Low-frequency variability of the arctic climate: the role of oceanic and atmospheric heat transport variations
- 800 Downloads
Changes in meridional heat transports, carried either by the atmosphere (HTRA) or by the ocean (HTRO), have been proposed to explain the decadal to multidecadal climate variations in the Arctic. On the other hand, model simulations indicate that, at high northern latitudes, variations in HTRA and HTRO are strongly coupled and may even compensate each other. A multi-century control integration with the Max Planck Institute global atmosphere-ocean model is analyzed to investigate the relative role of the HTRO and HTRA variations in shaping the Arctic climate and the consequences of their possible compensation. In the simulation, ocean heat transport anomalies modulate sea ice cover and surface heat fluxes mainly in the Barents Sea/Kara Sea region and the atmosphere responds with a modified pressure field. In response to positive HTRO anomalies there are negative HTRA anomalies associated with an export of relatively warm air southward to Western Siberia and a reduced inflow of heat over Alaska and northern Canada. While the compensation mechanism is prominent in this model, its dominating role is not constant over long time scales. The presence or absence of the compensation is determined mainly by the atmospheric circulation in the Pacific sector of the Arctic where the two leading large-scale atmospheric circulation patterns determine the lateral fluxes with varying contributions. The degree of compensation also determines the heat available to modulate the large-scale Arctic climate. The combined effect of atmospheric and oceanic contributions has to be considered to explain decadal-scale warming or cooling trends.
KeywordsDecadal to multidecadal climate variability Barents Sea Arctic climate Global coupled atmosphere-ocean modeling Coupled atmosphere-ocean-sea ice processes
The high northern latitudes have experienced strong climate variations during the last decades (McBean et al. 2005). Some polar regions exhibit, for example, a warming of as much as 2°C per decade (Rigor et al. 2000). This was accompanied by a record minimum in ice extent in summer 2007, which was only 4.28 × 106 km2 and thus about 50% lower than the conditions in the 1950s and 1970s (Stroeve et al. 2008). Climate change projections, such as those discussed in the International Panel of Climate Change (IPCC) Fourth Assessment Report (AR4) indicate that the already ongoing warming due to enhanced anthropogenic CO2 emissions is most pronounced over the continents and over the high latitudes (Meehl et al. 2007). On the other hand, large decadal to multidecadal variations have also been reported from the Arctic that are likely unrelated to external forcing (Overpeck et al. 1997; Polyakov and Johnson 2000; Schmith and Hansen 2003). Long-term variations of surface air temperature (SAT) are largest in the high northern latitudes (Johannessen et al. 2004; Kuzmina et al. 2008) with a pronounced warming in the 1930s and 1940s similar in magnitude to the late 20th/early 21st century. To discriminate between anthropogenically forced global warming and natural variability of the atmosphere-ocean-sea ice system it is therefore necessary to improve our understanding of the processes involved in the latter.
Various mechanisms have been proposed to explain the strong variations in the Arctic SAT record and a review was recently given by Goosse and Holland (2005). Most prominent are studies that relate the observed Arctic warming to recent trends in the North Atlantic Oscillation (NAO) or Arctic Oscillation (AO) (e.g. Dickson et al. 2000). More recently, analyzes have been extended to higher order empirical orthogonal functions (EOF) of the atmospheric circulation. Quadrelli and Wallace (2004) state that much of the Northern Hemisphere climate variations can be represented in terms of the two leading EOFs of sea level pressure (SLP) and Overland and Wang (2005) describe the Arctic climate variations for the second half of the 20th century by a phase space trajectory based on the two leading EOFs. Other studies emphasize the role of the Atlantic meridonal overturning circulation (AMOC). AMOC variations are related to pronounced changes in meridional heat transport and thus may influence Arctic climate (e.g. Polyakov and Johnson 2000). However, since observational data are scarce, this is mainly based on results from coupled ocean-atmosphere models (Delworth et al. 1993, 1997; Jungclaus et al. 2005). Ikeda et al. (2001), Bengtsson et al. (2004), and Goosse and Holland (2005) showed that increased heat transport into the Barents Sea affects the local interactions between atmosphere and ocean. The associated warming in the Barents Sea reduces the SLP that feed back positively on ocean circulation and heat transport. In summary, meridional heat transfer variations into the Arctic either through large-scale atmospheric or ocean circulation changes or through (locally) coupled processes play the dominant role. This is particularly evident during polar night, when the lateral fluxes determine surface air temperature fluctuation almost entirely.
Two recent publications, Shaffrey and Sutton (2006) and van der Swaluw et al. (2007) analyzed the decadal-scale variations in ocean and atmosphere heat transport using data from long control integrations with the HadCM3 climate model. They find that both components are anticorrelated between 30 and 80°N and that this anticorrelation is strongest around 70°N. Not only are the heat transports anticorrelated, their anomalies are also of the same size and the authors name the process “Bjerknes Compensation” (BC) referring to Bjerknes (1964). The mechanism behind the BC is given as follows: positive oceanic heat transport anomalies (related, for example to AMOC variations) causes enhanced warming and sea ice melt in high northern latitudes. The absence of sea ice leads to drastic ocean-to-atmosphere heat flux changes warming the overlying air. The reduced meridional temperature gradient in the atmosphere leads to reduced baroclinicity and reduced (eddy) heat transport (van der Swaluw et al. 2007).
Given the sensitivity of the high-latitude climate to changes in meridional heat transport, we wish to analyze the role of the BC mechanism in shaping the Arctic climate. If the heat transport compensation were complete, then this mechanism would tend to smooth Arctic-wide climate variations due to lateral changes, since variations in one component would be compensated by the other. Furthermore, BC, as a negative feedback, would act against positive feedbacks, such as those invoked to explain the 1930s to 1940s warming in the Arctic. Here we analyze the mechanisms of ocean and atmospheric heat transport variation in detail, demonstrate their regional impacts, and relate them to decadal to multidecadal Arctic climate variations.
2 Model and experiment description
3.1 Decadal-scale heat transport variations and ocean-atmosphere interaction
3.1.1 Mean oceanic and atmospheric heat and energy transports
3.1.2 Decadal scale heat transport variations and ocean atmosphere interaction
Overall, HTRO70 and HTRA70 anomalies tend to be of opposite sign and similar magnitude (Fig. 4a), but the compensation is not always perfect. In fact, comparing the HTRA70 time series with the respective one including all three terms from the RHS of Eq. (2), reveals that both the ocean heat storage term and the TOA terms are often of considerable amplitude. Indeed, differences between the black line and the blue line in Fig. 4a indicate periods where all terms of the RHS of Eq. (2) are needed to compensate HTRO70 variations. The respective contributions from the time-derivative of ocean heat content change and TOA radiation are of similar magnitude (Fig. 4b) but the amplitude of the residual appears to be more strongly controlled by the heat content changes. The most prominent example where there is no compensation is the period 2175–2225. HTRO70 anomalies are sustained positive (Fig. 4a) and lead to considerable increase in the heat content change (Fig. 4b) (the green curve is negative here because of the sign convention in Eq. (2)). From 2200 to 2225 HTRA70 anomalies are also positive and the heat surplus has to be radiated off to space leading to negative TOA anomalies (Fig. 4b). At other times, for example around the years 2525, HTRA70 is opposite in sign to HTRO70 but the higher amplitude of HTRA70 has to be compensated by contributions mainly from the TOA term (Fig. 4b).
The implied atmospheric heat transports calculated from Eq. (1) is very similar, but not identical to the advective energy transport according to Eq. (3) (Fig. 4c). This is due to different assumptions, e.g. neglecting diffusive transports in Eq. (3) or neglecting heat storage variations in the atmosphere in Eq. (2). Equation 3 allows us to discriminate between the dry static (first and second term in the integrand) and latent energy (third term) transports. For the decadally smoothed time series, Fig. 4c indicates that the variability is mostly determined by the dry static part. However, considerable contributions are seen, for example, around the year 2200 and the year 2225. This indicates that contributions from latent heat changes may disturb the compensation mechanism.
3.1.3 Long-term changes in the relation between oceanic and atmospheric heat transports
3.2 Regional response to heat transport variations
In summary, even though the interaction between ocean and atmosphere takes mainly place in the Atlantic sector, the Pacific sector plays a crucial role in the atmospheric response. The regional differences in the atmospheric circulation response lead to the dipolar temperature response in the Northern Hemisphere SATs (Fig. 6b).
3.3 Heat transport variations and pan-Arctic temperature changes
3.4 Variations of the large-scale atmospheric circulation patterns
A phase-space diagram that is based on the two leading EOFs (Fig. 11b) suggests that there is a stronger contribution from the second EOF pattern at times when the compensation rate is particularly low.
In summary, we find that there is an atmospheric circulation pattern associated with the compensation mechanism (Figs. 6a, 10a). In terms of the atmospheric heat transport variations at 70°N the compensation depends strongly on the Pacific sector where also the second EOF has a pronounced center of action and the relative importance of the leading patterns is changing with time. The reason for these shifts are quite hard to identify but the particular importance of the Pacific center of action suggest some similarity to the observed variation in the role of the first and second EOF for the SLP in the North Pacific (Wallace and Thompson 2002; Zhang et al. 2008; Zhao and Moore 2009). These authors talk about “regime shifts” but the mechanisms behind these shifts in the atmospheric circulation regimes are difficult to identify. This requires dedicated model studies, applying, for example idealized forcing or partially coupled experiments where the variability in certain region is suppressed. Recently, Jia et al. (2009) have performed experiments using a simplified atmospheric general circulation model and analyzed the response to different idealized tropical Pacific forcings. They find that there are different propagation pathways for disturbances in the eastern and western tropical Pacific, respectively. However, to find out if the mechanisms discussed by Jia et al. (2009) are also at work in our more complex model would require carrying out such sensitivity studies for which we don’t have the computational resources at hand.
3.5 Relation to observed climate variability
It is interesting to compare this sequence of events with the description of the “real” decades of the 1920s, 1930s, and 1940s by Overland and Wang (2005). They summarize that the early period, roughly 1920–1927, was dominated by a positive phase of AO, or more locally the North Atlantic Oscillation. The following years were characterized by a particular pattern of SLP and temperature anomalies with warm temperatures in both Europe and west Greenland associated with a low-pressure trough over Iceland and Canada, a feature which carries some elements of the third EOF of Northern Hemisphere SLP. The final period (1940–1942) had warm surface temperatures in Eastern Siberia, Alaska, and northeastern Canada related to SLP anomalies with resemblance to the positive PNA pattern. We do not claim that our experiment gives an exact picture of the real early 20th century climate evolution but the model results indicate that a strong warming event may be the result of a combination of climate patterns with different underlying mechanisms. The variations in the oceanic heat transport alone (e.g., Polyakov and Johnson 2000) or associated with local feed back effects (Bengtsson et al. 2004) would not be sufficient to explain a sustained warming. In the early part of the warming, the atmosphere responds to the HTRO70 anomalies with compensation (Fig. 4a), but, while HTRO70 anomalies remain positive, HTRA70 anomalies become also positive due to the shift in the circulation regime. Therefore we conclude that ocean and atmospheric heat transport must act in concert to achieve an Arctic-wide warming of the magnitude observed in the early to mid 20th century.
Sorteberg and Kvingedal (2006) studied the atmospheric forcing on the Barents Sea winter ice extent for the period 1967–2002 by analyzing the path and number of synoptic cyclones entering the Arctic. They identified two regions that influence the Barents Sea ice extent: the variability of the northward-moving cyclones traveling into the Arctic from the easternmost part of Siberia and the cyclone activity of northward moving synoptic system over the western Nordic Seas. They emphasize that the latter modulate the oceanic heat input into the Barents Sea. They also find that the Nordic Seas component co-varies with the AO/NAO while the East Siberian component does not. These findings are consistent with the model results discussed here. Ocean and atmosphere heat transport anomalies in the Atlantic sector clearly show a pronounced influence on the Barents Sea ice extent and Arctic-wide SAT but contributions from the East Siberian/Pacific sector of the Arctic need to be taken into account. They influence the Barents Sea ice extent by changing Arctic winds and ice advection.
4 Summary and conclusions
In this study we have investigated the decadal-scale variability of atmospheric and oceanic heat transports in a coupled climate model with a focus on their role in shaping the climate of the high northern latitudes. Recent model-based studies have demonstrated that the strong ocean-atmosphere coupling due to variations of the sea ice cover (and the associated enormous heat flux anomalies) could lead to the so-called Bjerknes Compensation where the atmosphere responds with reduced transient eddy transport to positive anomalies in oceanic heat transport. Decadal-scale variations in HTRO and HTRA are indeed anticorrelated most pronounced near 70°N in ECHAM5/MPIOM. The atmosphere responds to positive HTRO anomalies with a modified circulation that leads to increased heat outflow across 70°N towards Siberia and to reduced heat inflow over Alaska and Canada. This results in a dipolar temperature response and a moderation of pan-Arctic temperature changes.
The degree of compensation determines the heat transport anomalies available to modulate the large-scale Arctic climate (defined here as the 70°–90°N SAT average). The combined effect of HTRA and HTRO (HTRSUM) has to be considered to explain decadal-scale warming or cooling trends. In fact, the model simulation suggests that ocean and atmosphere have to act in concert to create drastic temperature changes, such as the early-20th century warming.
While the compensation mechanism is prominent, the compensation rate also changes over longer (century) time scales. These changes are related to shifts in the large-scale atmospheric circulation. The “compensation mode”, associated with the oceanic variations in the Atlantic sector, is characterized by an AO-like SLP pattern that is sometimes overruled by a more PNA-like EOF2 pattern. Consistent with observations, both the AO-like and the PNA-like SLP anomaly pattern influence the SAT and the circulation regimes related to particular combinations of the patterns vary over decadal and longer time scales. A similar conclusion can be drawn from observations of 20th century climate (Overland and Wang 2005). Moreover, analyzing data from 1864 to 1998, Vinje (2001) demonstrated that there is a fairly high correlation between April sea ice in the Barents Sea with the NAO, but this relationship is not stationary over times.
The reasons for these regime shifts are still unknown. One mechanism that has been proposed to be important for the Arctic circulation in general is the phase of the planetary-scale SLP wave 1 (Cavalieri and Häkkinen 2001). Further analysis of the model experiment here and additional sensitivity experiments together with an assessment of how realistically climate models simulate the characteristics of Arctic climate variations and their remote forcing have to be carried out. A key question is how changes (most likely in the tropical Pacific) may interact with the North Atlantic and the North Pacific (e.g., Jia et al. 2009). Another important question that will be addressed in an upcoming paper is the future evolution of the interaction between oceanic and atmospheric heat transports in a global warming context with much reduced ice cover and ice thickness.
TK was supported by the Deutsche Forschungsgemeinschaft through the Sonderforschungsbereich 512. The computations have been performed by the Deutsches Klima Rechenzentrum (DKRZ).
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.
- Bjerknes J (1964) Atlantic air-sea interaction advances in geophysics, vol 10. Academic Press, New York, pp 1–82Google Scholar
- Jia X, Lin H, Derome J (2009) The influence of tropical Pacific forcing on the Arctic Oscillation. Clim Dyn 32:495:509, doi: 10.1007/s00382-008-041-y
- Johannessen OM (2004) Arctic climate change–observed and modeled temperature and sea ice variability. Tellus 56A:328–341 Co authorsGoogle Scholar
- Johannessen OM, Myrmehl C, Olsen AM, Hamre T (2002) Ice cover data analysis—Arctic. AICSEX Tech Rep 2. Nansen Environmental and Remote Sensing Center, BergenGoogle Scholar
- Keith DW (1995) Meridional energy transport: uncertainty in zonal means. Tellus 47A:30–44Google Scholar
- Koenigk T, Mikolajewicz U, Jungclaus JH, Kroll A (2008) Sea ice in the Barents Sea: seasonal to interannual variability and climate feedbacks in a global coupled model. Clim Dyn published online. doi: 10.1007/s000382-008-0450-2
- Kuzmina SI, Johannessen OM, Bengtsson L, Aniskina OG, Bobylev LP (2008) High northern latitude surface air temperature: comparison of existing data and cration of a new gridded data set 1900–2000. Tellus 60A:289–304Google Scholar
- McBean G et al (2005) Arctic climate: past and present. In: Symon et al (eds) Arctic climate impact assessment, vol 2. Cambridge University Press, New York, pp 22–60 Co authorsGoogle Scholar
- Meehl GA et al (2007) Global climate projections. In: Solomon S et al (eds) Climate change 2007: the physical basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, vol 10. Cambridge University Press, Cambridge, UK, pp 747–845 Co authorsGoogle Scholar
- Overland JE, Turet P (1994) Variability of the atmospheric energy flux across 70 N computed from the GFDL data set. In: Johannessen OM (ed) The Polar Oceans and their role in shaping the global environment. American Geophysical, Union Geophysical Monograph 84, Washington, pp 313–325Google Scholar
- Roeckner E and coauthors (2003) The atmosphere general circulation model ECHAM5, part 1: Model description. Max-Planck-Institut für Meteorologie, Report No. 349: 127 ppGoogle Scholar
- Von Storch H, Zwiers FW (1999) Statistical analysis in climate research. Cambridge University Press, Cambridge, p 484Google Scholar
- Zhao H, Moore GWK (2009) Temporal variability in the expression of the Arctic Oscillation in the North Pacific. Accepted for publication, J ClimGoogle Scholar