Introduction
The present study shows the typical behaviour of a major sudden stratospheric warming (SSW) and its effects
on the atmospheric composition, thermal state, and dynamics. The SSW phenomenon
has already been investigated and reviewed in the past, e.g. by
, , and . In short, planetary wave-breaking
in the middle atmosphere leads to a reversal of the polar stratospheric
vortex in the winter hemisphere. SSWs are generally important for the
meridional circulation of the middle atmosphere. used a
high resolution numerical weather prediction model with a maximum of 84 km
altitude. They found that major SSWs can be predicted with a lead time of 12 days.
studied the response of the atmospheric composition to
a SSW and found ozone depletion and polar stratospheric clouds over Europe.
performed an ozone budget analysis and found that ozone inside
the polar vortex is enhanced by 26–28 DU due to the SSW.
reported that SSWs cause anomalous weather regimes 60 days after the SSW in
the troposphere.
In the present study, we follow the conservative definition of a SSW given by
. A major SSW is associated with a sudden warming (>20 K) of the stratosphere poleward of 60∘ N and at 10 hPa. Further, the
zonal wind reverses from eastward to westward flow for at least 5 days. The
date of the zonal wind reversal (u=0) is taken as the central date of the
SSW. The central date is ideally suited to be the timing mark of the
composite analysis. Different to previous studies, the composite analysis
is taken here for a longer time interval before and after the SSW (±90
days for the dynamic parameters and ±180 days for total ozone with
respect to the central date of the SSW). The long time interval is reasonable
since the life time of ozone in the lower stratosphere is several months. A
case study by showed that the stratopause is elevated
after a major SSW over a time interval of 90 days. In addition, one can argue
that the vortex onset is usually in November for the Northern Hemisphere,
while the SSW most often occurs in late January. The onset of the polar
vortex might already be related to the preconditioning of the stratospheric
flow.
simulated the effect of preconditioning on the occurrence of
major and minor SSWs. She found that an upward propagating, planetary wave
pulse can initialize a SSW when the lower stratospheric flow is
preconditioned. The area size of the polar vortex and the planetary wave
refractive index can indicate the preconditioning. The simulations of
were limited to the discussion of dynamical parameters and to
a time interval from 20 days before to 20 days after the SSW central date.
Evaluating meteorological reanalysis data with a composite analysis from 10 days
before to 10 days after the SSW central date, found
that a tropospheric blocking can initialize a SSW. However,
found no significant link between tropospheric blockings and SSWs.
investigated the life cycle of SSWs by using a composite
analysis of meteorological reanalysis data with epoch times from 40 days
before to 40 days after the SSW central date. The EP flux divergence reaches
an extrema in the stratosphere during the so-called SSW growth phase (from 22 to 8 days
before SSW central date). performed a composite
analysis (from 60 days before to 60 days after the SSW central date) based on
different meteorological reanalysis data sets. In difference to the present
study, they averaged out the high-latitude region, with the advantage of
studying the altitude dependence of SSWs. They found that the heat flux at 100 hPa
is increased during the 20 days before the SSW.
performed composite analysis for the time interval from 15 days before to 15 days
after the SSW central date. They looked particularly for mesospheric
precursors in the zonal wind, and they found longitudinal dependence of the
mesospheric flow before the central date of the SSW.
performed a study on the 3-D shape of the dynamical polar
vortex from 10 days before to 10 days after the SSW central date. As in other
studies, they distinguished between vortex-splitting SSWs and
vortex-displacement SSW so that they can discuss the longitudinal and
latitudinal variations of the SSW events in polar stereographic composite
plots. In the present study, no separation of major SSWs into
vortex-splitting and vortex-displacement events is performed since we focus
on the discussion of zonal means, and the effects of vortex-splitting and
vortex-displacement partly vanish in the zonal mean. Further, the separation
of the 20 major SSWs into two ensembles has the penalty that the number of
SSW events in each ensemble is reduced (only about 10 instead of 20). Thus
the standard deviation of the mean would become larger and it might be more
difficult to derive statistically significant results. Further, the number of
viewgraphs would be doubled which is also not desired for the present study.
In the following we describe the evolution of a major SSW by means of
composite analysis. As mentioned above, the time interval of the epoch time
is longer for the present study compared to previous studies. We also discuss
the behaviour of the parameter total ozone which was not covered by previous
composite analysis studies. The aim is to find SSW-related features in the
composites which might have been overseen by others. The detection of such
SSW-related features may enhance our understanding of the SSW phenomenon and
may lead to a better prediction of SSWs by the recognition of SSW precursors.
Data set and data analysis
Meteorological reanalysis data: ERA Interim
ERA Interim reanalysis provides vertical profiles of ozone volume mixing
ratio with a time resolution of 6 h from 1979 to present. While the
absolute values of the ozone profiles may have uncertainties of about 10 to
20 %, the spatial and temporal variations of the estimated ozone
distribution on larger scales (say horizontal scales >1000 km and temporal
scales > 5 days) are reliable since the ozone distribution of ERA Interim
is driven by observations from ozonesondes and satellites and by the dynamics
of the ECMWF global circulation model . The upper boundary of
the assimilation model at about 65 km (0.1 hPa) may lead to uncertainties in
the upper stratospheric circulation and ozone transport. However, ERA Interim
can be regarded as state-of-the-art meteorological reanalysis; the present study can take advantage of the numerous past improvements in
atmospheric modelling and observation techniques.
Climatology
In the present study, we discuss anomalies of the zonal mean data of ERA
Interim. The mean seasonal behaviour is derived from the 36 years of ERA
Interim data (1979 to 2015). The data are sorted and averaged for the day of
year. Then the mean seasonal series is obtained by means of a 60-day low pass
filter (digital non-recursive filter with a fast response time). An example
of the mean seasonal behaviour of TOC (total ozone column) at 80∘ N is shown by the red
line in Fig. .
Composite analysis and significance
We consider all major SSWs that have taken place in the Northern Hemisphere from 1980
to 2014. The SSW of 1979 is not taken since it lays at the edge of the
available ERA Interim data set. The timing marks of the 20 major SSWs are
given in Table . A timing mark corresponds to the central date of a
major SSW. This central date is the time when the zonal wind at 10 hPa and
poleward of 60∘ N reverses from eastward wind to westward wind. In
the present study we derive the central dates directly from the time series
of zonal wind and temperature from the ERA Interim data set. For each latitude
belt, we compute the anomalies of the different parameters at various
pressure levels. Anomaly data segments from 180 days before the central date
to 180 days after the central date are added and averaged so that the mean
behaviour of the atmosphere with respect to the 20 major SSWs is obtained at
each latitude belt and each pressure level.
Central dates of the 20 major SSWs
28 February 1980
25 February 1999
03 April 1982
01 February 2001
20 February 1984
28 December 2001
29 December 1984
15 February 2002
21 January 1987
16 January 2003
06 December 1987
02 January 2004
12 March 1988
21 January 2006
19 February 1989
23 January 2009
19 March 1992
23 January 2010
14 December 1998
06 January 2013
Time series of zonal mean TOC at 80∘ N. The blue line is
provided by ERA Interim reanalysis. The red line shows the mean seasonal
behaviour of TOC as derived from ERA Interim reanalysis data at 80∘ N
from 1979 to 2014. The vertical, magenta lines denote the central dates of
the major SSWs in 2006, 2009 and 2010. Please note the positive anomalies (>50 DU) of the blue line relative to the red line (climatology) after the
onsets of the major SSW events. Xticks are always on 1 January.
Composite of the anomaly ΔTOC as function of epoch time
(0 is the central date of SSW) and geographic latitude. Values are only shown if
they exceed the 2σ level where σ is the error of the mean of
the SSW ensemble. After the SSW, increased Δ TOC values of up to 90 DU
occur in the polar region. A negative anomaly of about -20 DU is present in
polar ozone 90 days before the SSW. The white contour lines have a spacing of
10 DU. The composite is based on 20 major SSWs.
For the ensemble of the 20 SSWs, we also derive the standard deviation of the
mean σ (error of the mean) as the function of epoch time and latitude for
each anomaly distribution. The mean anomalies which exceed the
2σ level are regarded to be statistically significant with a
confidence greater than 95 %. In the following, the viewgraphs and the
discussion are only performed for the mean anomalies exceeding the
2σ level.
Results
Composites of total ozone column
TOC density is derived by integration of the ozone
concentration over the height interval from 5 to 60 km. Figure shows
an example for zonal mean TOC at 80N latitude from 2005 to 2010. Three major
SSWs (January 2006, 2009 and 2010) happened during this time interval.
These SSWs are marked by the magenta lines and it is obvious that TOC (blue
line) is enhanced by about 50 DU with respect to the TOC climatology (red
line) after the onset of the SSWs.
Quite a similar behaviour is obtained at high latitudes for the composite of
the TOC anomaly of the 20 major SSWs. Figure shows the composite of
the anomaly ΔTOC as function of epoch time (0 = central date of SSW) and
as function of geographic latitude. After the SSW, increased ΔTOC
values of up to 90 DU occur in the polar region. A negative anomaly of about
-20 DU is present in polar ozone 90 days before the SSW. It is remarkable
that the positive anomaly exists for more than half a year. This is due to
the long life time of odd oxygen in the lower stratosphere and to the small
meridional mixing of the polar air masses. A simulation of for
the SSW in Arctic winter 2002–2003 gave the result that horizontal advection
of ozone into the polar vortex increases significantly due to the weakening
of the vortex by planetary waves. They found a pronounced increase (26–28 DU)
in the polar vortex ozone due to SSW events. Figure shows total
ozone derived from ozonesondes launched at Sodankylä (67∘ N) before and after
the major SSW of February 2001. High TOC values of about 450 DU are measured
in the months after the SSW.
Composites of the other parameters
Composites of the anomalies ΔT at 10 and at 1 hPa are shown in
Fig. . A temperature maximum of 27 K is reached at 10 hPa in the
polar region. Interestingly, the air temperature is enhanced in the 2
months before the SSW at 1 hPa. The cooling after the SSW at 1 hPa is
possibly associated with the cold layer below the elevated stratopause as
reported by . The elevated stratopause also remained for up
to 3 months after the SSW in their study. find that 71 %
of the major SSWs are accompanied by an upper mesospheric cooling. Composites
of the anomalies ΔPV (potential vorticity) at 10 and at 1 hPa are shown in the lower
panel of Fig. . In the months before the SSW, ΔPV is 2
times increased in the polar region at 10 hPa. After the major SSW,
ΔPV decreases, and this decrease spreads out from the polar to
mid-latitudes in the following 3 months. The weakening or breakdown of the
polar vortex after the SSW is of course reported in previous studies but it
is interesting that the circulation is perturbed over 3 months. At 1 hPa,
ΔPV at mid-latitudes is decreased 30 days before the SSW and this
negative anomaly propagates in the following time to high latitudes.
Total ozone derived from ozonesondes (blue line) launched at
Sodankyla (67∘ N). High TOC values (>450 DU) are reached after the
SSW of February 2001 (red line), similar as in the composite of Fig. . The black line denotes the long-term mean at Sodankyla as derived
from the ERA Interim reanalysis (1979–2014).
Composites of the anomalies ΔT at 10 hPa (upper left) and at
1 hPa (upper right). Values are only shown if they exceed the
2σ level. Temperature maximum of 27 K is reached at 10 hPa. Composites
of the anomalies ΔPV at 10 and at 1 hPa are shown in the lower
panel. The composites are based on 20 major SSWs.
Composites of the anomalies Δu at 10 hPa (upper left) and at
1 hPa (upper right). Values are only shown if they exceed the
2σ level. Eastward wind anomaly of -59 m s-1 is reached at 1 hPa.
Composites of the anomalies Δv at 10 and at 1 hPa are shown in the
lower panel.
Composites of the anomalies Δu at 10 and at 1 hPa are shown in
Fig. . A negative anomaly of eastward wind of -59 m s-1 is reached
at 1 hPa, connected with the breakdown and the reversal of the vortex after
the SSW. The negative values of Δu 30 days before the
SSW at mid-latitudes at 1 hPa are remarkable. It seems that the anomaly started at the
mid-latitude stratopause and propagated poleward until the SSW onset. At 10 hPa, Δu at mid- and high latitudes remains negative (westward anomaly)
up to 3 months after the SSW. In some of the previous composite studies
(e.g. , the zonal wind switches back to eastward direction 5
to 10 days after the SSW. This is certainly true if the composite is taken
for the absolute zonal wind values (and not for the anomalies with respect to
the climatological mean). The composite analysis of anomalies seems to foster
the recognition of precursors and the effects of SSWs. It should be noted
that several previous studies also discussed composites of anomalies.
However, the derivation of the anomalies was performed by means of leading
principal component (PC) time series analysis (e.g. )
which actually cannot fetch the climatological mean as the reference for the
SSW anomaly.
Composites of the anomalies Δv at 10 and at 1 hPa are shown in the
lower panel of Fig. . It is evident that a southward wind anomaly
(blue) prevailed in the polar region before the SSW. This is in agreement
with the enhanced meridional heat flux anomaly before the SSW, as reported by
.
Composites of the anomalies Δw (positive = downward) at 10 hPa
(upper left) and at 1 hPa (upper right). Values are only shown if they exceed
the 2σ level. Composites of the anomalies Δz at 10 hPa and at 1
hPa are shown in the lower panel. A maximal, upward shift of 2900 m is reached
for the 1 hPa geopotential height (lower right).
Composites of the anomalies Δw (positive = downward) at 10 hPa and at 1
hPa are shown in Fig. . In the polar region there is upwelling
(blue) before the SSW and downwelling (red) after the SSW. However, it
is surprising that the downwelling in the polar region starts 30 days after the
SSW. At mid-latitudes, the behaviour is vice versa, with downwelling (red)
before and during the SSW. Composites of the geopotential height anomalies
Δz at 10 and at 1 hPa are shown in the lower panel of Fig. . At 1 hPa, at mid-latitudes, the geopotential height (or pressure)
increases 30 days before the SSW. Similar to the zonal wind, the geopotential
height perturbation travels poleward. In addition, at 10 hPa and about 10 days
before the SSW, there is an increase in geopotential height. A maximal,
upward shift of 2900 m is reached for the 1 hPa geopotential height. The
pressure increase remains for 60 to 90 days after the SSW. The increase of
the geopotential height anomaly before, during, and after the SSW at 50 hPa is
shown in .
In summary, there are several precursors of the SSW, visible in the
stratopause region (u, z) and in the mid-stratosphere (PV, v). The
stratosphere seems to be most promising for prediction of major SSWs at mid-latitudes. Forecast of an individual SSW remains difficult since the
variability of the winter stratosphere might be larger than the precursor
signals.
Conclusions
The study showed that it is valuable to take climatology as the reference
for the SSW anomaly and to analyse the composites on a longer interval of
epoch time (e.g. from 3 months before, to 3 months after the SSW central
date). We quantitatively described the average behaviour of a major SSW in
the anomalies of total ozone column density (TOC), temperature (T),
potential vorticity (PV), eastward wind (u), northward wind (v), vertical
wind (w), and geopotential height (z). Interestingly, some parameters (T,u,z) already show an anomaly 1–2 months before the major SSW onset. For
example, the geopotential height of the 1 hPa level at mid-latitudes increases
more than 1 month before the major SSW. The meridional propagation of the
initial SSW anomaly from mid-latitudes to high latitudes is also seen in
Δu and might be based on the connection between Rossby waves and
SSWs. The significant features in the anomaly viewgraphs may help to further
understand the life cycle of SSWs. There might be a small chance for
long-term prediction of a major SSW if all characteristics are taken into
account, for example, the probability of a SSW could be calculated 1 or 2 months ahead. The anomaly of total ozone (Fig. ) showed the
long-lasting effect (> 6 months) of a major SSW on the atmospheric
composition in the polar stratosphere nicely.