University of Bergen, Geophysical Department, Bergen, Norway
The Atlantic is the most convective of the world oceans and plays a key role in the thermohaline circulation in which warm and saline surface waters flows poleward, cools, sinks to great depth and returns equatorward (e.g. Marshall, 1999). The high latitude open sea areas to the east and west of Greenland are the prime convective sites of the North-Atlantic. The convective activity of these sites is subject to substantial interannual and decadal variability which is likely to be associated with atmospheric variability (Dickson et al., 1996). The variability of the atmospheric forcing of the Nordic Seas through surface sensible heat loss is the topic of the following. The results presented herein documents that the variability in the sensible heat loss from the Nordic Seas is closely related to the variability of the meridional atmospheric flow component. The largest heat loss appears when anomalous northerly winds advects cold and dry air masses from the Arctic ice sheet out over the Nordic Seas. On a synoptic timescale such off-ice flow events are often referred to as Cold Air Outbreaks (CAOs). Here, the CAO variability will be adressed. The analysis is based on NCEP/NCAR reanalysed data (Kalnay et al. 1997) and anomalies are identified by removing the long term mean of each calendar month.
2. Leading extratropical modes of variability
Recently, a deep atmospheric mode of variability termed the "Arctic Oscillation" (AO) has been identified (Thompson and Wallace, 1998, hereafter TW98). The AO can be identified as the leading Empirical Orthogonal Function (EOF) of sea level pressure anomalies in the Northern Hemisphere extratropics in both monthly and seasonal data. EOF analyses identify recurring anomalous structures in datasets so that much of the variability of the data can be represented by a few EOFs. The principal component (PC) time series of the leading EOF may be regarded as the AO index. In the high index state (TW98 sign convention), sea level pressure (SLP) is low over the Arctic and high over the surrounding zonal ring, inducing a strong westerly geostrophic flow. By regressing geopotential height anomalies at pressure surfaces in the troposphere and stratosphere onto the AO index, TW98 demonstrated that the AO is a deep almost equivalent barotropic mode of variability, implying that a strong polar night jet in the lower stratosphere coincide with a strong zonal circulation in the troposphere. In a follow up paper, Thompson and Wallace (1999a) went on to demonstrate that the two leading modes of extratropical variability are very similar in the northern and southern hemisphere and they were referred to as "annular", emphasizing their zonally symmetric structures. Thompson and Wallace (1999b) focused on trends and the AO trend from the 1960s to the 1990s was shown to account for about 50% of the winter increase in surface air temperatures (SATs) over the area 40-70°N, 0-140°E in Eurasia. This may be regarded as an implication of that the increase in atmospheric CO2 may affect the probability distribution of natural atmospheric modes of variability, forcing them into one particular phase. The Eurasian heating was interpreted as a result of stronger westerlies, advecting warm and humid marine airmasses across the western seaboard of the Eurasian continent. Shindell et al. (1999) reports that only GCMs with sufficient vertical resolution in the stratosphere responds to a CO2 increase with an increase in the AO index. Baldwin and Dunkerton (1999) shows by analysing daily data that the AO in the "active season" (November through March; when the long planetary waves are allowed to propagate vertically into the stratosphere), seems to propagate downwards from the stratosphere to the troposphere in a timescale of weeks.
3. The Barents Sea Oscillation and its signatures in SLP, SAT and sensible heat loss
The striking resemblance between EOF 2 of extratropical northern hemisphere winter SLP anomalies and SLP composites based on low and high Nordic Seas heat loss months after the removal of the AO related variability has recently been pointed out (Skeie, submitted 1999). An EOF 2-like pattern is also found in SLP composites based on warm and cold Eurasian months, again after the removal of the AO variability. This indicates that EOF 2, referred to as the "Barents Sea Oscillation", is a recurring anomalous SLP pattern with strong influence on the heat loss of the Nordic Seas and on Eurasian SATs. In Figure 1, the AO and BO indices for December through March 1958-1999 are shown, along with their respective regression patterns in SLP and SAT.
The AO has already been commented and we limit the discussion to the BO index and its regression patterns. The BO pattern is comprised by three main SLP centers of action with the main one in the southeastern Barents Sea, a node aligned more or less with the Greenwich meridian, a centre action of opposite sign to the west of this node and with a centre of action in the eastern Pacific of same sign as the Barents Sea centre of action. The regression patterns shown is the pattern associated with one positive standard deviation away from zero of the normalized BO index, hence, it may be regarded as a typical anomaly. Note that the choice of sign convention is such that a positive BO index is associated with a positive SLP perturbation in the Barents Sea. This means that negative values of the BO index, in a gestrophical sence, are associated with anomalously northerly winds over the Nordic Seas. The SAT regression pattern of the BO has a dipole structure between the European Arctic and Eurasia with amplitudes of 3.5 and 2.5 K respectively. The time series of the BO has no apparent trends. However, it seems to fluctuate more strongly when the SLP is high over the Arctic, like in the 1960s according to the AO index. Strong northerly winds associated with the BO, occurred during the March months of 1961 and 1968. The latter falls together with the onset of the "Great Salinity Anomaly" (Dickson et al., 1988). The statistical significance of such extreme events is low however, and not conclusive. Normalised time series and averaging areas of Nordic Seas sensible heat loss and Eurasian SATs are shown in Figure 2.
Figure 1: Upper: The AO and BO indices for December-March 1958-1999. The thin red line indicates monthly data, the thicker blue is a 12 months (three winters) running mean. The seasons are indicated by vertical bars and December data are drawn on the boarder between green and white. Middle and lower: SLP and SAT regression patterns associated with one positive standard deviation of the AO and BO indices. Positive and negative contours are red and blue, respectively. The zero contour is yellow. Contour spacing is 1 hPa for SLP and 0.5 K for SAT.
Only the months Dec.-Mar. are considered and with 167 months (165 degrees of freedom), the 99% confidence level of correlations is 0.2. The heat loss and SAT time series is in the following referred to as the heat loss and SAT indices. The BO index correlates to the heat loss and SAT indices with r=0.76 and r=0.56, respectively. The corresponding correlations for the AO index is r=0.23 and 0.63. The latter was reported by TW98 as r=0.65 for the same area, but with a different SAT dataset. The time series formed by adding the AO and BO indices correlates to the SAT index with r=0.84, hence the AO and BO explains 70% of the variance of the SAT index suggesting that the bulk of the SAT variance over Eurasia is due to anomalous atmospheric circulation patterns. Figure 3 shows the long term mean SLP for January perturbed by the AO and BO regression patterns shown in Figure 1.
Figure 2: The normalised areally averaged sensible heat flux and surface air temperature anomaly time series for the Nordic Seas (blue curve and area) and Eurasia (red curve and area).
Figure 3: Long term mean SLP for January perturbed by the AO and BO regression patterns shown in Figure 1. The contour spacing is 8 hPa.
It is apparent that both AO and BO have a strong influence on the mean geostrophic flow across the western seashore of the Eurasian continent.
3. The authenticity of the BO
The time series of 167 months is too short for the BO EOF to be well separated from the subsequent EOFs according to the sampling error estimate of North et al. (1982). Hence, it is important to examine whether it may be found using other techniques as well. Figures 4 and 5 shows SLP composites based on months where the heat loss and SAT indices are more than one standard deviation off from zero.
Figure 4: SLP composites based on high and low sensible heat loss months. The line colors follows the convention introduced in Figure 1. Contour spacing is 1 hPa in the upper and 2 hPa in the lower panels. In the low minus high composite, the mean SLP of the low and high heat loss composites differs at the 99% confidence level according to a Students t-test inside the black dotted line.
The high heat loss SLP composite is formed by averaging the SLP over all the high heat loss months and so forth. Since an EOF must describe two phases, difference composites are constructed and by removing the AO related SLP variability, BO-like patterns arise, lending credence to the authenticity of EOF 2. The true EOFs of climate cannot be known, hence we must be satisfied with estimates of them based on a set of independent realisations (North et al., 1982). If EOF 2 is an authentic EOF of climate, one may speculate on the physics behind it. However, much more can be still be done at examining its authenticity, for instance by seeking signatures of it in higher layers of the atmosphere, or by examining a longer SLP dataset.
Figure 5: SLP composites based on warm and cold Eurasian months. The line colors follows the convention introduced in Figure 1. Contour spacing is 1 hPa in the upper and 2 hPa in the lower panels. In the cold minus warm composite, the mean SLP of the cold and warm composites differs at the 99% confidence level according to a Students t-test inside the black dotted line.
The link between BO variability and the sensible heat loss of the Nordic Seas as suggested, both by the highly significant temporal correlations, and by the heat loss SLP composites, implies that the BO or, at least the meridional atmospheric flow variability is important to diagnose and validate in coupled climate models. It is also interesting to examine whether the BO is reproduced in climate models. The ice export variability through the Fram Strait in winter has been shown to be closely related to variability in the atmospheric forcing (Vinje et al., 1998). It is interesting to note that while the AO SLP gradient across the Fram Strait is close to zero, the BO gradient associated with one standard deviation away from zero in the BO index is roughly 2 hPa.
Baldwin, M.P and Dunkerton, T.J., 1999, Propagation of the Arctic Oscillation from the stratosphere to the troposphere, J. Geo. Res. in press.
Dickson, R., Lazier, J., Meinke, J., Rhines, P. and Swift, J., 1996, Long-term coordinated changes in the convective activity of the North Atlantic, Prog. Oceanogr., 38, 241-295.
Dickson, R.R., Meincke, J., Malmberg, S.A. and Lee, A.J., 1988, The 'Great Salinity Anomaly' in the Northern North Atlantic 1968-1982, Prog. Oceanog., 20, 103-151.
Kalnay, E. and co-authors, 1996, The NCEP/NCAR 40-year reanalysis project, Bull. Amer. Met. Soc., 77, 437-471.
Marshall, J. and Schott, F., 1999 Open-ocean convection: Observations, theory, and models, Rev. Geophys., 37,1-64.
North, G.R., Bell, T.L., Cahalan, R.F. and Moeng, F.J., 1982, Sampling errors in the estimation of empirical orthogonal functions, Mon. Wea. Rev., 110, 699-706.
Shindell, D.T., Miller, R.L., Schmidt, G.A. and Pandolfo, L., 1999: Simulation of recent northern winter climate trends by greenhouse-gas forcing, Nature, 399, 452-455.
Skeie, P., 1999: The coupling between Cold Air Outbreaks over the Nordic Seas and Eurasian warm months. Geophys. Res. Lett., Submitted. (Postscript preprint).
Thompson, D.W.J, and Wallace, J.M., 1998: The Arctic Oscillation signature in the wintertime geopotential height and temperature fields. Geophys. Res. Lett., 25, 1297-1300.
Thompson, D.W.J. and Wallace, J.M., 1999: Annular modes in the extratropical circulation. Part I: Month-to-month variability. J. Climate, in press.
Thompson, D.W.J. and Wallace, J.M., 1999: Annular modes in the extratropical circulation. Part II: Trends. J. Climate, in press.
Vinje, T., Nordlund, N. and Kvambekk, A., 1998, Monitoring ice thickness in Fram Strait, J. Geo. Res., 103, C5, 10,437-10,449.