Interactive comment on “ The Arctic vortex in March 2011 : a dynamical perspective ”

This paper aims to examine dynamical behavior associated with the low Arctic ozone during March 2011. Results show that the low ozone was linked with cold vortex temperatures, low heat flux (dynamical forcing), and a delayed break-up of the vortex in springtime, and very similar behavior was observed for the anomalous low ozone conditions during March 1997. The authors examine systematic effects of ENSO and the QBO, and show that neither of these factors contributes significantly to the March 2011 behavior. However, significant effects are suggested to occur due to anomalous sea surface temperature (SST) in the North Pacific ocean, and the latitude-height pattern of composited temperatures for this mode is similar to observations during March 2011 (and also March 1997, when SST’s were also high).


Introduction
In the Arctic stratosphere, chemical ozone loss takes place each year in the late winter (WMO, 2011).Arctic ozone loss represents the interaction between chemistry and climate: heterogeneous ozone depletion on polar stratospheric clouds (PSCs) requires the presence of halogens, sunlight and temperatures below approximately 195 K. Rex et al. (2004 and2006) calculated that the severity of large ozone loss events has been increasing over the last few decades, and speculated that increased radiative cooling by greenhouse gases plays a role.
Severe ozone loss was observed in the Arctic stratosphere in 2011.On 14 March, the Alfred Wegener Institute in Germany reported that "unusually low temperatures in the Arctic ozone layer have recently initiated massive ozone de-Correspondence to: M. M. Hurwitz (margaret.m.hurwitz@nasa.gov)pletion" (http://www.awi.de/en/news/pressreleases).Manney et al. (2011) have since determined that the spring 2011 ozone loss was "unprecedented": During the 2010-2011 winter, the Arctic vortex was the most isolated and the timeintegrated PSC volume was the largest ever observed.These conditions enabled severe ozone loss in late winter.Vortex averaged lower stratospheric ozone was unusually low beginning in late February, and by March, reached values comparable to those recently observed in the Antarctic stratosphere in September.Figure 1a shows that March total ozone in the 60-80 • N region was the lowest of the satellite era (total ozone dataset updated from Stolarski and Frith, 2006).
Circulation patterns in the North Pacific sector have been linked to anomalous Arctic lower stratospheric conditions in winter.Orsolini et al. (2009) found that the largest observed Arctic PSC volumes were on average preceded by a weakening of the climatological, tropospheric low in the subpolar Far East and North Pacific regions.This type of "blocking" event in the North Pacific, also known as the positive phase of the Western Pacific teleconnection pattern (WP; Wallace and Gutzler, 1981), effectively weakens the upward propagation of upper tropospheric planetary waves (see e.g., Woollings et al., 2010).
In addition, two tropical phenomena contribute to interannual variability in the Arctic stratosphere in winter: El Niño/Southern Oscillation (ENSO) and the quasibiennial oscillation (QBO).Holton and Tan (1980) and Lu et al. (2008) showed that the phase of the QBO modulates the region in which planetary waves can propagate in the stratosphere, thus affecting the strength of the Arctic vortex in mid-winter.The vortex tends to be stronger during the westerly phase of the QBO than during the easterly phase.Similarly, planetary wave driving is relatively stronger during El Niño (ENSO warm phase) events than during La Niña (ENSO cold phase) events (e.g., Garfinkel and Hartmann, 2008).While the Arctic vortex was relatively strong throughout the winter, Manney et al. (2011) showed that the exceptional nature of 2010-2011 was not apparent until late February and March.Thus, the goals of this paper are to: (1) document the dynamical conditions that made possible the record ozone loss in March 2011; and (2) attribute these conditions to known sources of dynamical variability.Section 2 will describe the datasets and diagnostics used to perform this anal-ysis.In Sect.3, March 2011 will be examined in the context of the satellite era.The qualitative relationship of the March 2011 conditions in the Arctic stratosphere to ENSO and the phase of the QBO will be considered.In addition, the possible role of North Pacific sea surface temperature variability in the anomalous dynamical conditions in the Arctic vortex in March 2011 will be examined.The results are further discussed in Sect. 4. Section 5 provides a brief summary.429

Data and diagnostics
Sea surface temperature (SST) and atmospheric diagnostics are used to understand conditions in the Arctic stratosphere in March 2011.The present analysis spans the satellite era  and focuses on the Northern Hemisphere midto late winter (January through March).Zonal winds, temperature and eddy heat flux fields are derived from the Modern Era Retrospective-Analysis for Research and Applications (MERRA) reanalysis.The MERRA reanalysis is based on an extensive set of satellite observations and on the Goddard Earth Observing System Data Analysis System, Version 5 (GEOS-5) (Bosilovich et al., 2008;Rienecker et al., 2011).The MERRA reanalysis has vertical coverage up to 0.1 hPa, and for this study, is interpolated to 1.25 • ×1.25 • horizontal resolution.
The phase of the QBO is characterized by zonal winds in the equatorial region at 50 hPa.Monthly mean values of the 50-hPa QBO index (http://www.cpc.ncep.noaa.gov/data/indices/qbo.u50.index) are used in this study.
The springtime breakup of the Arctic vortex is calculated for each year.On the 450 K isentropic surface (i.e., in the lower stratosphere), the breakup date is defined as the date when the five-day running mean of zonal winds at the vortex edge falls below approximately 15.2 m s −1 , following the criteria of Nash et al. (1996).The present analysis considers breakup dates based on the NCEP-1 (Kalnay et al., 1996), NCEP-2 (Kanamitsu et al., 2002) and NOAA Climate Prediction Center (CPC) (Gelman et al., 1986;Nagatani et al., 1988;Finger et al., 1993) meteorological reanalyses.
Monthly mean SST fields are taken from the Hadley Centre Global Sea Ice and Sea Surface Temperature (HadISST1) dataset (Rayner et al., 2003).Sea surface temperature anomalies in the eastern equatorial Pacific are characterized by the Niño 3.4 index (see http://www.cpc.noaa.gov/data/indices).Trenberth (1997) defines a conventional El Niño event as a sustained period (usually six months or more) when the Niño 3.4 index exceeds 0.4, while a La Niña event is defined as a sustained period when the Niño 3.4 index is less than −0.4.

March 2011 in a historical context
In 2011, the Arctic vortex was colder, stronger and more persistent than usual.Figure 1 shows histograms of Arctic total ozone, polar cap temperature, breakup date of the Arctic vortex, ENSO index, QBO index and North Pacific SST index in 2011 with respect to the 1979-2011 period.A histogram of March mean temperatures for the Arctic polar cap at 50 hPa is shown in Fig. 1b.The March 2011 temperature of 207.3 K (indicated by the red outline) is more than two standard deviations lower than the climatological mean value (216.4K) and is the second-lowest value in the 1979-2011 period.The lowest value (204.9K, indicated by the blue outline) occurred in 1997.
The breakup of the Arctic vortex occurs in spring.A histogram of breakup dates at 450 K is shown in Fig. 1c.The breakup date in 2011 was 19 April in the NCEP-2 reanalysis, later than the mean date of 20 March in the NCEP reanalyses and 10 April in the CPC reanalysis.The breakup date in 2011 was, depending on the zonal wind dataset, either the third or fourth latest of the satellite era.The late breakup of the Arctic vortex is consistent with the low temperatures and total ozone observed in March 2011 (see Fig. 1a and b).
Unusually cold conditions in the Arctic stratosphere in March 2011 correspond with unusually weak planetary wave driving in February 2011.Newman et al. (2001) found that polar lower stratospheric temperature is correlated with midlatitude eddy heat flux at 100 hPa, with a 1-2 month lag; this finding suggests that weaker than usual eddy heat flux in February should correspond with a colder than usual Arctic lower stratosphere in March.March temperature anomalies in 2011 and 1997 are shown in Fig. 3a and b.In both 1997 and 2011, the Arctic stratosphere cooled strongly while the mid-latitudes and Arctic troposphere warmed weakly.Consistent with these temperature differences, zonal winds were relatively stronger at high latitudes; peak wind differences exceeded 20 m s −1 at 10 hPa at high latitudes (not shown).The magnitude of the stratospheric cooling was larger in 1997 (e.g., the polar cap temperature at 50 hPa was 11.5 K lower than the climatological mean) than in 2011 (9.1 K lower).Similarly, February eddy heat flux was weaker in 1997 (7.0 K m s −1 less than the climatological mean of 14.8 K m s −1 at 40-80 • N, 100 hPa; see Fig. 2, left) than in 2011 (5.5 K m s −1 less).

Influence of ENSO and the QBO on the Arctic stratosphere in March
La Niña and QBO-westerly conditions persisted through March 2011.The Niño 3.4 index was strongly negative in January through March 2011, indicating La Niña conditions (Fig. 1d).In March 2011, equatorial zonal winds at 50 hPa were approximately 6 m s −1 (Fig. 1e), indicating the westerly phase of the QBO.This section compares the temperature anomalies observed in March 2011 with those observed during typical La Niña conditions and during the westerly phase of the QBO.The March temperature response to La Niña events is estimated by comparing years when the Niño 3.4 index is equal to or less than −1 (as in 2011) with an ENSO neutral composite (i.e., winters when the Niño 3.4 index is between −0.4 and 0.4). Figure 3c shows that, in the Arctic stratosphere, the typical March temperature response to a La Niña event is a weak warming.Thus, the La Niña response is inconsistent with the observed strong polar cooling in both 1997 and 2011.
The QBO was in its westerly phase during the 2010-2011 winter season (Fig. 1e).The March temperature response to the phase of the QBO is estimated by comparing composites of QBO-westerly years and QBO-easterly years.The typical March temperature response is a relative warming of the Arctic stratosphere that increases with altitude (Fig. 3d).As for the La Niña response, the temperature response to QBOwesterly conditions is inconsistent with the observed strong polar cooling in both 1997 and 2011.
In summary, the patterns and magnitudes of the March 2011 temperature differences from climatology are similar to those seen in March 1997, but different from the Arctic response to both La Niña events and to the phase of the QBO.March zonal wind and February eddy heat flux differences are consistent with these conclusions.That is, the weak eddy heat flux in February and low temperatures in March 2011 are not related to either ENSO or the QBO.Note that, because the Niño 3.4 and QBO indices are slowly-varying, the above findings do not depend on the winter month and/or season used to represent the ENSO and QBO phases.

Influence of North Pacific SSTs on the Arctic stratosphere in March
As noted in Section 3.  mura et al. (1997).The subarctic mode is associated with SST variability at decadal timescales, caused by variability in the Kuroshio and Oyashio currents, and is not influenced by variability in the tropical Pacific (i.e., variability related to ENSO).Furthermore, the subarctic SST mode is not related to the Pacific Decadal Oscillation (PDO) (index updated from Mantua et al., 1997;Zhang et al., 1997).This section considers the influence of North Pacific SSTs on the Arctic troposphere and stratosphere in March.
The subarctic SST index was strongly positive in both 1997 and 2011 (Fig. 1f).The positive phase of the subarctic SST mode tends to weaken the Aleutian low and thus the Pacific-North American (PNA) and Western Pacific (WP) teleconnection patterns.Both the PNA and WP indices were strongly negative in December and January during the 1996-1997 and 2010-2011 winters.Equally, winters when the subarctic SST index was most negative (e.g., 1987-1988) include months with strongly positive values of the PNA and WP indices.Previous work has linked weakening of the PNA and WP teleconnection patterns with stratospheric variability: Orsolini et al. (2009) and Garfinkel et al. (2010) found that variability in the Aleutian low modulates the strength of the Arctic vortex in mid-and late winter.Nishii et al. (2010) found that extreme positive WP events in early winter can lead to persistent stratospheric cold periods and high PSC volumes in later months.
On average, the positive phase of the subarctic SST mode is associated with a relative cooling of the Arctic stratosphere.The subarctic SST index and March polar cap temperature at 50 hPa are anti-correlated (with a correlation coefficient of −0.45; Fig. 2, right).Figure 3e shows the difference between March temperatures in years when the subarctic SST index is strongly positive as compared with years when the index is strongly negative: The Arctic stratosphere is relatively colder (by approximately 6 K at 50 hPa), while below 500 hPa the Arctic is approximately 2 K warmer.The structure and magnitude of these temperature differences are broadly consistent with the March temperature anomalies

Discussion
Recent cooling of the Arctic lower stratosphere has been reported by e.g., Randel et al. (2009) and Kennedy et al. (2010).Rex et al. (2004 and2006) noted a pattern of recent Arctic "cold winters getting colder".In the MERRA reanalysis in March, polar cap temperature at 50 hPa decreased 0.17±0.14K yr −1 between 1979 and 2011.During this pe-riod, cooling of the Arctic lower stratosphere can be largely attributed to increased radiative forcing by greenhouse gases and to ozone depletion (Shine et al., 2003;Stolarski et al., 2010).However, this modest linear trend in March does not explain the anomalous conditions in 1997 and 2011, when the Arctic lower stratosphere was approximately 10 K below the climatological mean.
Similarly, the phase of the 11-yr solar cycle does not account for the anomalous conditions in March 2011.The solar cycle can be characterized by the solar flux at 2800 MHz (ftp: //ftp.ngdc.noaa.gov/STP/SOLARDATA/SOLAR RADIO/ FLUX/Penticton Observed/monthly/MONTHLY.OBS); both 1997 and 2011 were within a few years of solar minima.Since the QBO was easterly in 1997 but westerly in 2011, the product of the solar cycle and QBO anomalies had the opposite sign in 1997 as compared with 2011.Though this quantity is well correlated with polar variability (Haigh and Roscoe, 2006), it does not explain the anomalously strong vortex events in both 1997 and 2011.
ENSO and the QBO do not explain the unusual dynamical conditions in March 2011.While La Niña conditions tend to strengthen the Arctic vortex in mid-winter, the La Niña signal weakens and begins to reverse by March.In Goddard Earth Observing System Chemistry-Climate Model, Version 2 (GEOS V2 CCM) simulations (model formulation as described by Hurwitz et al., 2011; La Niña simulation as described by Garfinkel et al., 2011), the Arctic lower stratosphere is cooler in March under La Niña and QBOwesterly conditions, as compared with ENSO neutral and QBO-easterly; however; the magnitude of this cooling is an order of magnitude less than observed in March 2011.Furthermore, the structure and magnitude of dynamical anomalies in the Arctic stratosphere were similar in March 1997 and March 2011, despite different phases of the QBO.
Positive SST anomalies in the North Pacific likely contributed to the anomalous conditions in March 2011.Positive SST anomalies in the 40-50 • N, 160-200 • E region in January and February, such as those observed in 1997 and 2011, are anti-correlated with polar lower stratospheric temperature anomalies in March.Positive SSTs in this region tend to weaken the Aleutian low, leading to a reduced eddy heat flux entering the stratosphere (Garfinkel et al., 2010).However, the relationship between North Pacific SSTs and stratospheric variability is non-linear: While multiple linear regressions to either February eddy heat flux or March polar cap temperature show that the subarctic SST mode is, statistically, the dominant cause of dynamical variability, these linear regressions do not capture the extreme values seen in e.g., 1997 and 2011.

Conclusions
Unusual dynamical conditions were observed in the Arctic stratosphere in March 2011.Tropospheric planetary wave driving was unusually weak, consistent with a strong, stable Arctic vortex in late winter and a relatively late vortex breakup.From a zonal mean perspective, the dynamical conditions observed in 2011 were not unprecedented: February eddy heat flux was weaker and March polar cap temperature was lower in 1997 than in 2011.
While ENSO, the QBO and greenhouse gas-related climate change do not explain the unusual polar stratospheric conditions in March 2011, analysis of the subarctic SST mode suggests that unusually warm SSTs in the North Pacific may have contributed to the remarkable cooling of the Arctic lower stratosphere in both 1997 and 2011.A planned modelling study will, by comparing time-slice simulations of the positive and negative extremes of the subarctic SST mode, isolate the impact of North Pacific SSTs on dynamics and ozone in the Arctic winter and spring.
Figure 2 (left) shows that www.atmos-chem-phys.net/11/11447/2011/February eddy heat flux and March polar cap temperature at 50 hPa are indeed well correlated (with a correlation coefficient of 0.75), and highlights the unusually low values observed in 2011.

Fig. 3 .
Fig. 3. March temperature differences [K]: (a) 2011 from the 1979-2011 climatological mean (i.e., blue contours indicate regions where March 2011 is cooler than the climatology); (b) 1997 from the climatological mean; (c) composite of La Niña events from the climatological mean; (d) QBO-westerly years -QBO-easterly years; (e) strongly positive subarctic SST years -strongly negative subarctic SST years (further discussed in the text).In (c), (d) and (e) black Xs denote differences significant at the 95 % confidence level.Zero difference contours are shown in white.
2, SSTs in the tropical Pacific and March polar cap temperatures are not correlated.However, SSTs in the North Pacific are strongly negatively correlated with March polar cap temperatures.For this study, a subarctic SST index is defined as the January/February mean SST anomaly from the 1979-2011 climatology, in the 40-50 • N, 160-200 • E region.Variability in this region characterises the dominant mode of SST variability in the North Pacific in boreal winter i.e., the 'subarctic mode' identified by Naka-