1. Introduction
Climate variability can be generated internally by interactions within or between the individual climate subcomponents (e.g., atmosphere, ocean, and sea ice) and externally by, for example, volcanic eruptions, variations in the solar insolation at the top of the atmosphere, or changed atmospheric greenhouse gas concentrations in response to anthropogenic emissions. Examples of internal variations are the North Atlantic (NA) Oscillation (NAO), the El Niño–Southern Oscillation (ENSO), the Pacific decadal variability (PDV), and the Atlantic multidecadal variability (AMV). Internally generated variations may project on global or hemispheric surface air temperature (SAT), thereby masking and complicating detection of anthropogenic climate change.
Northern Hemisphere (NH) averaged SAT has risen by about 1 K during the twentieth century (Fig. 1a) and has contributed to about two-thirds of global warming over this period. The long-term warming trend during the twentieth century has been attributed with high confidence mostly to anthropogenic forcing, specifically the increase in atmospheric greenhouse gas concentrations (Hegerl et al. 2007). Strong multidecadal fluctuations, however, are superimposed on the long-term NH warming trend, which can be readily identified in the observed record of annual mean SAT (Fig. 1a) without applying further time filtering. The “midcentury warming” (MCW) during 1910–40, for instance, was replaced by a cooling trend in the period 1950–70, which is also clearly seen in the Arctic SAT record (Fig. 1b). Thereafter, NH SAT increased again at a rate considerably stronger than the mean trend during the twentieth century (Trenberth et al. 2007).
Contributions of anomalous solar forcing and changes in volcanic and anthropogenic aerosols were suggested to explain the multidecadal variations of the observed temperatures including MCW (Broccoli et al. 2003; Stott et al. 2000). However, when considering the ensemble mean of phase 3 of the Coupled Model Intercomparison Project (CMIP3) models forced by all known external forcing, anthropogenic and natural (Hegerl et al. 2007), the MCW in the NH cannot be fully explained (Fig. 1a). An even smaller fraction of MCW can be attributed to external forcing in the Arctic (Fig. 1b). Comprehensive analysis by Wang et al. (2007) demonstrated poor simulation of the Arctic warming by climate models when driven with estimates of both observed natural and anthropogenic forcing.
At the same time, the models are capable of producing strong internal multidecadal fluctuations of the global climate similar to the observed (Delworth and Knutson 2000; Delworth and Mann 2000; Knight et al. 2005; Latif et al. 2004), and they may even fully capture the MCW “by chance” in one of several climate change simulations with identical external forcing and different initial conditions (Delworth and Knutson 2000). These studies reveal the North Atlantic region as a source of strong multidecadal natural variability. Recent analyses of the multidecadal deviations between simulated (as given by the multimodel mean of the CMIP3 model ensemble) and observed global temperature changes in the last century (Kravtsov and Spannagle 2008; Ting et al. 2009) show that the strongest discrepancies are located in the North Atlantic and presumably related to internal oceanic variability in this region. The mechanism of the link between this regional climate variability and global climate changes has yet to be understood.
Another important feature of both MCW and the warming in recent decades is the so-called Arctic amplification, a relatively high rate of warming in the northern high latitudes. The mechanism behind this phenomenon is still under discussion (Alexeev et al. 2005; Serreze and Francis 2006). Climate models reproduce it in global warming simulations, some of which assign an important role to the oceanic poleward heat transport in accelerating high-latitude warming (Holland and Bitz 2003). For the last decades of the twentieth century, however, the CMIP3 multimodel ensemble mean shows less warming than observed in the northern North Atlantic and adjacent land areas (Figs. 2a,b). This may indicate that internal variability was important in addition to external forcing not only during MCW but also during the recent decades.
This study supports the hypothesis that a relatively strong contribution to Northern Hemisphere and hence global temperature change could possibly stem from internal multidecadal variability originating in the NA–Arctic sector. Previous climate model simulations (Zhang et al. 2007) report an important role of North Atlantic multidecadal sea surface temperature (SST) variability in shaping NH SAT during the twentieth century but did not consider the role of the Arctic. Here, we focus on two main questions: First, can internal multidecadal variability in the North Atlantic–Arctic sector considerably contribute to the Northern Hemisphere SAT change? And second, what is the role of the Arctic in this? Section 2 describes the surface heat fluxes associated with AMV changes and the implied heat transports in climate models and reanalysis. In section 3, the description of the experimental setup used to assess the impact of these fluxes on the Northern Hemisphere surface climate follows. The results of the numerical experiments are given in section 4. A discussion of our main findings and their implications concludes this paper.
2. Atlantic multidecadal variability and associated heat transport changes
The possibility is investigated that internal multidecadal variability originating in the North Atlantic (Manabe and Stouffer 1999; Schmittner et al. 2007) that is accompanied with a strong oceanic heat loss to the atmosphere in the North Atlantic and especially in the Arctic sector (Bengtsson et al. 2004) has considerably contributed to the NH SAT change during the recent decades. The observed SAT trend pattern during 1978–2007 (Fig. 2a) supports this, projecting strongly onto the positive phase of AMV (Kerr 2005; Knight et al. 2005, 2006; Sutton and Hodson 2005), which is characterized by anomalous warming in the North and anomalous cooling in the South Atlantic. AMV is hypothesized to be linked to the meridional overturning circulation (MOC) variability and consequently to changes in the northward oceanic heat transport in climate models (Delworth et al. 1993; Latif et al. 2004), and AMV-related North Atlantic SST changes do project onto NH and global SAT in these models. The latter is demonstrated by analysis of the correlation between AMV and global (hemispheric) SAT anomalies in control simulations with an extensive set of climate models (Table 1).
The multimodel ensemble mean SAT trend pattern during 1978–2007 obtained from twentieth-century climate model simulations with observed external forcing (Fig. 2b), natural and anthropogenic, shows considerable differences to the observed trend pattern, with less warming in the North Atlantic and more warming in the South Atlantic. Although other natural internal decadal climate modes may have contributed to the mismatch to the observed trend, the Atlantic signature of relatively warm North and cold South Atlantic suggests that the SAT trend observed during the recent decades may contain a considerable contribution from AMV. Multidecadal NH SAT variations such as those observed (Fig. 1) are simulated by climate models in control integrations without time-varying external forcing, as described earlier. An example (Fig. 2c) is shown from one particular model [Max Planck Institute for Meteorology (MPI), exp. D16], in which the multidecadal North Atlantic SAT changes are forced by MOC changes, as demonstrated by Latif et al. (2004).
The pattern of SAT difference between two 30-yr segments corresponding to a negative and a positive AMV extreme simulated by this model (Fig. 2c) resembles somewhat the SAT trend pattern observed during the recent decades (Fig. 2a). The particular MOC variation was selected because it was connected with coherent warming in the North Atlantic and the Arctic, similar to what was observed during the MCW and the warming in recent decades. In particular, the simulated warming during the chosen period displayed poleward amplification. The realization that we used below in our experiments, with a coupled atmospheric general circulation model (AGCM)–mixed layer ocean model (ML), is in terms of anomalous North Atlantic SST of the same order of magnitude as the observed warming trends during the MCW and the recent decades. The SST averaged over the region [40°–60°N, 50°–10°W, used as an index for AMV in Latif et al. (2004)] has varied by 0.7 K (minimum to maximum) for the MCW and by about 0.8 K for the increase from the mid-1970s to the present. Yet both warming trends most likely contain some contribution from external forcing, especially during the recent decades. The internal change that we have chosen to force the AGCM–ML exhibits an SST anomaly in the same region of 1.0 K, exceeding the observed variations by some 20%. The basinwide SST change in the North Atlantic (0°–60°N), however, amounts to 0.65 and 0.47 K for the observed trend during 1978–2007 and the coupled model trend, respectively.
We turn now to the heat fluxes associated to this particular realization of internal variability. The simulated anomalous heat loss to the atmosphere due to the turbulent sensible and latent heat fluxes is mostly concentrated in the northern North Atlantic and along the sea ice border in the Atlantic opening of the Arctic Ocean (Martin and Ruprecht 2007). The pattern of anomalous turbulent heat flux (Fig. 3) associated with the AMV variation in the climate model consists of two major maxima: in the North Atlantic midlatitudes where corresponding SST anomalies occur and between 70° and 80°N, indicating a role of the sea ice variability. This structure is further illustrated by zonal mean heat fluxes (Fig. 4). Among a number of control climate simulations (see Table 1), we selected those that exhibited North Atlantic AMV similar to that observed during the twentieth century with respect to multidecadal frequency, amplitude, and a sign of coherent changes in the North Atlantic and the Arctic. Associated zonal heat flux anomalies between selected AMV extremes are shown, including the anomaly from the MPI D16 simulation, which corresponds to the temperature trend pattern shown in Fig. 2c. Empirical estimates from the National Centers for Environmental Prediction (NCEP) reanalysis (Kalnay et al. 1996) may be illustrated by two recent transitions from the high AMV phase in 1951–55 and the low phase in 1968–76 and back to the high phase in 1998–2006 (Fig. 4b). The latter estimates are only indicative of internal AMV variation as (in contrast to the control climate model simulations) they contain an externally forced signal. However, the structure is somewhat similar and the amplitude of the anomalies compares well to the model estimates.
Changes in the Arctic sea ice, associated with the AMV, cause intense heat loss in wintertime and were found to have a strong impact on the atmosphere (van der Swaluw et al. 2007). Sea ice anomalies, in particular in the Barents Sea, induced by anomalous oceanic heat transport may be further amplified by a positive feedback between the heat inflow and sea ice (Bengtsson et al. 2004; Mysak and Venegas 1998; Semenov 2008), and this Arctic heat loss on the order of up to about 30 W m−2 annually in some regions (Fig. 3) is of particular importance here. The implied heat transport change over the NA–Arctic sector (40–90°N, 90°W–60°E) estimated from the turbulent heat fluxes corresponds to 0.09 PW in the MPI D16 simulation. Estimates between high and low AMV phases from control integrations with the other global climate models are also on the order of about 0.1 PW, corresponding to about 10% of the total northward heat transport by the MOC at 25°N. Hydrographic data from the second half of the twentieth century confirm this estimate (Huck et al. 2008). The implied heat transport change from the 1970s up to the present [(1998–2006) − (1968–76)] estimated from NCEP reanalysis amounts to about 0.14 PW. However, it should be pointed out again that observational estimates contain an externally driven component.
3. Model setup
The global climate response to an increase in heat transport in response to an internal multidecadal climate variation, which by definition does not contain an externally forced part, was studied with an atmospheric general circulation model (ECHAM5) coupled to a fixed-depth (50 m) ML (ECHAM5–ML; Roeckner et al. 2003) by driving it with the anomalous AMV-related heat flux from the MPI D16 simulation. This flux (monthly mean climatology) was calculated (as for SAT; see Fig. 2c) as a difference between the same two 30-yr periods corresponding to a negative and a positive AMV extreme. The annual mean pattern is shown in Fig. 3. The atmosphere model was run at T31 horizontal resolution with 19 vertical levels. The coupled model ECHAM5–ML uses Q flux derived from a stand-alone integration with ECHAM5 forced by observed SST/sea ice climatology [Atmospheric Model Intercomparison Project version 2 (AMIP2); Hurrell et al. 2008].
Three sensitivity integrations have been conducted. In one, the heat flux pattern is applied over the NA–Arctic sector (ocean area: 40°–90°N, 70°W–80°E) covering less than 6% of the NH area. In two others, the forcing was further restricted either to the NA (40°–60°N, 70°W–80°E) or Arctic region [(60°–90°N, 70°W–80°E); see denoted regions in Fig. 3]. All experiments described here have a length of 100 yr, and the last 80 yr were used in the analyses found later in the study. There is no long-term drift in the experiments but there is a considerable amount of interannual–decadal variability. The experimental setup allows estimating the effect of persistent regional surface heating of the atmosphere caused by deep ocean dynamics. Similar to Zhang et al. (2007), we attempt to quantify the contribution of AMV to NH and global SAT variability. Here, however, we focus solely on internally generated changes not only in the NA but also in the Arctic sector, and we assess the relative contributions of these two regions. We note that this experimental setup does not allow a separation of internally and externally forced SAT changes observed in the recent decades. Instead, it provides an assessment of whether internally generated AMV may have potentially contributed to the recent acceleration in NH warming since 1980.
4. Results
The annual mean SAT response to the imposed heat flux forcing averaged over the Northern Hemisphere amounts to 0.39°C and globally to 0.24°C (Table 2). The observed warming during 1978–2007 obtained from the Goddard Institute for Space Studies (GISS) dataset (Hansen et al. 1999) amounts to 0.68° and 0.45°C, respectively. The zonally averaged NH SAT response in all experiments with ECHAM5–ML is shown in Fig. 5 in comparison with NCEP reanalysis data, different observational estimates of the temperature trend for 1978–2007, and the multimodel mean trend simulated by the CMIP3 model ensemble (for the same period) in the twentieth-century integrations with observed natural and anthropogenic forcing. The linear superposition of the NA and Arctic experiments are also shown. CMIP3 multimodel mean yields stronger warming in the tropics and weaker warming in the Polar region than the observations. Although there is considerable spread among the observational estimates, this result appears to be robust. The horizontal SAT response pattern (Fig. 2d) compares well with the observed trend pattern during 1978–2007 (Fig. 2a) in the Northern Hemisphere, with the strongest warming over the NA–Arctic sector and a realistic latitudinal structure in the Atlantic. This lends further support to the picture that an internal multidecadal fluctuation in the North Atlantic could have considerably contributed to the recent Northern Hemisphere warming.
The two sensitivity experiments with the forcing restricted to either the NA or Arctic region reveal that the Arctic part of the anomalous forcing contributes about 60% to the total NH warming in the NA+Arctic experiment (Figs. 2f and 5; Table 2). Although the linear superposition of the two responses basically adds up to the NH and globally averaged SAT changes in the combined (NA+Arctic) experiment (Table 2), this is not the case everywhere. This is expected given the nonlinear nature of the atmospheric circulation and is not further discussed here. The results are qualitatively reproduced by a simple zonally averaged energy balance model with diffusive heat transport (North 1975), which is not shown here.
An MOC-driven heat release in the Northern Hemisphere may be compensated by an opposite anomaly in the Southern Hemisphere (SH), an issue which we did not directly address by our experiments. However, the comparison of the original climate model SAT anomalies (Fig. 2c) with those simulated by ECHAM5–ML when forced only in the NA–Arctic region (Fig. 2d) shows that mostly the Southern Hemisphere and thus global SAT will be affected by the compensation. Indeed, the SAT change averaged over the Northern Hemisphere is virtually identical in the two runs, whereas the globally averaged SAT change is, as expected, considerably weaker in the original climate model run (0.18 versus 0.24 K).
The Northern Hemisphere winter [December–February (DJF)] sea level pressure (SLP) response in the full (NA+Arctic) experiment (Fig. 6a) exhibits some similarities to the linear trend pattern observed during the recent decades as obtained from NCEP reanalysis (Fig. 6b). ECHAM5 ML simulates anomalously low pressure over the Arctic and anomalously high pressure farther to the south over the North Atlantic and North Pacific, features also seen in the NCEP trend pattern, although the latter is (as expected) characterized by a much “noisier” pattern. Furthermore, the simulated and observed centers of action do not correspond well. The model response in the centers of action is statistically significant at the 95% level applying a t test. The response pattern over the North Atlantic strongly projects onto the pattern of NAO; the observed SLP trend in the recent decades projects only weakly. We note, however, a rather strong sensitivity to the exact period chosen. The simulated teleconnection to the North Pacific may explain the surface warming in the western North Pacific (Fig. 2d) due to reduced cold air advection.
The vertical structure of the Northern Hemisphere temperature response is compared to the trend derived from NCEP reanalysis for 1978–2007 (Fig. 7). Consistent with NCEP, the model simulates considerable warming of the whole troposphere north of 30°N, with the strongest warming in the lower Arctic troposphere (Fig. 7a). The model’s troposphere warming is weaker, as is expected without the inclusion of external forcing, and it does not extend into the tropics and the Southern Hemisphere; however, overall, the results suggest that internal variability could have significantly contributed to the observed change in the vertical temperature structure. The stratosphere cooling, a prominent feature of the observed trend pattern (Fig. 7b) and one of the most robust fingerprints of enhanced greenhouse warming (Hegerl et al. 2007), cannot be simulated with our model setup considering only the effect of internal variability. This indicates the important role anthropogenic forcing played during the recent decades in driving global climate change.
5. Discussion
We have investigated the atmospheric response to surface heat flux anomalies in the North Atlantic–Arctic sector originating from a peak-to-peak change in AMV. This was performed by forcing a coupled atmosphere–ocean mixed layer model by regionally confined surface heat fluxes that were simulated in a control run with a global climate model in association with one particular AMV realization. The latter is by definition internally driven and does not require external forcing. The heat flux forcing was restricted to the North Atlantic–Arctic sector to specifically study the impact of that region on Northern Hemisphere and global climate. The results show that such an internal and regionally confined climate variation can drive relatively large surface climate anomalies on regional, hemispheric, and even global scales. The Arctic plays an important role in this, explaining about 60% of the total Northern Hemisphere SAT response in our experiment.
What are the implications of our work? The first implication concerns the hypothesis of whether a strong AMV change could have considerably contributed to the Northern Hemisphere surface warming during the recent decades. Our model experiments suggest that natural internal multidecadal variability originating in the North Atlantic and associated processes in the Arctic could have played an important role as the response is relatively large and statistically significant and because the simulated patterns resemble some of the large-scale parts of the observed trend patterns during the recent decades. Ocean model simulations with prescribed observed surface forcing indeed show a strengthening of the MOC during the recent decades and stabilization during the recent years (Böning et al. 2006). One may therefore speculate that there is an increasing probability that the AMV may return back to a neutral or negative state within the next few years or decades, counteracting the long-term anthropogenic warming trend. This could potentially offset global warming for some time until the AMV swings back, which would be in line with the prediction of Keenlyside et al. (2008). However, our results should be considered as suggestive but by no means as proof for internal variability being a major contributor to the global warming acceleration in the recent decades. We have shown that AMV changes do at least have the potential to strongly affect Northern Hemisphere averaged SAT, which is also confirmed by a number of global climate models. We would like to emphasize that this study does not question the existence of a long-term anthropogenic warming trend during the twentieth century.
A second implication relates to tipping points (Lenton et al. 2008), delicate thresholds where a slight rise in the earth’s temperature can cause a dramatic change in the environment that itself triggers a far greater increase in global temperatures. The Arctic sea ice loss has been dramatic during the recent years and a critical threshold may have been crossed (Lindsay and Zhang 2005; Stroeve et al. 2007), which would lead to an unavoidable complete loss of sea ice in the Arctic summer during the next decades. If a threshold has really been reached, it may not have been crossed so soon without the internal fluctuation. This indicates that the presence of strong internal variability could lead to extreme climate changes even in the presence of an only moderate anthropogenic signal.
Acknowledgments
Discussions with J. Willebrand, C. Böning, and R. Greatbatch are acknowledged. This work was supported by the European Union’s project THOR, the project “Nordatlantik” of BMBF, the SFB 754 of the German Research Foundation, and NATO Collaborative Linkage Grant and Russian Foundation for Basic Research. This paper is a contribution to the Excellence Cluster “The Future Ocean.” The model runs were performed at the Northern Germany High-Performance Computing Center (HLRN).
REFERENCES
Alexeev, V. A., P. L. Langen, and J. R. Bates, 2005: Polar amplification of surface warming on an aquaplanet in “ghost forcing” experiments without sea ice feedbacks. Climate Dyn., 24 , 655–666.
Bengtsson, L., V. A. Semenov, and O. M. Johannessen, 2004: The early twentieth-century warming in the Arctic—A possible mechanism. J. Climate, 17 , 4045–4057.
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. doi:10.1029/2006GL026906.
Broccoli, A. J., K. W. Dixon, T. L. Delworth, T. R. Knutson, R. J. Stouffer, and F. R. Zeng, 2003: Twentieth-century temperature and precipitation trends in ensemble climate simulations including natural and anthropogenic forcing. J. Geophys. Res., 108 , 4798. doi:10.1029/2003JD003812.
Brohan, P., J. J. Kennedy, I. Harris, S. F. B. Tett, and P. D. Jones, 2006: Uncertainty estimates in regional and global observed temperature changes: A new dataset from 1850. J. Geophys. Res., 111 , D12106. doi:10.1029/2005JD006548.
Delworth, T. L., and T. R. Knutson, 2000: Simulation of early 20th century global warming. Science, 287 , 2246–2250.
Delworth, T. L., and M. E. Mann, 2000: Observed and simulated multidecadal variability in the Northern Hemisphere. Climate Dyn., 16 , 661–676.
Delworth, T. L., S. Manabe, and R. J. Stouffer, 1993: Interdecadal variations of the thermohaline circulation in a coupled ocean–atmosphere model. J. Climate, 6 , 1993–2011.
Hansen, J., R. Ruedy, J. Glascoe, and M. Sato, 1999: GISS analysis of surface temperature change. J. Geophys. Res., 104 , 30997–31022.
Hegerl, G. C., and Coauthors, 2007: Understanding and attributing climate change. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 663–745.
Holland, M. M., and C. M. Bitz, 2003: Polar amplification of climate change in coupled models. Climate Dyn., 21 , 221–232.
Huck, T., A. Colin de Verdière, P. Estrade, and R. Schopp, 2008: Low-frequency variations of the large-scale ocean circulation and heat transport in the North Atlantic from 1955–1998 in situ temperature and salinity data. Geophys. Res. Lett., 35 , L23613. doi:10.1029/2008GL035635.
Hurrell, J. W., J. J. Hack, D. Shea, J. M. Caron, and J. Rosinski, 2008: A new sea surface temperature and sea ice boundary dataset for the Community Atmosphere Model. J. Climate, 21 , 5145–5153.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77 , 437–471.
Keenlyside, N. S., M. Latif, J. Jungclaus, L. Kornblueh, and E. Roeckner, 2008: Advancing decadal-scale climate prediction in the North Atlantic sector. Nature, 453 , 84–88.
Kerr, R. A., 2005: Climate change: Atlantic climate pacemaker for millennia past, decades hence? Science, 309 , 41–43.
Knight, J. R., R. J. Allan, C. K. Folland, M. Vellinga, and M. E. Mann, 2005: A signature of persistent natural thermohaline circulation cycles in observed climate. Geophys. Res. Lett., 32 , L20708. doi:10.1029/2005GL024233.
Knight, J. R., C. K. Folland, and A. A. Scaife, 2006: Climate impacts of the Atlantic multidecadal oscillation. Geophys. Res. Lett., 33 , L17706. doi:10.1029/2006GL026242.
Kravtsov, S., and C. Spannagle, 2008: Multidecadal climate variability in observed and modeled surface temperatures. J. Climate, 21 , 1104–1121.
Latif, M., and Coauthors, 2004: Reconstructing, monitoring, and predicting multidecadal-scale changes in the North Atlantic thermohaline circulation with sea surface temperature. J. Climate, 17 , 1605–1614.
Lenton, T. M., H. Held, E. Kriegler, J. W. Hall, W. Lucht, S. Rahmstorf, and H. J. Schellnhuber, 2008: Tipping elements in the earth’s climate system. Proc. Natl. Acad. Sci. USA, 105 , 1786–1793.
Lindsay, R. W., and J. Zhang, 2005: Thinning Arctic Sea ice: Have we passed a tipping point? Bull. Amer. Meteor. Soc., 86 , 325–326.
Manabe, S., and R. J. Stouffer, 1999: The role of thermohaline circulation in climate. Tellus, 51A , 91–109.
Martin, T., and E. Ruprecht, 2007: Decadal variation of the North Atlantic meridional heat transport and its relation to atmospheric processes. Geophys. Res. Lett., 34 , L04703. doi:10.1029/2006GL028438.
Mysak, L. A., and S. A. Venegas, 1998: Decadal climate oscillations in the Arctic: A new feedback loop for atmosphere–ice–ocean interactions. Geophys. Res. Lett., 25 , 3607–3610.
North, G. R., 1975: Theory of energy-balance climate models. J. Atmos. Sci., 32 , 2033–2043.
Roeckner, E., and Coauthors, 2003: The atmospheric general circulation model ECHAM 5. Part I: Model description. Max Planck Institute for Meteorology Rep. 249, 140 pp.
Schmittner, A., J. Chiang, and S. Hemming, Eds. 2007: Ocean Circulation: Mechanisms and Impacts. Geophys. Monogr., Vol. 173, Amer. Geophys. Union, 392 pp.
Semenov, V. A., 2008: Influence of oceanic inflow to the Barents Sea on climate variability in the Arctic region. Dokl. Earth Sci., 418 , 91–94.
Serreze, M. C., and J. A. Francis, 2006: The arctic amplification debate. Climatic Change, 76 , 241–264.
Stott, P. A., S. F. B. Tett, G. S. Jones, M. R. Allen, J. F. B. Mitchell, and G. J. Jenkins, 2000: External control of 20th century temperature by natural and anthropogenic forcings. Science, 290 , 2133–2137.
Stroeve, J., M. M. Holland, W. Meier, T. Scambos, and M. Serreze, 2007: Arctic sea ice decline: Faster than forecast. Geophys. Res. Lett., 34 , L09501. doi:10.1029/2007GL029703.
Sutton, R. T., and D. L. R. Hodson, 2005: Atlantic Ocean forcing of North American and European summer climate. Science, 309 , 115–118.
Ting, M., Y. Kushnir, R. Seager, and C. Li, 2009: Forced and internal twentieth-century SST trends in the North Atlantic. J. Climate, 22 , 1469–1481.
Trenberth, K. E., and Coauthors, 2007: Observations: Surface and atmospheric climate change. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 235–336.
van der Swaluw, E., S. S. Drijfhout, and W. Hazeleger, 2007: Bjerknes compensation at high northern latitudes: The ocean forcing the atmosphere. J. Climate, 20 , 6023–6032.
Wang, M. Y., J. E. Overland, V. Kattsov, J. E. Walsh, X. D. Zhang, and T. Pavlova, 2007: Intrinsic versus forced variation in coupled climate model simulations over the Arctic during the twentieth century. J. Climate, 20 , 1093–1107.
Zhang, R., T. L. Delworth, and I. M. Held, 2007: Can the Atlantic Ocean drive the observed multidecadal variability in Northern Hemisphere mean temperature? Geophys. Res. Lett., 34 , L02709. doi:10.1029/2006GL028683.
Observed (a) NH and (b) Arctic annual mean SAT anomalies (°C) from the surface temperature dataset compiled by the Climatic Research Unit of the University of East Anglia (Brohan et al. 2006), CRUTEM3 (red), and the ensemble of twentieth-century simulations with the CMIP3 models using both anthropogenic and natural radiative forcing. The ensemble mean is given by the thick black line; the shading shows the range in which 90% of the individual model realizations lie. Model data were masked (in respect to missing values) as the observational data.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3347.1
(a) The SAT trend pattern [°C (30 yr)−1] during 1978–2007 from GISS observations and (b) the multimodel ensemble mean in comparison to the simulated SAT anomalies (°C): (c) the climate model (coupled GCM) control run from which the forcing for ECHAM5–ML was derived; (d) the NA+ARCTIC experiment in which the forcing was applied in both the North Atlantic and the Arctic; (e) the NA-only experiment in which the forcing was applied only in the North Atlantic; and (f) the Arctic-only experiment in which the forcing was applied only in the Arctic. (upper right of panel) Numbers are global (NH)-averaged values.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3347.1
Annual mean turbulent fluxes (W m−2) from the MPI D16 control simulation, associated with a transition from cold to warm AMV phase, used in the experiments with the ECHAM5–ML model. Also shown are the different regions where the forcing was applied in the different experiments. A positive sign indicates a flux from the ocean to the atmosphere.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3347.1
(a) Zonally averaged surface turbulent heat flux anomalies (W m−2) in the Atlantic sector (70°W–80°E) corresponding to differences between high and low AMV phases (represented by averages for 15-yr periods around corresponding extremes) simulated in different control experiments (see legends) with global climate models. The bold black line is the anomaly used in the NA+ARCTIC experiment. (b) Zonally averaged surface turbulent heat flux anomalies from NCEP reanalysis corresponding to the transitions from the high AMV phase in 1951–55 and the low phase in 1968–76 (magenta) and back to the high phase in 1998–2006 (red). The bold black line again shows the forcing used in the NA+ARCTIC experiment.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3347.1
The zonally averaged observed SAT changes (°C) in the period 1978–2007 from reanalysis [NCEP (red)], two observational datasets [that of the Hadley Centre and Climatic Research Unit of the University of East Anglia, HadCRUT3, (blue) and GISS (purple)], and from the ensemble mean of the CMIP3 models (black) driven with observed natural and anthropogenic forcing. The thin lines denote the individual simulations from the CMIP3 model ensemble. Also shown are the different numerical climate model experiments with ECHAM5–ML. The NA+Arctic experiment is shown in green.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3347.1
(a) The simulated DJF SLP response (hPa) over the Northern Hemisphere in the NA+Arctic experiment and (b) the SLP trend (hPa) from NCEP reanalysis for the period 1978–2007.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3347.1
The simulated temperature anomalies in the NA+Arctic experiment, representing a typical decadal-scale change from a low to a high AMV phase. (a) Simulated zonal mean annual temperature changes (°C) as a function of height and (b) the observed trends (°C) for 1978–2007 as given by NCEP reanalysis.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3347.1
Correlations and linear regression coefficients (°C °C−1) between the North Atlantic SST index and Northern Hemisphere and globally averaged SAT in the different climate model control integrations. The lengths of the integrations are shown in the last column.
SAT changes (relative to the control simulation, in °C) averaged for NH, SH, and the globe (GLOB) as simulated by the ECHAM5–ML model forced by the anomalous surface turbulent heat fluxes over water in the northern North Atlantic and Arctic regions. The NA–Arctic region is defined as the average over 40°–90°N and 90°W–60°E, the northern NA region as the Atlantic portion of the region 40°–60°N, and the Arctic as 60°–90°N, 90°W–60°E. The last row denoted by obs is the observed trend in 1978–2007 computed from the GISS dataset. Numbers in parentheses indicate the model SAT changes after masking the model data using the same missing data regions as in GISS data.
