The Arctic ocean is unique due in part to its perennial ice cover. The sea ice has a large seasonal cycle, reaching a maximum extent of approximately 15 million km2 in March and reducing to approximately 7 million km2 by September. The ice exhibits high spatial variability with many different surface types and ice thickness occurring in a relatively small region (Figure 1). The thickest ice cover of greater than 6m, occurs along the Canadian Arctic Archipelago and the north Greenland coast. The Eurasian coastline has the thinnest ice cover of less than 1 m (Figure 2). This results in a mean thickness of approximately 3m within the Arctic basin. The spatial distribution of ice thickness is largely determined by the ice motion which consists of an anti-cyclonic Beaufort gyre and a trans-polar drift stream. The ice motion field is primarily wind-driven and causes the removal of ice from the Siberian coast and the transport of ice from the Arctic to the Greenland Sea via Fram Strait. This results in an average net transport of approximately 2800 km3 per year (Aagaard and Carmack, 1989) into the northern North Atlantic.
The Arctic region is important in that there are large feedback mechanisms associated with changes in snow and sea ice which amplify climate change and variability. Variations in Arctic sea ice are forced by both atmospheric and oceanic conditions. In turn, sea ice variations modify the atmosphere and ocean systems through a number of processes. Firstly, sea ice is highly reflective which modifies the global energy budget. Second, it is an efficient insulator between the atmosphere and ocean, affecting the exchange of heat, moisture and momentum. And finally, ice growth and melt affects the ocean buoyancy. When ice grows from the sea water, the salt is rejected which increases the salinity and density of the underlying ocean. Ice melt produces the opposite effect. Increasing atmospheric CO2 scenarios have shown that these processes modify the global climate and that an amplified warming occurs at high latitudes (Figure 3) largely due to a reduction in sea ice (IPCC, 1995). This suggests that climate change may first be detected in polar regions.
Observational studies suggest that the Arctic region has indeed undergone substantial changes over the last 30-40 years (Dickson, 1999). These include trends in the ice thickness, ice extent, ocean temperature and salinity, and atmospheric circulation and temperature. Using a comparison of submarine sea-ice draft data from two time periods, Rothrock et al. (1999) found that the mean ice draft had decreased by greater than a meter over most of the central Arctic from 1958-1976 to the 1990s (Figure 4). The ice extent also appears to be decreasing in the Arctic region (Figure 5). For example, Maslanik et al. (1996) found a net trend in the summertime ice cover of the Arctic basin and Greenland-Iceland-Norwegian (GIN) Seas region of -0.6% per year from 1978 to 1995. This trend had the largest signature in the Siberian region of the Arctic. As discussed by Parkinson et al. (1999) the decreasing trends in ice concentration occur in all seasons, although they are largest in the spring and fall. Additionally, some regions have experienced an increase in ice extent, including the Labrador Sea.
The trends in Arctic sea ice appear to coincide with atmospheric circulation trends. As discussed by Deser et al. (2000), the relationship between the atmospheric and sea ice variability is consistent with the atmosphere forcing changes in the ice cover. It appears that the increasing trend in the North Atlantic Oscillation (NAO) (Figure 6) or the prevalence of low atmospheric pressure over the Arctic region, is in part responsible for the observed sea ice trends. It also appears that the NAO modifies the export of ice from the Arctic to the GIN Sea region (Kwok and Rothrock, 1999). Additionally, Hilmer and Jung (2000) found that this relationship has changed over time, with low correlations present from 1958-1977 and high correlations present in the last ~20 years (Figure 7) . This is due to changes in the location of the Icelandic low. Thus, not only the strength but the structure of the sea level pressure anomalies are important for determining the ice export and resulting Arctic ice mass balance.
Recent changes have also occurred in the Arctic ocean (Figure 8). In particular, a warming of the Atlantic layer within the Arctic has been observed (Carmack et al., 1995). This appears related to an increased temperature (Swift et al., 1998) and strength (Zhang et al., 1998) of the Atlantic inflow into the Arctic basin. Again, the trends in the NAO may be in part responsible because of the associated air temperature and circulation anomalies it represents over the GIN sea region. Additionally, the cold halocline layer, which insulates the sea ice from the relatively warm Atlantic waters, appears to have retreated from the Eurasian Basin in recent years (Steele and Boyd, 1998). This has important consequences for ice/ocean heat exchange and ice growth rates. The cause of the modified halocline layer is likely related to a diversion of Russian river runoff caused by atmospheric circulation anomalies.
Substantial variability also exists in the Arctic, making the signal of potential climate change difficult to detect above the Arctic noise. For example, Parkinson et al. (1999) found that ice extent in the Arctic Ocean had a positive trend from 1990-1996, although a negative trend was present from 1978-1996. Proshutinsky and Johnson (1997) have identified two circulation regimes of the Arctic Ocean which appear to have a signature in a number of different ice, ocean, and atmospheric variables. These regimes appear to persist for 5-7 years and are related to the location and intensity of the Icelandic low and Siberian high.
If we are to make conclusions about global warming based on Arctic change, a better understanding of sea ice variability in the Arctic is necessary. This lecture presents two different studies which are designed to examine: 1) sea ice related feedbacks that modify climate change and variability and 2) potential effects of sea ice variability on the climate system - particularly the ocean thermohaline circulation.
As mentioned above, sea ice modifies the global energy budget and ice/ocean/atmosphere exchange. This leads to substantial feedbacks that are associated with the ice cover. Model simulations are a useful way of quantifying these feedbacks. This is possible because certain feedback mechanisms can be "turned off"; in model simulations and compared to integrations where all feedbacks are active. For example, the influence of variable ocean conditions (or ice/ocean feedbacks) can be ignored by using a constant annual cycle of oceanic forcing to drive the ice conditions.
The albedo feedback mechanism (Figure 9) is positive, indicating that it amplifies perturbations in the system. If a warming perturbation occurs, this modifies the sea ice state, resulting in a lower surface albedo. This allows more solar radiation to be absorbed and enhances the initial warming perturbation. Here we examine the influence of the albedo feedback on modifying climate variability and change. Climate variability simulations are performed using an ice/ocean mixed layer model for the Arctic. These simulations are driven by variable atmospheric forcing from 1948-1998. Climate change simulations are performed using a global model which includes an dynamic/thermodynamic sea ice component, an ocean general circulation model, and an atmospheric energy-moisture balance model. These simulations are run under increasing atmospheric CO2 conditions.
In the variability simulations, we find that the albedo feedback has a very large effect on the ice area and volume variance in the Arctic. Figure 10 shows the fraction of the Arctic ice area and volume variance that can be attributed to the feedback mechanism. The albedo feedback is most strongly felt in the summer months and is responsible for up to 80% of the variance in the summertime ice area. The feedback has an influence on the ice volume variance throughout the year, with a minimum influence in April and a maximum influence in August. The feedback strength is affected by the ice model parameterizations which are used due to their influence on mean ice conditions and open water formation rates. The results from this study are discussed further in Holland (2001c).
The albedo feedback mechanism also has an influence on atmospheric and oceanic conditions under climate change scenarios (Holland et al, 2001b). Figure 11 shows model results from coupled ice/ocean/atmosphere simulations that have been run for 500 years under increased atmospheric CO2 conditions. The air temperatures have warmed considerably in this simulation, with a maximum 4.5 oC warming present in the Arctic region. By year 500 of the model run, approximately 20% of the global air temperature increase can be attributed to the albedo feedback. The oceanic conditions are also impacted by the albedo feedback mechanism which accounts for approximately 16% of the warming in the upper 1500m of the ocean.
The northern North Atlantic is one of the primary regions of deep water formation in the world's oceans (Figure 12). Changes in the density of the water in these regions can lead to variability in the deep water formation and consequently in the global thermohaline circulation. This in turn can influence the northward transport of heat by the global ocean.
Sea ice plays an important role for the heat and fresh water budgets in the Greenland/Iceland/Norwegian (GIN) and Labrador Sea regions. The most notable example of this in the observational record is the Great Salinity Anomaly that occurred in the late 1960s. This event was a widespread freshening of the northern North Atlantic that was first observed north of Iceland and then propagated around the subpolar gyre (Dickson et al., 1988), reducing convection in the Labrador Sea as it moved through (Lazier, 1980). This freshening of the subpolar gyre was caused by anomalous sea ice and fresh water export from the Arctic (Hakkinen, 1993) which were partly driven by anomalous wind forcing (Serreze et al., 1992).
In this study (Holland et al., 2001a), we examine the influence of Arctic sea ice on the variability of the simulated North Atlantic climate. A global coupled ice-ocean-atmosphere model (Fanning and Weaver, 1996) is used and variability is induced through a stochastic wind stress forcing that is applied to the sea ice motion. These stochastic anomalies have observed spatial patterns but are random in time. Integrations are run for 1000 years from an equilibrium state and the influence of the sea ice variations on the thermohaline circulation (THC) are examined.
We find that the North Atlantic overturning circulation, which is defined to be the maximum value of the meridional overturning streamfunction, has a standard deviation of approximately 2 Sv and preferred interdecadal timescales (Figure 13). This variability is related to anomalous sea surface temperature and salinity within the northern North Atlantic (Figure 14), with warm and saline conditions leading changes in the overturning by 2-3 years. This lead time suggests that the sea surface conditions are forcing the THC variations. The overturning variability is also related to ocean temperature conditions within the Arctic basin, which is reminiscent of recent observations which show a warming of the Arctic ocean (Carmack et al., 1995).
In order to determine the factors which drive the overturning variability, a regression analysis of the ocean density within the northern North Atlantic was performed (Figure 15). The results from this analysis show that 2-3 years prior to a high overturning event, the density is anomalously large due to changes in the ocean salinity. This forces increased deep water formation and an increase in the THC. Coincident with and following a high overturning event, the density decreases due to anomalously warm conditions which are forced by modifications to the oceanic heat transport. Thus, the temperature variations in the northern North Atlantic have a negative feedback and act to damp the overturning strength.
Because we are forcing variability in the wind-driven sea ice motion, we would expect that the ice/ocean interactions might be responsible for the variations in ocean salinity. Figure 16 shows the timeseries of anomalous ice export from the Arctic Basin. The standard deviation of this timeseries is 1200 km3 per year, which is reasonable based on the limited observations which are available (Kwok and Rothrock, 1999). The timeseries is largely white in time, although some significant peaks are present at interdecadal timescales. The ice export variations are due to anomalies in both the velocity and thickness of the ice which leaves the Arctic basin. For high frequencies, the variability in ice velocity dominates, but at low frequencies variations in velocity and thickness are equally responsible for driving the ice export variations. The ice thickness variations are in turn partially determined by changes in ocean heat transport due to the variable THC.
In a correlation of the ice export and overturning timeseries (Figure 17), we find that low ice export leads changes in the overturning by approximately 5 years. This reduction in ice export leads to less ice melt in the northern North Atlantic which destabilizes the water column causing more deep water formation to occur. An increase in the THC results, leading to anomalously high ocean heat transport and reduced ice growth the following winter. The reduced ice growth and warmer ocean counteract the anomalous ocean density and reduce the anomalous ocean circulation. So, the temperature anomalies damp the THC variability due to both direct and indirect (ice growth) effects on the ocean density.
A further integration was performed to examine how the sea ice induced THC variability may change in a climate warming scenario (Figure 18). Under increased atmospheric CO2 forcing, the overturning variability is significantly decreased. This is partially due to lower ice export variability resulting from the thinner ice which is present. Additionally, the primary North Atlantic ice melt location moves northward in the climate warming integration and is further from the primary convective region. This results in a less efficient influence of the ice melt anomalies on the THC strength.
Aagaard, K and E. C. Carmack, 1989: The role of sea ice and other fresh water in the Arctic circulation. J. Geophys. Res., 94, 14485-14498.
Bourke, R.H., and R.P. Garret, 1987: Sea ice thickness distribution in the Arctic Ocean, Cold Regions Sci and Tech., 13, 259-280.
Carmack, E. C., R. W. Macdonald, R. G. Perkin, F. A. McLaughlin and R. J. Pearson, 1995: Evidence for warming of Atlantic water in the southern Canadian Basin of the Arctic Ocean: Results from the Larsen-93 expedition. Geophys. Res. Lett., 22, 1061-1064.
Deser, C., J. Walsh and M. Timlin, 2000: Arctic sea ice variability in the context of recent atmospheric circulation trends. J. Climate, 13, 617-633.
Dickson, R.R., 1999: All change in the Arctic. Nature, 397, 389-391.
Dickson, R.R. , J. Meincke, S. A. Malmberg and A. J. Lee, 1988: The `Great Salinity Anomaly' in the northern North Atlantic 1968-1982. Prog. Oceanogr., 20, 103-151.
Fanning, A.F. and A. J. Weaver, 1996: An atmospheric energy-moisture balance model: Climatology, interpentadal climate change, and coupling to an ocean general circulation model, J. Geophys. Res., 101, 15,111-15,128.
Holland, M.M., C.M. Bitz, M.Eby and A.J. Weaver, in press 2001a: The role of ice-ocean interactions in the variability of the North Atlantic thermohaline circulation. J. Climate.
Holland, M.M., C.M. Bitz and A.J. Weaver, submitted 2001b: The influence of sea ice physics on simulations of climate change. J. Geophys. Res.
Holland, M.M., submitted 2001c: The influence of ice/ocean coupling feedbacks on Arctic sea ice variability, J. Climate
Hurrell, J.W., 1995: Decadal trends in the North Atlantic Oscillation regional temperatures and precipitation, Science, 269, 676-679.
IPCC, 1995: Climate Change 1995: The Science of Climate Change. Cambridge University Press, Cambridge.
Kwok, R. and D.Rothrock, 1999: Variability of Fram Strait ice flux and North Atlantic Oscillation. J. Geophys. Res., 104, 5177-5189.
Lazier, J.R.N., 1980: Oceanographic conditions at Ocean Weather Ship Bravo, 1964-1974. Atmos. Ocean, 18, 227-238.
Maslanik, J.A., M.C. Serreze and R.G. Barry, 1996: Recent decreases in Arctic summer ice cover and linkages to atmospheric circulation anomalies. Geophys. Res. Lett., 23, 1677-1680.
Morison, J, M. Steele, and R. Anderson, 1998: Hydrography of the upper Arctic Ocean measured from the nuclear submarine U.S.S. Pargo, Deep-Sea Res., 45, 15-38.
Parkinson, C.L., D.J. Cavalieri, P.Gloersen, H.J. Zwally and J.C. Comiso, 1999: Arctic sea ice extents, areas, and trends, 1978-1996. J. Geophys. Res., 104, 20837-20856.
Proshutinsky, A.Y. and M.A. Johnson, 1997: Two circulation regimes of the wind-driven Arctic Ocean. J. Geophys. Res., 102, 12493-12514.
Rothrock, D., Y.Yu and G.Maykut, 1999: Thinning of the Arctic sea-ice cover. Geophys. Res. Lett., 26, 3469-3472.
Serreze, M.C., J. A. Maslanik, R.G. Barry and T.L. Demaria, 1992: Winter atmospheric circulation in the Arctic Basin and possible relationships to the Great Salinity Anomaly in the northern North Atlantic, Geophys. Res. Lett., 19, 293-296.
Steele, M. and T.Boyd, 1998: Retreat of the cold halocline layer in the Arctic Ocean. J. Geophys. Res., 103, 10419-10435.
Swift, J.H., E.P. Jones, K.Aagaard, E.C. Carmack, M.Hingston, R.W. MacDonald, F.A. McLaughlin and R.G. Perkin, 1998: Waters of the Makarov and Canada basins. Deep Sea Res, 44, 1503-1529.
Zhang, J., D.A. Rothrock and M.Steele, 1998: Warming of the Arctic Ocean by a strengthened Atlantic inflow: Model results. Geophys. Res. Lett., 25, 1745-1748.
Figure 1:A photograph of the surface of the summertime Arctic ice cover taken during the SHEBA experiment. Visible are surface meltwater ponds, drainage channels, a lead with individual ice floes, and crumpled snow which indicates ridged ice cover. Photo courtesy of Dr. Don Perovich.
Figure 2: The average ice thickness in (a) winter and (b) summer based on upward looking sonar observations (Bourke and Garret, 1987).
Figure 3: The simulated surface air temperature difference of a 2XCO2 simulation minus a 1XCO2 simulation. The warming is amplified at high latitudes where it reaches a maximum of 4.5 degrees C.
Figure 4: Mean ice draft difference, 1993-1997 conditions minus 1958-1976 conditions, obtained from upward looking sonar observations from submarine tracks. This figure is taken from Rothrock et al., 1999.
Figure 5: The timeseries of Arctic ice extent for the Winter and Summer. This figure is taken from Deser et al., (2000).
Figure 6: The timeseries of the North Atlantic Oscillation Index defined as the difference of normalized pressures between Lisbon, Portugal and Stykkisholmur, Iceland. This figure was taken from Hurrell, 1995.
Figure 7: This shows the simulated and observed Fram Strait ice area flux anomaly. Hilmer and Jung (2000) found that this timeseries was correlated to the NAO at 0.06 from 1958-1977 and 0.70 from 1978-1997.
Figure 8: The (a) temperature and (b) salinity difference of a 1993 SCICEX cruise minus the climatology. This shows the anomalous warmth and salinity of the Atlantic layer within the Arctic basin. This figure was taken from Morison et al., 1998.
Figure 9: The sea ice/albedo feedback mechanism.
Figure 10: The fraction of the variance in Arctic (a) ice area and (b) ice volume that can be attributed to the albedo feedback mechanism. The different lines denote simulations which use different sea ice physics.
Figure 11: The change in globally averaged air temperature over 500 years of integration for increasing atmospheric CO2 simulations. The CO2 increases at 1% per year until it stabilizes at 2XCO2 at year 70 of the simulation. The black line is for a case when all feedbacks are active. The red line is for a case in which the albedo feedback mechanism is inactive.
Figure 12: This schematic of the North Atlantic thermohaline circulation shows the surface currents in red. These are converted to deep water within the Arctic basin, the GIN Seas, and the Labrador Sea. The deep currents, shown in blue, return south as North Atlantic Deep Water (NADW).
Figure 13: The (a) timeseries and (b) spectrum of the simulated overturning index. In panel (b), the red line denotes the theoretical red spectrum and the dashed line denotes the 95% significance level.
Figure 14: The first EOF of (a) sea surface temperature (SST) and (b) sea surface salinity (SSS) from the model simulations. Panel (c) shows the correlation of the EOF timeseries with the overturning index. A negative lag indicates that the SST and SSS conditions lead the overturning timeseries
Figure 15: A regression of the ocean density on the overturning index. The black line is the total density changes due to both temperature and salinity variations, the red line is the density variations due to salinity changes, and the blue line is the density variations due to temperature changes.
Figure 16: The timeseries of the simulated anomalous ice export.
Figure 17: The correlation of the ice export and overturning timeseries. A negative lag indicates that the export leads the overturning.
Figure 18: The timeseries of the simulated (a) overturning index and (b) Fram Strait ice export. The first 1000 years of the integration are run with present day forcing. The CO2 is then allowed to increase at 1% per year, until it reaches doubling at year 1070. It then remains at 2XCO2 values for the remainder of the integration.