However, mean state differences in the subpolar North Atlantic are small.
The difference between the ensemble mean of the FCM and the ensemble mean of the MDM isolates the role of externally forced changes in wind-driven ocean circulation, with the assumption that changes due to internal variability driven by anomalous wind stress forcing (such as ENSO) projecting onto the ensemble mean are small. Globally averaged, the FCM and MDM ensembles warm at slightly different rates, with the FCM warming 17% more than the MDM over 1979–2014 (McMonigal et al. 2023). This is due to wind-driven ocean circulation changes altering the pattern of the warming, including amplifying warming in the eastern tropical Pacific due to southward Ekman advection driven by off-equatorial westerly wind anomalies (Fu et al. 2024). The FCM warming pattern reduces low cloud coverage and amplifies global-mean warming relative to the MDM (McMonigal et al. 2023). This warming rate difference is fairly constant in time after 1979. For simplicity, we do not remove the global averaged warming from the figures shown unless otherwise specified. Anomalies are calculated by subtracting a reference period of 1930–60 within each experiment. The NAWH region is defined as 48°–60°N, 50°–10°W, chosen to align with previous work (Gervais et al. 2019, 2020; Rahmstorf et al. 2015; Keil et al. 2020).
b. Oceanic analyses
Both the time-varying Ekman mass transport and time-varying potential temperature are important components of Eq. (1). Substituting a time mean Ekman layer potential temperature gives significantly different results (not shown). This suggests that analyses which only consider changes to Ekman velocity, and not temperature, may be inaccurate in the NAWH region under external forcing due to the large projected changes in horizontal SST gradients caused by the warming hole. This is in contrast to analyses that occur over shorter time periods (likely with smaller changes to SST gradients), where the velocity component of anomalous Ekman transport dominates (Buckley et al. 2015).
To investigate the role of AMOC in the warming hole development, we use the meridional overturning streamfunction output from CESM2, which is defined in depth space. Additionally, we use the output variable for the northward heat transport in the Atlantic (N_HEAT). The northward heat transport is the total advective northward heat transport, e.g., the sum of Eulerian-mean (resolved) advection, mesoscale advection, and submesoscale advection. The diffusive term is undefined at the southern boundary of the Atlantic and is not included in the CESM2 output.
c. Atmospheric analyses
The NAO index is defined using the station-based index, where the NAO is calculated as the difference of normalized sea level pressure between Lisbon, Portugal, and Reykjavik, Iceland (Hurrell et al. 2001). We use December–February (DJF) means when showing the NAO index.
For the atmospheric analysis, results are presented as anomalies relative to a reference period of 2015–40 because daily output from the MDM is not available for the period of 1850–2015. All seasons are considered because the NAWH region SST difference between the FCM and the MDM is similar in magnitude between December–February and June–August (not shown).
d. Statistical significance
3. Results
a. Warming hole emergence
The ensemble means of the NAWH SST in the FCM and MDM show very similar evolution prior to anthropogenic warming, in the early 1900s (Fig. 1). The spread of ensemble members, calculated as two ensemble standard deviations divided by the square root of the number of ensemble members, is larger in MDM than in FCM. This is due to the smaller ensemble size in MDM than in FCM. Both ensembles show the NAWH region warming from 1970 to 2000 and cooling after 2000 (first vertical line in Fig. 1). The SST trend patterns over 1970–2000 and 2000–40 are broadly similar in FCM and MDM (Figs. 2a,b,d,e). This shows that, in CESM2, changes to the wind-driven ocean circulation do not lead to the emergence of the externally forced component of the NAWH. The timing of externally forced cooling beginning in 2000 is consistent with the “delay” of warming hole formation by aerosol forcing in CMIP6 models (Dagan et al. 2020). In 2040, the warming hole in the FCM model begins to undergo amplified cooling, seen as an amplified negative slope of the NAWH SST (second vertical line in Fig. 1). This amplification of the warming hole persists until 2070, when the FCM SST tendency reverses from positive to negative and the NAWH SST begins to warm (third vertical line in Fig. 1). In contrast, the MDM ensemble mean does not show a period of amplified cooling. Instead, the MDM continues to cool at a relatively steady rate over 2000–2100. This leads to significantly cooler NAWH SSTs in the FCM than in the MDM over 2061–75 (thick line in Fig. 1b). The amplified cooling in FCM relative to MDM is widespread across the subpolar North Atlantic (Fig. 2i). By construction, the amplified cooling tendency in FCM relative to MDM, and the significantly cooler SSTs over 2061–75, must be related to changes in wind-driven ocean circulation.
(a) Time series of the SST anomaly in the NAWH region (48°–60°N, 50°–10°W) in the two model experiment ensemble means (thick lines). Shading shows plus and minus two ensemble standard deviations divided by the square root of the number of ensemble members. (b) Difference between SST anomalies in FCM ensemble mean minus MDM ensemble mean. Thicker line shows when the difference is statistically significant at the 95% threshold. Vertical lines are shown at years 2000, 2040, and 2070 for reference.
Citation: Journal of Climate 38, 11; 10.1175/JCLI-D-24-0227.1
Spatial map of SST trends in (a)–(c) FCM ensemble mean, (d)–(f) MDM ensemble mean, and (g)–(i) FCM − MDM over time periods (a),(d),(g) 1970–2000, (b),(e),(h) 2000–40, and (c),(f),(i) 2040–70. Boxes show the NAWH region used for the spatial average in Fig. 1.
Citation: Journal of Climate 38, 11; 10.1175/JCLI-D-24-0227.1
The timing of FCM − MDM differences in the NAWH is similar when the global-mean warming is subtracted (Fig. S1 in the online supplemental material) or regressed out (Fig. S2), indicating the changes in the NAWH region are distinct from global-mean differences. The possible impact of the different ensemble sizes of the FCM and MDM is quantified by considering the chance that the ensemble mean of 10 randomly selected FCM ensemble members would show a different NAWH region SST trend from the MDM ensemble mean over 2040–70. Using a Monte Carlo simulation with 100 000 iterations, the chance is 99.999%. This can be explained through a histogram of all ensemble member NAWH SST trends over 2040–70 (Fig. S3). Only 8 out of the 50 FCM ensemble members have a NAWH SST trend which is greater than the ensemble mean MDM NAWH SST trend, and the FCM ensemble members have a wide distribution. If five FCM ensemble members are selected in the Monte Carlo simulation, the chance that the FCM and MDM ensemble means are different drops to 99.4%. Thus, it is extremely likely that it would be possible to detect the significant difference between the ensembles using only five FCM ensemble members.
b. Role of the Atlantic meridional overturning circulation
A weakening AMOC is often implicated as the cause of the NAWH (Caesar et al. 2018; Rahmstorf et al. 2015; Drijfhout et al. 2012). To investigate the role of AMOC in the cooling of the NAWH that begins in 2000 in our model, and in the amplified cooling that occurs from 2040 to 2070, we show several metrics for AMOC strength (Fig. 3). The time evolution of the AMOC streamfunction across 48°N in both experiments is similar (Fig. 3a). Both experiments show a weakening of the AMOC due to external forcing beginning just prior to 2000. This is in agreement with previous studies showing that the projected AMOC weakening is buoyancy forced (Cheng et al. 2013; Levang and Schmitt 2020). The AMOC across 48°N continues to decline until 2100, in both experiments. Small but significant differences between the experiments emerge in 2050. This broadly agrees with findings demonstrating the AMOC strength is sensitive to wind nudging (Roach et al. 2022), although our results specifically isolate the role of wind stress trends (rather than turbulent heat fluxes) in altering AMOC strength. The advective ocean heat transport across 48°N follows a slightly different trajectory than AMOC across 48°N, with a decline in both the FCM and MDM ocean heat transports beginning around 2000 and an amplified decline in the FCM relative to the MDM around 2040 (Fig. 3b). The difference between the evolution of AMOC and heat transport at 48°N suggests one of two possibilities. First, the AMOC variable output by CESM2 is calculated in depth space. In the subpolar North Atlantic, isopycnals slope from east to west, such that most of the heat transport is carried by warm flow in the east returned at the same depth by cooler flow in the west (Buckley et al. 2023; Zhang and Thomas 2021). Thus, when AMOC is calculated in depth space from observations in the subpolar gyre, nearly all of the heat transport is carried by a horizontal, rather than a vertical, flow (Lozier et al. 2019). Therefore, the difference in AMOC in depth space (from the model output) and advective northward heat transport may be due to the depth-based definition of AMOC in CESM2. Second, even when using a density-based definition, some of the ocean heat transport variability in the North Atlantic subpolar gyre is carried by the horizontal gyre circulation. In observations, 13% of monthly variance in ocean heat transport is not captured by the density space overturning (Lozier et al. 2019). Thus, it is also possible that the difference between AMOC and heat transport seen in Fig. 3 is due to heat transport variability unrelated to the overturning circulation. In either case, a declining AMOC (and heat transport) at 48°N occurs at year 2000, the same time as the NAWH cooling begins in both models, suggesting the AMOC decline is a likely cause of the externally forced warming hole. The surface net heat flux into the ocean Qnet (defined as positive into the ocean) increases beginning around year 2000 in both models (Figs. 4a,b) and thus acts to dampen the reduction in ocean heat transport. Changes to Qnet are mostly due to turbulent heat flux changes (Figs. 4a,b).
Time series of AMOC streamfunction anomalies at (a) 48°N, (c) 60°N, and (e) the convergence between 48° and 60°N in the two model experiment ensemble means (thick lines). Shading shows plus and minus two ensemble standard deviations divided by the square root of the number of ensemble members. Shading in (e) is present, but minuscule, due to very small ensemble standard deviation. Units are given in Sverdrups, defined as 106 m3 s−1. (b) Time series of full depth ocean heat transport anomalies across (b) 48°N, (d) 60°N, and (f) the convergence between 48° and 60°N in the two model experiment ensemble means (thick lines). Units are given in petawatts, defined as 1015 Watts. Vertical lines are shown at years 2000, 2040, and 2070 for reference.
Citation: Journal of Climate 38, 11; 10.1175/JCLI-D-24-0227.1
The Qnet components averaged over the NAWH region (48°–60°N, 50°–10°W) in the (a) FCM, (b) MDM, and (c) FCM − MDM ensemble means. Vertical lines are shown at years 2000, 2040, and 2070 for reference.
Citation: Journal of Climate 38, 11; 10.1175/JCLI-D-24-0227.1
Next, we investigate the AMOC streamfunction across 60°N. Both experiments show a decline beginning in 2000, with FCM showing less of a decline than MDM over 2000–40 (Fig. 3c). In contrast, advective heat transport across 60°N (Fig. 3d) shows a similar decline over 2000–40 in both experiments. Finally, we combine the streamfunctions and heat transport to estimate volume and heat convergence between 48° and 60°N (Figs. 3e,f). Both volume and heat convergence decline over time starting in 2000. In other words, the overturning circulation and heat transport decline more at 48°N than at 60°N. The convergence trajectories show differences between FCM and MDM that begin during the 2000–40 period. Advective ocean heat transport is only significantly different between the two experiments for a few years after 2070 (Fig. 3f, significant differences occur when the shading is not overlapping). When the amplified cooling period begins in FCM, the FCM ensemble mean heat convergence anomaly is less negative than the MDM ensemble mean. Thus, ocean heat convergence differences are unable to explain the amplified cooling in the FCM model. Additionally, the vertical structure of the FCM minus MDM temperature anomalies shows amplified cooling in the upper 150 m and amplified warming below 150 m (Fig. 5c). This suggests that the amplified cooling is not due to a full depth heat transport change but is instead due to a process modifying primarily the upper water column with partial compensation at depth. A similar reversal of sign with depth is seen in the FCM minus MDM salinity anomalies over the same period (Fig. 5f).
Vertical temperature anomalies over 2040–70 as a function of depth and latitude along 40°W in (a) the FCM, (b) the MDM, and (c) FCM − MDM ensemble means and vertical salinity anomalies over 2040–70 as a function of depth and latitude along 40°W in (d) the FCM, (e) the MDM, and (f) FCM − MDM ensemble means. The black line shows the mixed layer depth in FCM, and the green line shows the mixed layer depth in MDM. Stippling shows statistically significant differences.
Citation: Journal of Climate 38, 11; 10.1175/JCLI-D-24-0227.1
In conclusion, a decline in AMOC and ocean heat convergence can explain the formation of a warming hole in each experiment. However, the FCM minus MDM difference between in ocean heat convergence cannot explain the different evolutions of the NAWH over 2040–70. Additionally, AMOC in depth space tells a slightly different story than heat transport, suggesting caution when using AMOC in depth space in the subpolar gyre. Because ocean heat convergence (and AMOC) differences cannot explain the amplified cooling in FCM compared to MDM, we next consider other possible causes.
c. Causes of amplified cooling
To investigate the cause of the amplified cooling in FCM relative to MDM, we next investigate the latitudinal range where amplified cooling begins (Fig. 6). When the warming hole begins in 2000, both FCM and MDM show cooling centered just south of the latitude of maximum westerlies, at approximately 55°N (Figs. 6b,d; dashed line shows the location of maximum westerlies). Temperatures north of 60°N show small changes, and temperatures south of 48°N continue to warm. A different pattern emerges when we consider the FCM minus MDM difference. Differences are small and insignificant at most latitudes before 2040 (Fig. 6f). However, shortly after 2040, significant amplified cooling in FCM relative to MDM begins at 60°N. This difference spreads to a wider latitude range over the following decades. Nonetheless, SST trend differences over 2040–70 are evident across the entire North Atlantic subpolar gyre, with maximum differences occurring near the Greenland coast (Fig. 2i).
Zonal mean and ensemble mean SSTs over 50°–10°W as a function of latitude in (a) FCM, (c) MDM, and (e) FCM minus MDM. (g) Mean TAUX as a function of latitude in FCM. Hovmoeller plots of anomalous SST in (b) FCM, (d) MDM, and (f) FCM minus MDM. (h) Hovmoeller plot of anomalous TAUX in FCM. Stippling in (f) and (h) shows statistical significance. Dashed lines in (b), (d), (f), and (h) show the time-varying latitude of maximum westerly wind strength in the FCM ensemble mean. Solid vertical and horizontal lines denote 48°N, 60°N, and years 2000, 2040, and 2070 for reference.
Citation: Journal of Climate 38, 11; 10.1175/JCLI-D-24-0227.1
Because the amplified cooling signal begins at the northern edge of the subpolar gyre, near the Greenland coast, we construct a Greenland index showing the areal average SST anomaly between 58°–65°N and 55°–30°W (Fig. 7, area shown as green box in Fig. 2). Statistically significant differences between FCM and MDM emerge earlier when considering the Greenland coastal region. Figure 7b shows that statistically significant differences here begin before 2050, while Fig. 1b shows that the significant differences in the larger NAWH region do not begin until 2060.
(a) Time series of SST in the Greenland index region (58°–65°N, 30°–55°W) in the two model experiment ensemble means (thick lines). Shading shows plus and minus two ensemble standard deviations divided by the square root of the number of ensemble members. (b) Difference between SST anomalies in FCM ensemble mean minus MDM ensemble mean. Thicker line shows when the difference is statistically significant. Vertical lines are shown at years 2000, 2040, and 2070 for reference.
Citation: Journal of Climate 38, 11; 10.1175/JCLI-D-24-0227.1
In 2030, the northern edge of the NAWH, near the Greenland coast, begins to cool more in FCM than in MDM. This implies that wind-driven ocean circulation changes act to amplify cooling near the coast. The vertical pattern of the FCM cooling is consistent with wind-driven upwelling (Fig. 5a). Additionally, the vertical pattern of the FCM minus MDM temperature and salinity differences are consistent with enhanced coastal upwelling in the FCM (Figs. 5c,f). Wind stress anomalies over 2030–59 are generally along the coast line, with northwesterly anomalies off the west coast of Greenland and southwesterly anomalies off the east coast of Greenland (black arrows in Fig. 8). These wind anomalies become significant before 2040, as seen in the zonal mean zonal wind stress (Fig. 6h). The alongshore wind anomalies are expected to lead to offshore Ekman transport and coastal upwelling. The wind-driven upper 100-m ocean velocity anomalies are small (Fig. 8c; green arrows) but consistent with increasing upper ocean velocity divergence near the coast (Fig. 9). Mixed layer depth changes show more shoaling in FCM than MDM (Fig. 10i), also consistent with enhanced upwelling. The Qnet trend differences (Figs. 4c and 11i) are the wrong sign to explain the amplified cooling, implying a damping effect from the atmosphere. Horizontal Ekman heat flux convergence anomalies also suggest cooling to the south and east of Greenland (colors in Fig. 8c), but the locations of cooling are much more confined to the coast than the SST trend difference pattern (Fig. 2). In some regions where FCM cools more than MDM, such as west of Greenland, the horizontal Ekman heat flux convergence anomalies in FCM minus MDM are large and positive. This is the incorrect sign to explain the SST difference (Fig. 2i). We conclude that upwelling of cold waters from anomalous wind stresses near the Greenland coast is the only term with the correct sign and a spatial pattern extending to both sides of Greenland to explain the initial amplified cooling in the Greenland index region.
Horizontal Ekman heat flux convergence anomalies relative to 1930–60 (colors), wind stress anomalies relative to 1930–60 (black arrows), and upper 100-m ocean velocity anomalies relative to 1930–60 (green arrows) over 2030–59 (a) in FCM, (b) in MDM, and (c) in FCM − MDM.
Citation: Journal of Climate 38, 11; 10.1175/JCLI-D-24-0227.1
FCM ensemble mean upper 100-m ocean velocity divergence trend over 2030–60 (m s−1 century−1).
Citation: Journal of Climate 38, 11; 10.1175/JCLI-D-24-0227.1
Spatial map of mixed layer depth trends in (a)–(c) FCM ensemble mean, (d)–(f) MDM ensemble mean, and (g)–(i) FCM − MDM over time periods (a),(d),(g) 1970–2000, (b),(e),(h) 2000–40, and (c),(f),(i) 2040–70. Black boxes show the NAWH region used for the spatial average in Fig. 1, and green box shows the Greenland area used for the spatial average in Fig. 6.
Citation: Journal of Climate 38, 11; 10.1175/JCLI-D-24-0227.1
The Qnet trends in (a)–(c) FCM ensemble mean, (d)–(f) MDM ensemble mean, and (g)–(i) FCM − MDM over time periods (a),(d),(g) 1970–2000, (b),(e),(h) 2000–40, and (c),(f),(i) 2040–70. Black boxes show the NAWH region used for the spatial average in Fig. 1, and green box shows the Greenland area used for the spatial average in Fig. 6.
Citation: Journal of Climate 38, 11; 10.1175/JCLI-D-24-0227.1
Over 2020–40, the NAO index increases in both the FCM and MDM (Fig. 12). It is possible that this shift toward a positive NAO could drive the enhanced coastal upwelling in FCM compared to MDM. Greenland tip jets are synoptic time-scale events which are more common during positive NAO events (Våge et al. 2009; Bakalian et al. 2007). Tip jets enhance deep convection in the Irminger Sea due to anomalous surface heat fluxes (Bakalian et al. 2007; Pickart et al. 2003), an effect that would be present in both FCM and MDM. However, tip jets can also cause anomalous westerly winds leading to coastal upwelling and transport of cool, freshwaters offshore (Doyle and Shapiro 1999; Pacini and Pickart 2023), an effect which would only be present in FCM. Thus, the difference between FCM and MDM would isolate a cooling effect from tip jets and other anomalous wind-driven ocean circulations. We hypothesize that enhanced frequency of Greenland tip jets due to a persistent shift toward NAO positive conditions may explain the wind stress anomalies that drive enhanced coastal upwelling.
Station-based, DJF NAO index in the FCM and MDM ensemble means (thin lines). Thick lines show 15-year rolling means.
Citation: Journal of Climate 38, 11; 10.1175/JCLI-D-24-0227.1
Enhanced coastal upwelling can lead to the initial cooler temperatures near the Greenland coast but not to the cooling in the broader Labrador Sea region. In the center of the Labrador Sea, horizontal Ekman heat flux convergence anomalies are the wrong sign to contribute to cooling (Fig. 8c). The negative FCM minus MDM SSTs in this region can instead be explained through a positive feedback mechanism that links negative NAO events to reduced deep convection (Gu et al. 2024). Gu et al. (2024) note that, beginning in 2035, some FCM ensemble members cool much more than others, leading to a larger internal variability. They propose that the difference between the cold group and warm group members can be explained by a positive feedback loop, where stochastic atmospheric forcing leads to less upward mixing of heat in the Labrador Sea, which cools SSTs. The cooler SSTs spread to a wider region through the mean large-scale circulation, leading to a deceleration of sea ice melting, which reduces mechanical stirring and weakens deep convection, forming a positive feedback. This feedback loop cannot operate in the MDM because the prescribed preindustrial climatological wind stress forcing shuts off the ability for mechanical stirring to change over time. There is a reduction in sea ice melt over 2040–70 in FCM relative to MDM (Fig. S4), consistent with cooler SSTs leading to less melting, which then reduces mechanical stirring. We hypothesize that, as each FCM ensemble member is triggered by stochastic forcing from negative NAO events, it undergoes amplified cooling. As more and more FCM ensemble members undergo amplified cooling, and no MDM ensemble members do, eventually a significant ensemble mean difference emerges between the two experiments. While there is a general trend toward a positive NAO state (Fig. 12), negative NAO events will still occur and trigger the feedback loop. The simultaneous amplified cooling in the FCM when the internal variability amplifies both supports the proposed positive feedback loop and strengthens the argument that the feedback hinges on wind-driven mechanical stirring. It also suggests that the subpolar North Atlantic climate is more sensitive to negative NAO events than to positive NAO events. While a trend toward positive NAO can produce cooling near the Greenland coast, the intermittent negative NAO events are able to kick off a positive feedback that cools a broader region of the Labrador Sea.
Last, the AMOC evolution over time in the two experiments is consistent with the positive feedback proposed by Gu et al. (2024). The feedback completes when reduced mechanical stirring reduces deep convection and further weakens AMOC. FCM shows a larger AMOC decline than MDM over 2040–70 (Fig. 3). This is particularly pronounced at 60°N (Fig. 3c), near where the positive feedback occurs. Thus, we propose that the AMOC differences described in the previous section are a consequence, rather than a cause, of the amplified SST cooling due to wind-driven ocean circulation changes.
The amplified cooling near the Greenland coast leads to different meridional SST gradients in the two models. This is notable when comparing FCM and MDM SST as a function of latitude during 2040–70 (Fig. 13a). The FCM SST gradient is negative over the latitude range of 48°–54°N (Fig. 13b) and small yet positive over the latitude range of 54°–57°N. This is important because, prior to the warming hole emergence, the SST gradient in this latitude range is always negative (not shown). The change of sign of the SST gradient has substantial impacts on ocean heat advection. The prevailing winds are westerly, such that the Ekman heat transport is dominated by the effect of zonal winds advecting the meridional SST gradient. If the wind direction stays the same, but the SST gradient reverses sign, this reverses the direction of Ekman heat transport.
(a) Mean SST as a function of latitude between 50° and 10°W over 2040–70. (b) Mean meridional SST gradient as a function of latitude between 50° and 10°W over 2040–70. Solid horizontal lines denote 48° and 60°N. Dashed horizontal line denotes the mean location of the maximum westerly winds.
Citation: Journal of Climate 38, 11; 10.1175/JCLI-D-24-0227.1
With a negative meridional SST gradient, Ekman heat convergence will tend to cool the area. With a positive meridional SST gradient, Ekman heat convergence will induce a warming. Thus, the more positive SST gradient near 54°N in MDM compared to FCM will contribute to heating (or reduced cooling). This alteration of the meridional SST gradient allows the amplified cooling to spread to a broader latitude range. Both the FCM (Fig. 14a) and the MDM (Fig. 14b) show reduced horizontal Ekman heat convergence in the middle of the subpolar gyre. The MDM changes in Ekman heat convergence are solely due to changes in the temperature gradient because the wind stress anomalies are zero by definition. The effect of changing temperature gradients is sizable in the warming hole region, as seen by the magnitude of trends in Fig. 14b. This is in contrast to previous studies which found that anomalous velocities advecting mean temperature gradients dominate Ekman heat transport variability in the subpolar gyre (Buckley et al. 2015). However, it makes physical sense that a region of cooling, when surrounded by anthropogenic warming, would substantially alter meridional temperature gradients and thus Ekman heat convergence. To isolate only the role of wind stress anomalies in Ekman heat convergence anomalies, we consider the FCM minus MDM difference (Fig. 14c). Weakened westerlies in the FCM lead to predominately northwestward surface ocean velocity trends in FCM minus MDM (green arrows in Fig. 14c).
Horizontal Ekman heat convergence trends (colors), wind stress trends (black arrows), and upper 100-m ocean velocity trends (green arrows) over 2040–70 (a) in FCM, (b) in MDM, and (c) in FCM − MDM.
Citation: Journal of Climate 38, 11; 10.1175/JCLI-D-24-0227.1
We note that other wind-driven ocean dynamics may play some role in the amplified cooling in FCM relative to MDM. The subpolar gyre circulation weakens more in the FCM than the MDM (Fig. S5), which could lead to differences in temperature advection. The importance of geostrophic transports at time scales of longer than 1 year in the subpolar North Atlantic upper ocean heat budget has been noted previously (Buckley et al. 2014; Piecuch et al. 2017). Enhanced weakening of the subpolar gyre could also be a consequence of the larger weakening of AMOC (Böning et al. 2006; Larson et al. 2020) triggered through the positive feedback loop. Next, we address whether the wind-driven ocean circulation–related amplification of the warming hole has significant feedbacks onto the atmosphere during the time period when the FCM minus MDM SST anomaly is significantly different.
d. Atmospheric feedback of wind-driven SST cooling
To assess whether the different NAWHs in each model lead to different atmospheric responses, we compare sea level pressure and precipitation anomalies in each model, focusing on the period of 2061–75, when FCM SST is significantly different than MDM SST (Fig. 15). Cooler SST anomalies in FCM compared to MDM (Fig. 15a) lead to locally higher sea level pressure anomalies in FCM compared to MDM (Fig. 15b). This is the expected direct linear response to a cool, shallow midlatitude thermal anomaly (Hoskins and Karoly 1981; Hendon and Hartmann 1982; Kushnir et al. 2002) and consistent with the linear response of the atmosphere to the NAWH (Gervais et al. 2019). The significant sea level pressure signal is confined to a small area near the tip of Greenland.
(a) SST anomaly difference (FCM − MDM) over 2061–75. (b) SLP anomaly difference (FCM − MDM) over 2061–75. (c) Precipitation rate anomaly difference (FCM − MDM) over 2061–75. Hatching shows regions where the difference is statistically significant.
Citation: Journal of Climate 38, 11; 10.1175/JCLI-D-24-0227.1
Similarly, reduced precipitation is collocated with higher pressure in FCM (Fig. 15c), such that FCM has reduced precipitation anomalies compared to MDM. Model experiments that isolate the impact of the warming hole onto the atmosphere show that cooler SSTs lead to reduced precipitation over the warming hole region (Gervais et al. 2020). Thus, the reduced precipitation in FCM relative to MDM is consistent with the expected response from an amplified warming hole. This precipitation response is a local effect mostly confined to the southern tip of Greenland, and no significant effects are seen over western Europe.
Why is the atmospheric response to the cooler SST anomalies in FCM locally confined, when the atmospheric response to a warming hole has a broader regional pattern? In addition to the linear response, a nonlinear response to the increased meridional SST gradient caused by the NAWH can enhance transient eddy activity near the surface, and this can feed the eddy-driven jet in the upper troposphere (Gervais et al. 2019; Kushnir et al. 2002). To address whether this mechanism is operating, we compute the atmospheric eddy heat transport and E vector anomalies (Fig. 16).
The 850-hPa eddy heat transport anomalies (2061–75 relative to 2015–35) and 200-hPa E vector anomalies in (a) FCM, (b) MDM, and (c) FCM − MDM.
Citation: Journal of Climate 38, 11; 10.1175/JCLI-D-24-0227.1
Both FCM and MDM show an increase in eddy heat transport and divergent E vectors in the NAWH region (Figs. 16a,b), in agreement with the expected response to increased meridional SST gradients. Positive values of lower-tropospheric eddy heat transport indicate a westward tilt with height and an upward propagation of eddy activity. Divergent upper-tropospheric E vectors indicate a transient eddy-forced eastward acceleration of the mean flow. However, the difference between FCM and MDM eddy heat transport and E vector anomalies is small (Fig. 16c). FCM shows a larger increase in eddy heat transport everywhere. This is not clearly related to the NAWH and may instead be due to the amplified global warming rate in FCM compared to MDM (McMonigal et al. 2023). Changes to the North Atlantic jet can be driven by tropical heating, Arctic amplification, and the NAWH itself (Gervais et al. 2019). We cannot clearly distinguish the role of each factor on the jet differences between the FCM and MDM models. The FCM − MDM E vectors are not obviously divergent. This suggests that, although there are some differences in the atmospheric eddy heat transport between the experiments, the differences in eddy heat transport are not feeding into the mean flow. This is likely due to the location of the difference in FCM minus MDM SST gradients. The largest difference in FCM versus MDM SST gradients occurs near 52°–58°N (Fig. 13b). This is north of the eddy-driven jet, such that the enhanced SST gradients cannot alter the jet strength. This is likely why there is not a significant difference in western European precipitation anomalies between the experiments (Fig. 15c).
4. Discussion and implications
The externally forced component of the emergence of the NAWH in CESM2 is not caused by changes to wind-driven ocean circulation. Neither the timing nor magnitude of the initial warming hole is altered when wind-driven ocean circulation is constrained to its preindustrial state (Fig. 1). Rather, in CESM2, a decline in buoyancy-forced ocean heat convergence causes the externally forced NAWH (Fig. 3f), with compensation by turbulent heat fluxes (Fig. 4). It is possible that some component of the observed historical warming hole is related to wind-driven ocean circulation changes, as argued by previous studies (He et al. 2022; Hu and Fedorov 2020; Karnauskas et al. 2021). In particular, a component of the observed NAWH may be related to internal variability, which is not explored in our analysis.
In CESM2, wind-driven ocean circulation changes impact the future evolution of the externally forced warming hole. This consists of an amplified cooling over 2040–70, which manifests over a broad region (Fig. 2i). Coastal upwelling leads to amplified cooling of FCM relative to MDM near Greenland, perhaps linked to enhanced Greenland tip jets due to the trend toward an NAO positive state. However, even with the overall trend toward a positive NAO state, sporadic negative NAO events can kick off amplified cooling of individual ensemble members in the central Labrador Sea. Eventually, each FCM ensemble member has been triggered by a negative NAO event, which mixes less heat upward, cooling SSTs. This spreads via the mean ocean circulation, decelerating sea ice melt and reducing mechanical stirring and further weakening AMOC (Gu et al. 2024). Due to the reliance on mechanical stirring, this feedback is absent in the MDM. Thus, the FCM simulates cooler SSTs (Figs. 1 and 2), enhanced sea ice extent (Fig. S4), and reduced AMOC strength (Fig. 3) compared to the MDM.
It is likely that the exact timing and magnitude of the impact of wind-driven ocean circulation changes is model dependent. For example, the externally forced warming hole in CMIP6 emerges later than in CMIP5, likely due to aerosol feedback effects (Dagan et al. 2020). This may impact the timing of amplified cooling, as the positive feedback loop between SSTs and mechanical stirring depends on the background salinity gradient (Gu et al. 2024), which is dependent on freshwater changes due to melting ice. It is also possible that higher resolution models, which better resolve mixing and deep convection, would show different impacts of wind-driven ocean circulation changes. Nonetheless, these results show that North Atlantic SST trends in CMIP6-era models are sensitive to wind-driven ocean circulation changes, with implications for climate projections.
Despite significantly different SST anomalies due to wind-driven ocean circulation changes, the feedback onto the atmosphere is locally confined (Fig. 15). This is because the SST anomaly differences are located far north of the North Atlantic jet. The atmosphere only responds through the linear response to a midlatitude thermal forcing and not through the nonlinear transient eddy mechanism (Fig. 16). In CESM2, the amplified cooling due to wind-driven ocean circulation changes does not significantly alter sea level pressure or precipitation over western Europe (Fig. 15). However, if wind-driven ocean circulation changes occur farther south in other models, it is possible that feedbacks onto the atmosphere would lead to significant changes in precipitation over western Europe.
We also find a substantial impact of the changing horizontal temperature gradient on Ekman heat convergence changes. The MDM, which has no changes to the wind stress forcing on the ocean, still shows relatively large Ekman heat convergence changes (Figs. 8b and 14b). This illustrates that, in cases with a warming hole, it is inappropriate to assume that Ekman heat convergence changes are dominated by wind stress changes over long time periods. Several analyses of the subpolar North Atlantic make the assumption that Ekman heat convergence variability or trends are solely due to wind stress changes (Hu and Fedorov 2020; He et al. 2022; Buckley et al. 2014). On relatively short time scales, it is likely reasonable to assume a constant temperature gradient. However, on the century-long time scales analyzed here, this is a poor assumption. Studies investigating the future of the warming hole need to consider possible changes to the meridional ocean temperature gradient.
These results illustrate that wind-driven ocean circulation changes can alter the evolution of the NAWH under external forcing, even if the NAWH is not initially caused by wind-driven ocean circulation changes. In CESM2, buoyancy forcing causes the externally forced NAWH, as evidenced by the warming hole existence in the MDM. However, the later evolution of the warming hole is dictated by wind stress changes induced by a shift toward a positive NAO state and a positive feedback loop kicked off by sporadic negative NAO events. Thus, the effect of NAO on the NAWH region SSTs appears to be asymmetrical. It is crucial that models of the future NAWH adequately simulate future atmospheric circulation changes and the impact of such changes onto the ocean circulation.
Acknowledgments.
This work is supported by NSF Grant AGS-1951713 (KM and SML). The authors acknowledge the high-performance computing support from Cheyenne provided by NCAR’s Computing and Information Services Lab, sponsored by NSF. We thank Martha Buckley for the insightful discussion and feedback.
Data availability statement.
The CESM2 FCM large ensemble is publicly available at https://www.cesm.ucar.edu/community-projects/lens2/data-sets. The MDM ensemble fields used to produce the presented results are available at https://doi.org/10.5281/zenodo.10999760. Code to reproduce the MDM simulations is available at https://zenodo.org/records/6678286.
REFERENCES
Bakalian, F., S. Hameed, and R. Pickart, 2007: Influence of the Icelandic low latitude on the frequency of Greenland tip jet events: Implications for Irminger Sea convection. J. Geophys. Res., 112, C04020, https://doi.org/10.1029/2006JC003807.
Bellomo, K., M. Angeloni, S. Corti, and J. von Hardenberg, 2021: Future climate change shaped by inter-model differences in Atlantic meridional overturning circulation response. Nat. Commun., 12, 3659, https://doi.org/10.1038/s41467-021-24015-w.
Böning, C. W., M. Scheinert, J. Dengg, A. Biastoch, and A. Funk, 2006: Decadal variability of subpolar gyre transport and its reverberation in the North Atlantic overturning. Geophys. Res. Lett., 33, L21S01, https://doi.org/10.1029/2006GL026906.
Buckley, M. W., R. M. Ponte, G. Forget, and P. Heimbach, 2014: Low-frequency SST and upper-ocean heat content variability in the North Atlantic. J. Climate, 27, 4996–5018, https://doi.org/10.1175/JCLI-D-13-00316.1.
Buckley, M. W., R. M. Ponte, G. Forget, and P. Heimbach, 2015: Determining the origins of advective heat transport convergence variability in the North Atlantic. J. Climate, 28, 3943–3956, https://doi.org/10.1175/JCLI-D-14-00579.1.
Buckley, M. W., M. S. Lozier, D. Desbruyères, and D. G. Evans, 2023: Buoyancy forcing and the subpolar Atlantic meridional overturning circulation. Philos. Trans. Roy. Soc., A381, 20220181, https://doi.org/10.1098/rsta.2022.0181.
Caesar, L., S. Rahmstorf, A. Robinson, G. Feulner, and V. Saba, 2018: Observed fingerprint of a weakening Atlantic Ocean overturning circulation. Nature, 556, 191–196, https://doi.org/10.1038/s41586-018-0006-5.
Chemke, R., L. Zanna, and L. M. Polvani, 2020: Identifying a human signal in the North Atlantic warming hole. Nat. Commun., 11, 1540, https://doi.org/10.1038/s41467-020-15285-x.
Cheng, W., J. C. H. Chiang, and D. Zhang, 2013: Atlantic Meridional Overturning Circulation (AMOC) in CMIP5 models: RCP and historical simulations. J. Climate, 26, 7187–7197, https://doi.org/10.1175/JCLI-D-12-00496.1.
Dagan, G., P. Stier, and D. Watson-Parris, 2020: Aerosol forcing masks and delays the formation of the North Atlantic warming hole by three decades. Geophys. Res. Lett., 47, e2020GL090778, https://doi.org/10.1029/2020GL090778.
Danabasoglu, G., and Coauthors, 2020: The Community Earth System Model Version 2 (CESM2). J. Adv. Model. Earth Syst., 12, e2019MS001916, https://doi.org/10.1029/2019MS001916.
Doyle, J. D., and M. A. Shapiro, 1999: Flow response to large-scale topography: The Greenland tip jet. Tellus, 51A, 728–748, https://doi.org/10.1034/j.1600-0870.1996.00014.x.
Drijfhout, S., G. J. van Oldenborgh, and A. Cimatoribus, 2012: Is a decline of AMOC causing the warming hole above the North Atlantic in observed and modeled warming patterns? J. Climate, 25, 8373–8379, https://doi.org/10.1175/JCLI-D-12-00490.1.
Fan, Y., W. Liu, P. Zhang, R. Chen, and L. Li, 2023: North Atlantic Oscillation contributes to the subpolar North Atlantic cooling in the past century. Climate Dyn., 61, 5199–5215, https://doi.org/10.1007/s00382-023-06847-y.
Fasullo, J. T., J.-F. Lamarque, C. Hannay, N. Rosenbloom, S. Tilmes, P. DeRepentigny, A. Jahn, and C. Deser, 2022: Spurious late historical-era warming in CESM2 driven by prescribed biomass burning emissions. Geophys. Res. Lett., 49, e2021GL097420, https://doi.org/10.1029/2021GL097420.
Ferster, B. S., A. Simon, A. Fedorov, J. Mignot, and E. Guilyardi, 2022: Slowdown and recovery of the Atlantic meridional overturning circulation and a persistent North Atlantic warming hole induced by Arctic sea ice decline. Geophys. Res. Lett., 49, e2022GL097967, https://doi.org/10.1029/2022GL097967.
Fiedler, S., and D. Putrasahan, 2021: How does the North Atlantic SST pattern respond to anthropogenic aerosols in the 1970s and 2000s? Geophys. Res. Lett., 48, e2020GL092142, https://doi.org/10.1029/2020GL092142.
Fu, S., S. Hu, X.-T. Zheng, K. McMonigal, S. Larson, and Y. Tian, 2024: Historical changes in wind-driven ocean circulation drive pattern of Pacific warming. Nat. Commun., 15, 1562, https://doi.org/10.1038/s41467-024-45677-2.
Fu, Y., F. Li, J. Karstensen, and C. Wang, 2020: A stable Atlantic Meridional overturning circulation in a changing North Atlantic Ocean since the 1990s. Sci. Adv., 6, eabc7836, https://doi.org/10.1126/sciadv.abc7836.
Gervais, M., J. Shaman, and Y. Kushnir, 2018: Mechanisms governing the development of the North Atlantic warming hole in the CESM-LE future climate simulations. J. Climate, 31, 5927–5946, https://doi.org/10.1175/JCLI-D-17-0635.1.
Gervais, M., J. Shaman, and Y. Kushnir, 2019: Impacts of the North Atlantic warming hole in future climate projections: Mean atmospheric circulation and the North Atlantic Jet. J. Climate, 32, 2673–2689, https://doi.org/10.1175/JCLI-D-18-0647.1.
Gervais, M., J. Shaman, and Y. Kushnir, 2020: Impact of the North Atlantic warming hole on sensible weather. J. Climate, 33, 4255–4271, https://doi.org/10.1175/JCLI-D-19-0636.1.
Gu, Q., M. Gervais, G. Danabasoglu, W. M. Kim, F. Castruccio, E. Maroon, and S.-P. Xie, 2024: Wide range of possible trajectories of North Atlantic climate in a warming world. Nat. Commun., 15, 4221, https://doi.org/10.1038/s41467-024-48401-2.
He, C., A. C. Clement, M. A. Cane, L. N. Murphy, J. M. Klavans, and T. M. Fenske, 2022: A North Atlantic Warming Hole Without Ocean Circulation. Geophys. Res. Lett., 49, e2022GL100420, https://doi.org/10.1029/2022GL100420.
Hendon, H. H., and D. L. Hartmann, 1982: Stationary waves on a sphere: Sensitivity to thermal feedback. J. Atmos. Sci., 39, 1906–1920, https://doi.org/10.1175/1520-0469(1982)039<1906:SWOASS>2.0.CO;2.
Hoskins, B. J., and D. J. Karoly, 1981: The steady linear response of a spherical atmosphere to thermal and orographic forcing. J. Atmos. Sci., 38, 1179–1196, https://doi.org/10.1175/1520-0469(1981)038%3C1179:TSLROA%3E2.0.CO;2.
Hoskins, B. J., I. N. James, and G. H. White, 1983: The shape, propagation and mean-flow interaction of large-scale weather systems. J. Atmos. Sci., 40, 1595–1612, https://doi.org/10.1175/1520-0469(1983)040<1595:TSPAMF>2.0.CO;2.
Hu, A., G. A. Meehl, and W. Han, 2007: Causes of a fresher, colder northern North Atlantic in late 20th century in a coupled model. Prog. Oceanogr., 73, 384–405, https://doi.org/10.1016/j.pocean.2006.07.008.
Hu, S., and A. V. Fedorov, 2020: Indian Ocean warming as a driver of the North Atlantic warming hole. Nat. Commun., 11, 4785, https://doi.org/10.1038/s41467-020-18522-5.
Huang, D., A. Dai, and J. Zhu, 2020: Are the transient and equilibrium climate change patterns similar in response to increased CO2? J. Climate, 33, 8003–8023, https://doi.org/10.1175/JCLI-D-19-0749.1.
Hurrell, J. W., Y. Kushnir, and M. Visbeck, 2001: The North Atlantic Oscillation. Science, 291, 603–605, https://doi.org/10.1126/science.1058761.
IPCC, 2023: Ocean, cryosphere and sea level change. Climate Change 2021: The Physical Science Basis, Cambridge University Press, 1211–1362, https://doi.org/10.1017/9781009157896.011.
Jahn, A., and M. M. Holland, 2013: Implications of arctic sea ice changes for North Atlantic deep convection and the meridional overturning circulation in CCSM4-CMIP5 simulations. Geophys. Res. Lett., 40, 1206–1211, https://doi.org/10.1002/grl.50183.
Karnauskas, K. B., L. Zhang, and D. J. Amaya, 2021: The atmospheric response to North Atlantic SST trends, 1870–2019. Geophys. Res. Lett., 48, e2020GL090677, https://doi.org/10.1029/2020GL090677.
Keil, P., T. Mauritsen, J. Jungclaus, C. Hedemann, D. Olonscheck, and R. Ghosh, 2020: Multiple drivers of the North Atlantic warming hole. Nat. Climate Change, 10, 667–671, https://doi.org/10.1038/s41558-020-0819-8.
Kilbourne, K. H., and Coauthors, 2022: Atlantic circulation change still uncertain. Nat. Geosci., 15, 165–167, https://doi.org/10.1038/s41561-022-00896-4.
Kusakabe, Y., and T. Takemura, 2023: Formation of the North Atlantic Warming Hole by reducing anthropogenic sulphate aerosols. Sci. Rep., 13, 8, https://doi.org/10.1038/s41598-022-27315-3.
Kushnir, Y., W. A. Robinson, I. Bladé, N. M. J. Hall, S. Peng, and R. Sutton, 2002: Atmospheric GCM response to extratropical SST anomalies: Synthesis and evaluation. J. Climate, 15, 2233–2256, https://doi.org/10.1175/1520-0442(2002)015<2233:AGRTES>2.0.CO;2.
Larson, S. M., M. W. Buckley, and A. C. Clement, 2020: Extracting the buoyancy-driven Atlantic meridional overturning circulation. J. Climate, 33, 4697–4714, https://doi.org/10.1175/JCLI-D-19-0590.1.
Larson, S. M., K. McMonigal, Y. Okumura, D. Amaya, A. Capotondi, K. Bellomo, I. R. Simpson, and A. C. Clement, 2024: Ocean complexity shapes sea surface temperature variability in a CESM2 coupled model hierarchy. J. Climate, 37, 4931–4948, https://doi.org/10.1175/JCLI-D-23-0621.1.
Levang, S. J., and R. W. Schmitt, 2020: What causes the AMOC to weaken in CMIP5? J. Climate, 33, 1535–1545, https://doi.org/10.1175/JCLI-D-19-0547.1.
Li, L., M. S. Lozier, and F. Li, 2022: Century-long cooling trend in subpolar North Atlantic forced by atmosphere: An alternative explanation. Climate Dyn., 58, 2249–2267, https://doi.org/10.1007/s00382-021-06003-4.
Liu, W., A. V. Fedorov, S.-P. Xie, and S. Hu, 2020: Climate impacts of a weakened Atlantic Meridional Overturning Circulation in a warming climate. Sci. Adv., 6, eaaz4876, https://doi.org/10.1126/sciadv.aaz4876.
Lobelle, D., C. Beaulieu, V. Livina, F. Sévellec, and E. Frajka-Williams, 2020: Detectability of an AMOC decline in current and projected climate changes. Geophys. Res. Lett., 47, e2020GL089974, https://doi.org/10.1029/2020GL089974.
Lozier, M. S., and Coauthors, 2019: A sea change in our view of overturning in the subpolar North Atlantic. Science, 363, 516–521, https://doi.org/10.1126/science.aau6592.
Luongo, M. T., N. G. Brizuela, I. Eisenman, and S.-P. Xie, 2024: Retaining short-term variability reduces mean state biases in wind stress overriding simulations. J. Adv. Model. Earth Syst., 16, e2023MS003665, https://doi.org/10.1029/2023MS003665.
Marshall, J., J. R. Scott, K. C. Armour, J.-M. Campin, M. Kelley, and A. Romanou, 2015: The ocean’s role in the transient response of climate to abrupt greenhouse gas forcing. Climate Dyn., 44, 2287–2299, https://doi.org/10.1007/s00382-014-2308-0.
McMonigal, K., S. Larson, S. Hu, and R. Kramer, 2023: Historical changes in wind-driven ocean circulation can accelerate global warming. Geophys. Res. Lett., 50, e2023GL102846, https://doi.org/10.1029/2023GL102846.
Menary, M. B., and R. A. Wood, 2018: An anatomy of the projected North Atlantic warming hole in CMIP5 models. Climate Dyn., 50, 3063–3080, https://doi.org/10.1007/s00382-017-3793-8.
Menary, M. B., and Coauthors, 2020: Aerosol-forced AMOC changes in CMIP6 historical simulations. Geophys. Res. Lett., 47, e2020GL088166, https://doi.org/10.1029/2020GL088166.
Nummelin, A., M. Ilicak, C. Li, and L. H. Smedsrud, 2016: Consequences of future increased arctic runoff on arctic ocean stratification, circulation, and sea ice cover. J. Geophys. Res. Oceans, 121, 617–637, https://doi.org/10.1002/2015JC011156.
O’Neill, B. C., and Coauthors, 2016: The Scenario Model Intercomparison Project (ScenarioMIP) for CMIP6. Geosci. Model Dev., 9, 3461–3482, https://doi.org/10.5194/gmd-9-3461-2016.
Pacini, A., and R. S. Pickart, 2023: Wind-forced upwelling along the West Greenland Shelfbreak: Implications for Labrador sea water formation. J. Geophys. Res. Oceans, 128, e2022JC018952, https://doi.org/10.1029/2022JC018952.
Pickart, R. S., M. A. Spall, M. H. Ribergaard, G. W. K. Moore, and R. F. Milliff, 2003: Deep convection in the Irminger Sea forced by the Greenland tip jet. Nature, 424, 152–156, https://doi.org/10.1038/nature01729.
Piecuch, C. G., R. M. Ponte, C. M. Little, M. W. Buckley, and I. Fukumori, 2017: Mechanisms underlying recent decadal changes in subpolar North Atlantic Ocean heat content. J. Geophys. Res. Oceans, 122, 7181–7197, https://doi.org/10.1002/2017JC012845.
Rahmstorf, S., J. E. Box, G. Feulner, M. E. Mann, A. Robinson, S. Rutherford, and E. J. Schaffernicht, 2015: Exceptional twentieth-century slowdown in Atlantic Ocean overturning circulation. Nat. Climate Change, 5, 475–480, https://doi.org/10.1038/nclimate2554.
Riahi, K., and Coauthors, 2017: The Shared Socioeconomic Pathways and their energy, land use, and greenhouse gas emissions implications: An overview. Global Environ. Change, 42, 153–168, https://doi.org/10.1016/j.gloenvcha.2016.05.009.
Roach, L. A., E. Blanchard-Wrigglesworth, S. Ragen, W. Cheng, K. C. Armour, and C. M. Bitz, 2022: The impact of winds on AMOC in a fully-coupled climate model. Geophys. Res. Lett., 49, e2022GL101203, https://doi.org/10.1029/2022GL101203.
Roberts, M. J., and Coauthors, 2020: Sensitivity of The Atlantic Meridional Overturning Circulation to model resolution in CMIP6 HighResMIP simulations and implications for future changes. J. Adv. Model. Earth Syst., 12, e2019MS002014, https://doi.org/10.1029/2019MS002014.
Rodgers, K. B., and Coauthors, 2021: Ubiquity of human-induced changes in climate variability. Earth Syst. Dyn., 12, 1393–1411, https://doi.org/10.5194/esd-12-1393-2021.
Shaw, T. A., and Coauthors, 2016: Storm track processes and the opposing influences of climate change. Nat. Geosci., 9, 656–664, https://doi.org/10.1038/ngeo2783.
Våge, K., T. Spengler, H. C. Davies, and R. S. Pickart, 2009: Multi-event analysis of the westerly Greenland tip jet based upon 45 winters in ERA-40. Quart. J. Roy. Meteor. Soc., 135, 1999–2011, https://doi.org/10.1002/qj.488.
Weijer, W., W. Cheng, O. A. Garuba, A. Hu, and B. T. Nadiga, 2020: CMIP6 models predict significant 21st century decline of the Atlantic Meridional Overturning Circulation. Geophys. Res. Lett., 47, e2019GL086075, https://doi.org/10.1029/2019GL086075.
Winton, M., S. M. Griffies, B. L. Samuels, J. L. Sarmiento, and T. L. Frölicher, 2013: Connecting changing ocean circulation with changing climate. J. Climate, 26, 2268–2278, https://doi.org/10.1175/JCLI-D-12-00296.1.
Zhang, R., and M. Thomas, 2021: Horizontal circulation across density surfaces contributes substantially to the long-term mean northern Atlantic Meridional Overturning Circulation. Commun. Earth Environ., 2, 112, https://doi.org/10.1038/s43247-021-00182-y.
