1
2
3
El Niño-Southern Oscillation in Tropical and Mid-Latitude
4
Column Ozone
5
6
Jingqian Wang,1* Steven Pawson,2 Baijun Tian,3 Mao-Chang Liang,4,5,6
7
Run-Lie Shia,6 Yuk L. Yung,6 and Xun Jiang1
8
9
10
11
12
1 Department of Earth and Atmospheric Sciences, University of Houston, TX 77204,
13
USA.
14
2 Global Modeling and Assimilation Office, NASA GSFC, Code 610.1, Greenbelt, MD
15
20771, USA.
16
17
18
3 Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Dr.,
Pasadena, CA 91109, USA.
4 Research Center for Environmental Changes, Academia Sinica, Taipei, Taiwan.
19
5 Graduate Institute of Astronomy, National Central University, Jhongli, Taiwan.
20
21
22
6 Division of Geological and Planetary Sciences, California Institute of Technology,
1200 East California Boulevard, Pasadena, CA 91125, USA.
23
* To whom all correspondence should be addressed. E-mail: jwang3@mail.uh.edu
24
25
Submitted to JAS April 30 2010; Revise Jan 12, 2011
26
-1-
27
Abstract
28
The impacts of the El Niño-Southern Oscillation (ENSO) on the tropical total
29
column ozone, the tropical tropopause pressure, and on the 3.5-year ozone signal in the
30
mid-latitude total column ozone are examined using the Goddard Earth Observing
31
System Chemistry-Climate Model (GEOS CCM). Observed monthly-mean sea surface
32
temperature and sea ice between 1951 and 2004 were used as boundary conditions for the
33
model. Since the model includes no solar cycle, Quasi-Biennial Oscillation or volcanic
34
forcing, the ENSO signal dominates the tropical total column ozone variability. Principal
35
component analysis is applied to the detrended, deseasonalized, and low-pass filtered
36
model outputs. The first mode of model total column ozone captures 65.8% of the total
37
variance. The spatial pattern of this mode is similar to that in Total Ozone Mapping
38
Spectrometer (TOMS) observations. This ENSO signal in the total column ozone is also
39
well simulated in WACCM3. There is also a clear ENSO signal in the tropical tropopause
40
pressure in the GEOS CCM, which is related to the ENSO signal in the total column
41
ozone. The regression coefficient between the model total column ozone and the model
42
tropopause pressure is 0.78 DU/hPa. GEOS CCM is also used to investigate a possible
43
mechanism for the 3.5-year signal observed in the mid-latitude total column ozone. The
44
3.5-year signal in the GEOS CCM column ozone is very similar to that in the
45
observations, which suggests that a model with realistic ENSO can reproduce the 3.5-
46
year signal. Hence, it is likely that the 3.5-year signal is caused by ENSO.
47
48
-2-
49
1. Introduction
50
The influence of El Niño-Southern Oscillation (ENSO) on the sea surface temperature
51
(SST), surface pressure, winds, and convection is well known (e.g., Trenberth and Shea,
52
1987; Trenberth, 1997). During El Niño events, stratospheric temperature is reduced in
53
the tropics and increased in the Arctic (Garcia-Herrera et al., 2006; Free and Seidel,
54
2009). ENSO can also affect the total column ozone abundance (Bojkov 1987, Shiotani
55
1992, Zerefos 1992, Kayano 1997, Kita et al. 2000, Camp et al. 2003). Zerefos (1992)
56
found the ozone reductions in tropical regions extend to middle and high latitudes
57
following large El Niño events. The decreases in total column ozone in middle and high
58
latitudes lag the ENSO events by several months (Bojkov, 1987). Cagnazzo et al. (2009)
59
showed that many chemistry-climate models capture the global-scale impacts of ENSO in
60
the stratospheric temperature and total ozone fields. Kayano (1997) showed that the first
61
two modes of an empirical orthogonal function study of the total column ozone between
62
70º N and 70º S are related to ENSO. Hood et al. (2010) used a multiple regression
63
method and found that ozone anomalies are negative in the lower stratosphere and
64
positive in the middle stratosphere during El Niño. Chandra et al. (1998) and Thompson
65
and Hudson (1999) found the ENSO signal in the tropospheric column ozone derived
66
from Total Ozone Mapping Spectrometer (TOMS). Tropospheric column ozone
67
decreased in the eastern Pacific and increased in the western Pacific as a result of the
68
eastward shift of the tropical convective activity (Chandra et al., 1998). Kita et al. (2000)
69
used total column ozone from TOMS and found significant increase of total ozone over
70
Indonesia during El Nino periods, which is related to the decreased precipitation and
71
extensive forest fires occurring in the dry season during the El Niño periods. More
-3-
72
recently, Camp et al. (2003) used the total column ozone observed by TOMS and
73
revealed that there is an ENSO signal in the fourth mode of the total column ozone in the
74
tropical region. The power spectral estimate of the PC time series for the fourth mode
75
reveals that most signals are related to the ENSO signal in total column ozone. There are
76
negative total column ozone anomalies in the eastern tropical Pacific and positive total
77
column ozone anomalies in the western Pacific during the El Niño events. The influence
78
of ENSO on total column ozone can be explained by the variation in the tropopause
79
height driven by changes of tropical deep convection and the changes of the Brewer-
80
Dobson circulation (Schubert and Munteanu, 1988; Shiotani, 1992; Hasebe, 1993;
81
Garcia-Herrera et al., 2006; Free and Seidel, 2009). Such a causal relationship between
82
total column ozone and the tropopause height is also evident in the Madden-Julian
83
Oscillation (Tian et al., 2007). However, a quantitative demonstration of the impact of
84
ENSO on total column ozone and tropopause height has not been carried out using model
85
simulations.
86
87
In this paper, we will focus on the ENSO signal in the tropopause pressure and the
88
total column ozone simulated by Version 1 of the Goddard Earth Observing System
89
Chemistry-Climate Model (GEOS CCM) (Pawson et al., 2008). The Whole Atmosphere
90
Community Climate Model version 3 (WACCM3) is also used in this paper for exploring
91
the influence of ENSO on total column ozone (See Appendix). Principal component
92
analysis will be applied to model total column ozone and the spatial pattern of the
93
modeled ENSO signal will be compared with that from Camp et al. (2003). We will also
94
investigate the mechanism for the 3.5-year signal in mid-latitude total column ozone. The
95
3.5-year signal was first found in the total column ozone observations from Arosa and the
-4-
96
same signal also appears in the first leading mode of the northern hemispheric TOMS
97
observations (Jiang et al., 2008). However, the mechanism for the 3.5-year signal in the
98
total column ozone remained unknown: it is explored in this paper.
99
100
2. The GEOS Chemistry-Climate Model and Data
101
The GEOS CCM (Version 1) (Pawson et al., 2008) integrates the GEOS-4 general
102
circulation model with a stratospheric chemistry model (Douglass et al., 2003; Stolarski
103
et al., 2006). Interaction between atmospheric composition and the dynamics is through
104
the radiative heating. In the model radiation module, three-dimensional constituent
105
concentrations (CO2, O3, H2O, CH4, N2O, CFC-11, and CFC-12) vary as a function of
106
time. The model run analyzed in this study, which is the run P1 of Pawson et al. (2008),
107
spanned 54 years (1951-2004) at 2º 2.5º (latitude by longitude) with 55 layers between
108
the surface and about 80km. Zonal-mean fields from this run were also included in
109
Cagnazzo et al. (2009). At the lower boundary, monthly mean SST and sea ice are
110
prescribed from observations (Rayner et al., 2003), along with time-dependent
111
concentrations of greenhouse gases and chloroflorocarbons (see Eyring et al., 2006).
112
Neither the solar cycle nor volcanic aerosol variations were included in the model. This
113
version of the model does not simulate a QBO signal (see Horinouchi et al., 2003). The
114
signals of these three sources of ozone variability are therefore all absent from the GEOS
115
CCM simulation.
116
117
In the isolation of the ENSO signal in the GEOS CCM, two simulated monthly
118
mean fields will be examined: the tropopause pressure (PT) and the total column ozone
-5-
119
(ΩS). The former field is obtained by using the lapse-rate definition tropopause. The latter
120
is the integrated profile of the total amount of ozone in the atmosphere. A comparative
121
analysis of the imposed monthly mean SSTs between 1951 and 2004 will also be shown.
122
123
For comparison with the simulation, monthly mean total column ozone data from
124
ground
stations
will
be
analyzed.
(Data
can
be
downloaded
from
125
http://www.woudc.org/data_e.html). Twenty-one stations with continuous records longer
126
than 20 years will be used in this paper, so as to include several full cycles of the 3.5-year
127
ozone signal. Total column ozone time series from these stations will be compared with
128
the GEOS CCM model simulation.
129
130
3. Methods
131
The filtering and analysis technique of Jiang et al. (2008) is applied to the GEOS
132
CCM model outputs, which are the gridded monthly mean SST, tropopause pressure (PT),
133
and total column ozone time series (ΩS). The mean annual cycle and the linear trend
134
computed for the 54-year monthly time series were subtracted from the original time
135
series to create the detrended monthly anomalies for each month. The mean annual cycle
136
is calculated from the monthly mean fields for 54 years and the linear trend is calculated
137
as linear regression of the 54-year time series. A spectral filter was then applied to these
138
monthly anomalies to separate the interannual variability (IAV) from higher frequency
139
oscillations. As in Jiang et al. (2008), the filter was constructed to extract signals at
140
periods longer than 15 months by convolving a step function with a Hanning window.
141
-6-
142
These time-filtered anomalies of GEOS CCM SST, PT, and ΩS were used in a
143
principal component analysis (PCA) (Preisendorfer, 1988, Camp et al., 2003) to isolate
144
the leading spatial empirical orthogonal functions (EOFs) and their associated time-
145
dependent amplitudes (the principal component time series, PCs). PCA can reduce the
146
dimensionality of the original data set, and small sets of orthogonal functions are much
147
easier to understand and analyze (Dunteman, 1989). The EOFs are the orthogonal
148
eigenfunctions of the covariance matrix of the dataset sorted by the decreasing values of
149
associated eigenvalues. Since these eigenvalues represent the variance captured by each
150
EOF, PCA guarantees that the leading EOFs capture more of the total variance of the
151
dataset than any other orthogonal vectors.
152
153
Observed total column ozone at different stations were deseasonalized and
154
detrended before the power spectral analysis was applied. Spectral analysis was also
155
applied to the model deseasonalized and detrended total column ozone. To obtain the
156
statistical significance of signals in a power spectrum, the amplitudes of the spectral
157
peaks were compared with a red noise spectrum (Gilman et al., 1963). The 10%, and 5%
158
significance levels for the power spectrum are found using F-statistics to compare the
159
spectrum to the red noise spectrum.
160
161
-7-
162
4. ENSO and 3.5-year Signals in Total Column Ozone
163
4.1 ENSO Signals
164
The leading mode of the PCA of filtered tropical (25ºS to 25ºN) anomalies
165
capture 46.9%, 33.9%, and 63.8% of the total variance in GEOS CCM SST, P T, and ΩS,
166
respectively (Table 1). The Southern Oscillation Index (SOI, the sea level pressure
167
difference between Tahiti and Darwin) is used to calculate the correlations between
168
ENSO
169
http://www.cpc.noaa.gov/data/indices/soi. For a fair comparison, we applied a lowpass
170
filter to the detrended SOI index. Comparison of the time series of the detrended and
171
lowpass filtered SOI and the amplitude of the first mode (PC1) of the SST (Fig 1) reveals
172
a good correlation, with in-phase variations. Note that negative/positive SOI values
173
denote the El Niño/La Niña events. The correlation between the two time series is 0.88
174
and it is significant at the 0.1% level. The significance level was computed by a Monte
175
Carlo method (Press et al., 1992): a small numerical value denotes high statistical
176
significance. Time series of PC1 for PT and ΩS reveal a similar low-frequency oscillation
177
as for lowpass filtered SOI, with some higher-frequency variations superimposed. This
178
leads to somewhat lower correlations (0.53 and 0.65) between these time series and the
179
lowpass filtered SOI, but these remain highly significant (see Table 1). The correlation
180
between PC1s of PT and ΩS is 0.82. The power spectra of the PC1s of SST, PT and ΩS
181
(Fig. 2) all show strong spectral peaks near 3-7 years, which have the same power
182
spectral signals as the ENSO variations. The higher frequency variations evident in PT
183
and ΩS also lead to prominent peaks near 17 months, which might be a beat frequency
184
between the annual cycle and the ENSO signal.
and
leading
PC
time
series.
-8-
The
SOI
is
downloaded
from
185
186
Figure 3 shows the spatial patterns of the first mode of the PCA analysis for
187
GEOS CCM SST, PT, and ΩS, which describe their ENSO-related variability. For SST,
188
the spatial pattern (Fig. 3a) is similar to the first mode of ENSO-related variability
189
described in Trenberth et al. (2002, see Figure 11), which corresponds to the SST
190
anomalies in the Niño-3.4 region (covering 120°W-170°W, 5°N-5°S). Negative
191
amplitudes of this mode correspond to the El Niño events, with warm SST anomalies in
192
the central and eastern Pacific and cold anomalies in the western Pacific. During El Niño
193
periods, the convection will move eastward to the central Pacific and impact the
194
longitudinal structure of the tropopause (e.g., Gage and Reid, 1987, Gettelman et al.
195
2001; Kiladis et al. 2001). The spatial pattern of the first mode of PT (Fig. 3b) is
196
consistent with this behavior. For the negative amplitudes of this mode, corresponding to
197
El Niño events, there are large negative tropopause pressure anomalies (higher
198
tropopause height) in the central and eastern Pacific and weak positive tropopause
199
pressure anomalies (lower tropopause heights) in the equatorial western Pacific. Please
200
note the tropopause pressure anomalies are relatively small over the equatorial regions.
201
On the other hand, the large tropopause pressure anomalies are found over the subtropics,
202
which is the Rossby-wave response of the equatorial ENSO convection anomaly
203
(Trenberth et al., 1998). In the positive amplitude of this mode, that is, La Niña events,
204
there are weak negative tropopause pressure anomalies (higher tropopause heights) in the
205
equatorial western Pacific and large positive tropopause pressure anomalies (lower
206
tropopause heights) in the central and eastern Pacific. The present results are in general
207
agreement with previous studies (e.g., Kiladis et al., 2001; Gettelman et al., 2001) on the
-9-
208
tropopause pressure and height variability associated with the ENSO based on reanalysis
209
data. This indicates that the vertical movement of the tropopause associated with the
210
ENSO is reasonably captured by the GEOS CCM.
211
212
One important aspect of the spatial pattern of the PT anomaly is that it has the
213
same sign everywhere except for a confined region centered over the Equator to the
214
North of Indonesia (Fig. 3b). This contrasts with the existence of positive and negative
215
anomalies in the SST pattern. This means that almost the entire tropical tropopause is
216
displaced upwards (to lower pressure) in El Niño events, with a discernable signal in the
217
zonal mean. A similar result holds for the ΩS pattern (Fig. 3c), in that substantial zonal
218
anomalies are superimposed on a monotonic signal, which implies the forcing of the
219
Brewer-Dobson circulation (Garcia-Herrera et al., 2006) in addition to the forcing by
220
tropical convection during El Niño events. Note that the ENSO signal dominates the
221
variability in ΩS because irradiance changes associated with the solar cycle were not
222
included in the simulation and because this version of GEOS CCM does not simulate a
223
QBO. The pattern is very similar to the ENSO spatial pattern in the TOMS total column
224
ozone data found by Camp et al. (2003, see Figure 5), although there are some
225
discrepancies in the southern hemisphere. The discrepancies in the total column ozone in
226
the southern hemisphere might arise because that the model has an unrealistic
227
representation of the Antarctic polar vortex, with too much interannual variabilitry
228
compared to observations (Pawson et al., 2008). The fourth mode of TOMS total column
229
ozone obtained by Camp et al. (2003) is not entirely explained by the ENSO signal.
230
There are other weak signals (e.g. decadal signal) existing in the fourth mode of TOMS
- 10 -
231
ozone (Camp et al. 2003), while there is no decadal signal in the results from model
232
ozone. This might explain some discrepancies between model and observation. The
233
spatial pattern of ENSO signal in the model total column ozone is also consistent with the
234
second mode (mature stages of the El Niño) for the total column ozone from Kayano et
235
al. (1997, see Figure 1). The value is relatively low in the western Pacific and relatively
236
high in the eastern Pacific.
237
238
A broadly consistent picture arises from the leading anomalies, with (in El Niño
239
events) an eastward migration of warm SST anomalies that enhance convection over the
240
eastern Pacific where the tropopause rises and ΩS decreases. This decrease in total
241
column ozone is consistent with the changes of the tropopause. It is also consistent with
242
a faster vertical flow through the ozone source region in the tropical stratosphere, which
243
tends to reduce the total column ozone. During El Niño, vertical propagation of Rossby
244
wave and divergence of Eliassen-Palm flux will be enhanced, which will accelerate the
245
Brewer-Dobson circulation (Garcia-Herrera et al., 2006). Thus the vertical flow in the
246
tropical stratosphere will be accelerated. It will lead to a decrease in total column ozone
247
in the tropical region and an increase of total column ozone in the middle to high
248
latitudes.
249
250
Since there is strong correspondence between total column ozone and tropopause
251
pressure, we regress the first mode of the total column ozone on the first mode of the
252
tropopause pressure. The mean value for the first mode of the total column ozone in the
253
Niño-3.4 region (120°W-170°W, 5°N-5°S) is plotted against the mean value for the first
- 11 -
254
mode of the tropopause pressure in Fig. 4a. The Niño-3.4 region is chosen for its
255
proximity to the Pacific warm pool and main centers of convection (Trenberth, 1997).
256
When the Niño 3.4 region SST anomaly is larger than +0.4°C, it corresponds to an El
257
Niño event. Time series of total column ozone and tropopausre pressure averaged in the
258
Niño-3.4 region correlate well. The correlation coefficient is 0.82 and it is significant at
259
the 0.1% level. The scatter plot for the two time series is shown in Fig. 4b. The regression
260
coefficient between the total column ozone and tropopause pressure is 0.71 DU/hPa.
261
Assuming that the tropopause height is 10.5 km and scale height is 7 km, the regression
262
coefficient translates -15.5 DU/km, which is consistent with the previous results that
263
range from -10 to -25 DU/km (Meetham, 1937; Reed, 1950; Hoinka et al., 1996;
264
Steinbrecht et al., 1998).
265
266
4.2 The 3.5-year Ozone signal
267
Total column ozone data at twenty-one stations are chosen to investigate the 3.5-
268
year signal. These stations are chosen for their relatively long time series (more than 20
269
years). Spectral power analysis is applied to the deseasonalized and detrended total
270
column ozone at all stations. In the twenty-one stations, only eight stations display a
271
significant 3.5-year signal. The other stations have signals at 4-5 years. There is also no
272
preference of the locations for the stations that have the 3.5-year signal versus stations
273
with signals at 4-5 years in the total column ozone. The eight stations with significant
274
3.5-year signal are Potsdam, Belsk, Hradec, Hohenpeissenberg, Arosa, Toronto, Sapporo,
275
and Nashville. The geographical locations and record lengths of the eight stations are
276
listed in Table 2. The power spectra and statistical significances of the peaks are shown in
- 12 -
277
Figure 5 for the eight stations. Dotted lines are the red-noise spectra. Dash-dot lines and
278
dashed lines show 10% and 5% significance levels, respectively. In Fig. 5a, there are
279
about five significant spectral peaks (10% significance level) in the low-frequency
280
regions: the residual annual cycle, QBO-Annual beat, QBO, 3.5-year signal, and a
281
decadal oscillation. Since the mechanisms for the QBO-Annual beat, QBO, and the
282
decadal oscillation are well known (Tung and Yang, 1994a; Tung and Yang, 1994b;
283
Jiang et al., 2005; Baldwin et al., 2001; Soukharev and Hood, 2006), we only focus on
284
the 3.5-year signal in this paper. Among eight stations, there are significant peaks
285
between 3-4 years. The power spectra of Fig. 5a-Potsdam, d-Hohenpeissenberg, and g-
286
Sapporo lie within the 10% to 5% range of significance. The significance of the 3.5-year
287
signals in the other five stations, Fig. 5b-Belsk, c-Hradec, e- Arosa, f-Toronto, h-
288
Nashville, exceeds the 5% level.
289
290
In order to verify the mechanism for the 3.5-year cycle and explain the results
291
presented above, we investigated the total column ozone signal in the GEOS-CCM.
292
Spectral analysis of the simulated total column ozone (Fig. 6) reveals statistically
293
significant 3.5-year signals at the eight stations. The significance levels are within 10%-
294
5% for the total column ozone at Potsdam, Belsk, Hradec, Hohenpeissenberg, Arosa, and
295
Sapporo (Figs. 6a-6e and Fig. 6g), while at Toronto and Nashville (Fig. 6f and Fig. 6h),
296
the significance levels are better than 5%.
297
298
Because the GEOS CCM simulation is forced by observed SST and sea-ice
299
distributions, its atmosphere does represent variability related to these boundary forcings,
- 13 -
300
which include ENSO events. The absence of both the QBO and solar cycle in the model
301
means that total column ozone variations related to these phenomena are not captured.
302
Since the 3.5-year signal isolated in total column ozone observations is captured at a
303
significant level in the GEOS CCM, this suggests that this signal originates from the
304
ENSO signal in SST forcing, rather than from other forced variations in stratospheric
305
ozone, although on the basis of this analysis we cannot preclude the possibility that
306
internal (unforced) dynamics cause this signal.
307
308
5. Conclusion
309
In this paper, we have studied the ENSO signal in the tropical total column ozone
310
and tropical tropopause pressure in the GEOS CCM from Jan 1951 to Dec 2004. The first
311
modes in the model tropical total column ozone and tropopause pressure capture 63.8%
312
and 33.9% of the total variances, respectively. The PC1 for the first modes in the model
313
total column ozone and tropopause pressure correlate well with the lowpass filtered SOI
314
index and leading PC time series for the SST. The spatial pattern of the first mode for the
315
model total column ozone is similar to the ENSO signal in the TOMS total column
316
ozone, although there are some discrepancies for the total column ozone in the southern
317
hemisphere. The regression coefficient between the model total column ozone and the
318
model tropopause pressure is consistent with that found from observations.
319
320
In the middle latitudes, we applied spectral analysis to the observed monthly
321
mean total column ozone data and find the 3.5-year signal, exceeding the 10%
322
significance level, at eight stations. The 3.5-year signal is also found in the GEOS CCM
- 14 -
323
model total column ozone. Since GEOS-CCM does have a realistic ENSO signal and
324
does not have a QBO or a solar cycle signals, the results suggest that the 3.5-year signal
325
in total column ozone is related to the ENSO signal. The linkage between 3.5-year ozone
326
signal and surface ENSO needs to be investigated in more detail in the future, for
327
example using controlled model experiments in which the ENSO forcing is removed.
328
329
Acknowledgments. We thank Bernhard Rappenglueck, Barry Lefer, and three
330
anonymous reviewers for useful inputs and helpful comments. We thank World Ozone
331
and Ultraviolet Radiation Data Centre for providing the ozone observation data.
332
- 15 -
333
334
335
APPENDIX
336
In this appendix, we provide the information on El Niño-Southern Oscillation
337
(ENSO) signal in the monthly mean total column ozone simulated by Whole Atmosphere
338
Community Climate Model version 3 (WACCM3).
339
340
a. WACCM3 Model Description
341
WACCM3 is a chemistry-climate model, which is based on the software
342
framework produced by the National Center for Atmospheric Research’s Community
343
Atmospheric Model (CAM). WACCM3 includes most physical and chemical
344
mechanisms. The model run analyzed in this study, which spanned 42 years (1960-2001)
345
at 4º 5º (latitude by longitude) with 66 layers between the surface and about 610-6 hPa.
346
At the lower boundary, monthly mean SST and sea ice are prescribed from observations.
347
Solar cycle is also included in the model simulation. Quasi-biennial Oscillation signal is
348
not included in the WACCM3.
349
350
b. ENSO Signal in WACCM3 Total Column Ozone
351
352
Principal component analysis (PCA) (Preisendorfer, 1988) was performed on the
353
low-pass filtered and detrended tropical (25ºS to 25ºN) monthly mean total column ozone
354
from WACCM3. The filter was constructed to obtain a full signal from periods above 15
355
months and no signal from periods below 12.5 months. The first mode captures 67.7% of
356
the total variance in WACCM3 total column ozone. Spatial pattern of the first mode of
357
the total column ozone is shown in Fig A1a. The spatial pattern is very similar to the
358
ENSO spatial pattern in the TOMS total column ozone data in Camp et al. (2003, see
359
Figure 5). It is also consistent with the ENSO spatial pattern in total column ozone from
360
GEOS-CCM, except that the ENSO signal in WACCM3 total column ozone are larger
- 16 -
361
than those in GEOS-CCM. A double-ridged symmetric structure with ridges at 15º N and
362
10º S is found, which is similar to that from Camp et al. (2003). The value is relatively
363
low in the western Pacific and relatively high in the eastern Pacific.
364
365
c. Discussion
366
PC1 time series for the first mode of the WACCM3 total column ozone is plotted
367
against the lowpass filtered SOI in Fig. A1b. These two time series correlate with each
368
other well. The correlation coefficient is 0.46 (0.1% significance level). The power
369
spectrum of the PC1 (Fig. A1c) indicates strong spectral peaks at 3-7 years, which are
370
typical of ENSO variations. The peaks near 17 months might be a beat frequency
371
between the annual cycle and ENSO signal. There is also a significant peak between 3-4
372
years, which suggests that the 3.5-year signal in total column ozone might originate from
373
ENSO. There is also a decadal signal in the power spectrum, for there is solar forcing in
374
the WACCM3 model.
375
- 17 -
376
377
REFERENCES
378
Baldwin M. P., and Coauthors, 2001: The quasi-biennial oscillation. Rev. Geophys., 39,
379
380
381
179-229.
Bojkov, R. D., 1987: The 1983 and 1985 anomalies in ozone distribution in perspective.
Mon. Wea. Rev., 115, 2187-2201.
382
Chandra, S., J. R. Ziemke, W. Min, and W. G. Read, 1998: Effects of 1997-1998 El Niño
383
on tropospheric ozone and water vapor. Geophys. Res. Lett.,25, 3867-3870.
384
Cagnazzo, C., et al. 2009: Northern winter stratospheric temperature and ozone responses
385
to ENSO inferred from an ensemble of Chemistry Climate Models. Atmospheric
386
Chemistry and Physics, 9, 8935–8948.
387
Camp, C. D., M. S. Roulston, and Y. L. Yung, 2003: Temporal and spatial patterns of the
388
interannual variability of total ozone in the tropics. J. Geophys. Res, 108, D4643,
389
doi:10.1029/2001JD001504.
390
Douglass, A. R., M. R. Schoeberl and R. B. Rood, 2003: Evaluation of transport in the
391
lower tropical stratosphere in a global chemistry and transport model. J. Geophys.
392
Res., 108, D4259, doi:10.1029/2002JD002696.
393
394
Dunteman, G. H., 1989: Principal Components Ahanlysis, International Education and
Professional Publisher, 98pp.
395
Eyring, V., and Coauthors, 2006: Assessment of temperature, trace species, and ozone in
396
chemistry-climate model simulations of the recent past. J. Geophys. Res., 111,
397
D22308, doi:10.1029/2006JD007327.
- 18 -
398
Free, M., and D. J. Seidel, 2009: Observed El Niño-Southern Oscillation temperature
399
signal
in
the
stratosphere.
400
doi:10.1029/2009JD012420.
J.
Geophys.
Res.,
114,
D23108,
401
Gage, K.S., and G. C. Reid, 1987: Longitudinal variations in tropical tropopause
402
properties in relation to tropical convection and El Niño-Southern Oscillation. J.
403
Geophys. Res., 92, 14,197-14,203.
404
Garcia-Herrera, R., N. Calvo, R. R. Garcia, and M. A. Giorgetta, 2006: Propagation of
405
ENSO temperature signals into the middle atmosphere: A comparison of two
406
general circulation models and ERA-40 reanalysis data. J. Geophys. Res., 111,
407
D06101, doi:10.1029/2005JD006061.
408
Gettelman, A., W. J. Randel, S. Massie, F. Wu, W. G. Read, and J. M. Russell III, 2001:
409
El Niño as a natural experiment for studying the tropical tropopause region. J.
410
Climate, 14, 3375-3392.
411
412
413
414
415
416
Gilman, D. L., F. J. Fuglister, and J. M. Mitchell, 1963: On the Power Spectrum of “Red
Noise”. J. Atmos. Sci., 20, 182-184.
Hasebe, F., 1993: Dynamical response of the tropical total ozone to sea surface
temperature changes. J. Atmos. Sci., 50, 345-356.
Hoinka, K. P., H. Claude, and U. Kohler, 1996: On the correlation between tropopause
pressure and ozone above central Europe. Geophys. Res. Lett., 23, 1753-1756.
417
Hood, L. L., B. E. Soukharev, and J. P. McCormack, 2010: Decadal variability of the
418
tropical stratosphere: secondary influence of the El Nino-Southern Oscillation. J.
419
Geophys. Res., 115, D11113, doi:10.1029/2009JD012291.
- 19 -
420
Horinouchi, T., S. Pawson, K. Shibata, U. Langematz, E. Manzini, M. A. Giorgetta, F.
421
Sassi, R. J. Wilson, K. P. Hamilton, J. de Grandpré, A. A. Scaife, 2003: Tropical
422
Cumulus Convection and Upward Propagating Waves in Middle Atmospheric
423
GCMs. J. Atmos. Sci., 60, 2765-2782.
424
Jiang, X., D. B. A. Jones, R. Shia, D. E. Waliser, and Y. L. Yung, 2005: Spatial patterns
425
and mechanisms of the Quasi-Biennial Oscillation-Annual Beat of ozone. J.
426
Geophys. Res., 110, D23308, doi:10.1029/2005JD006055.
427
Jiang, X., S. Pawson, C. D. Camp, E. Nielsen, R. Shia, T. Liao, K. Jeev, V. Limpasuvan,
428
and Y. L. Yung, 2008: Interannual variability and trends in extratropical ozone.
429
Part I: Northern hemisphere. J. Atmos. Sci., 65, 3013-3029.
430
Jiang, X., Q. Li, M. Liang, R. L. Shia, M. T. Chahine, E. T. Olsen, L. L. Chen, and Y. L.
431
Yung, 2008: Simulation of upper troposphere CO2 from chemistry and transport
432
models. Global Biogeochem. Cycles, 22, doi:10.1029/2007GB003049.
433
Kayano, M. T., 1997: Principal modes of the total ozone on the Southern Oscillation
434
timescale and related temperature variations. J. Geophys. Res., 102, 25797-25806.
435
Kita, K., M. Fujiwara, and S. Kawakami, 2000: Total ozone increase associated with
436
forest fires over the Indonesian region and its relation to the El Nino-Southern
437
oscillation. Atmos. Envir., 34, 2681-2690.
438
439
Kiladis, G. N., Straub K. H., Reid G. C., and Gage K. S., 2001: Aspects of interannual
440
and intraseasonal variability of the tropopause and lower stratosphere. Quart. J.
441
Roy. Meteor. Soc., 127, 1961-1983.
442
443
Meetham, A. R., 1937: The correlation of the amount of ozone with other characteristics
of the atmosphere. Quart. J. Roy. Meteor. Soc., 63, 289-307.
- 20 -
444
Pawson, S., R. S. Stolarski, A. R. Douglass, P. A. Newman, J. E. Nielsen, S. M. Frith,
445
and M. L. Gupta, 2008: Goddard Earth Observing System chemistry-climate
446
model simulations of stratospheric ozone-temperature coupling between 1950 and
447
2005. J. Geophys. Res., 113, D12103, doi:10.1029/2007JD009511.
448
449
Preisendorfer, R. W., 1988: Principal component analysis in meteorology and
oceanography. Dev. in Atmos. Sci., 17, 425 p.
450
Press, W., S. Teukolsky, W. Vetterling, and B. Flannery, 1992: Numerical Recipes in
451
Fortran 77: The Art of Scientific Computing, 933 pp. Cambridge Univ. Press,
452
New York.
453
Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander, D. P. Rowell,
454
E. C. Kent, and A. Kaplan, 2003: Global analyses of sea surface temperature, sea
455
ice, and night marine air temperature since the late nineteenth century. J.
456
Geophys. Res., 108, D4407, doi:10.1029/2002JD002670.
457
458
459
460
461
462
Reed, R. J., 1950: The role of vertical motions in ozone-weather relationship. J. Meteor.,
7, 263-267.
Schubert, S. D., and M. J. Munteanu, 1988: An analysis of tropopause pressure and total
ozone correlations. Mon. Wea. Rev., 116, 569-582.
Shiotani, M., 1992: Annual, quasi-biennial, and El Niño-Southern Oscillation (ENSO)
timescale variations in equatorial total ozone. J. Geophys. Res., 97, 7625-7633.
463
Soukharev, B. E., and L. L. Hood, 2006: Solar cycle variation of stratospheric ozone:
464
Multiple regression analysis of long-term satellite data sets and comparisons with
465
models. J. Geophys. Res., 111, D20314, doi:10.1029/2006JD007107.
- 21 -
466
Steinbrecht, W., H. Claude, U. Kohler, and K. P. Hoinka, 1998: Correlations between
467
tropopause height and total ozone: Implications for long-term changes. J.
468
Geophys. Res., 103, 19183-19192.
469
Stolarski, R. S., A. R. Douglass, S. E. Steenrod, S. Pawson, 2006: Trends in stratospheric
470
ozone: Lessons learned from a 3D chemical transport model. J. Atmos. Sci., 63,
471
1028-1041.
472
Tian, B., Y. L. Yung, D. E. Waliser, T. Tyranowski, L. Kuai, E. J. Fetzer and F. W. Irion,
473
2007: Intraseasonal variations of the tropical total ozone and their connection to
474
the
475
doi:10.1029/2007GL029451.
Madden-Julian
Oscillation,
Geophys.
Res.
Lett.,
34,
L08704,
476
Thompson, A. M., and R. D. Hudson, 1999: Tropical tropospheric ozone (TTO) maps
477
from Nimbus 7 and Earth Probe TOMS by the modified-residual method:
478
Evaluation with sondes, ENSO signals, and trends from Atlantic regional time
479
series. J. Geophys. Res., 104, 26961-26975.
480
481
Trenberth, K. E. and D. J. Shea, 1987: On the evolution of the Southern Oscillation event.
Mon. Wea. Rev., 115, 3078-3096.
482
Trenberth, K. E., 1997: The Definition of El Niño. Bull. Amer. Met. Soc., 78, 2771-2777.
483
Trenberth, K. E., 1998: Progress during TOGA in understanding and modeling global
484
teleconnections associated with tropical sea surface temperatures. J. Geophys.
485
Res., 103, 14291-14324.
486
Trenberth, K. E., J. M. Caron, D. P. Stepaniak, and S. Worley, 2002: Evolution of El
487
Niño-Southern Oscillation and global atmospheric surface temperatures. J.
488
Geophys. Res., 107, doi: 10.1029/2000JD000298.
- 22 -
489
490
491
492
Tung, K. K., and H. Yang, 1994a: Global QBO in circulation and ozone: 1.
Reexamination of observational evidence. J. Atmos. Sci., 51, 2699-2707.
Tung, K. K., and H. Yang, 1994b: Global QBO in circulation and ozone: 2. A simple
mechanistic model. J. Atmos. Sci., 51, 2708-2721.
493
Zerefos, C. S., A. F. Bais, I. C. Ziomas, and R. D. Bojkov, 1992: On the Relative
494
importance of Quasi-Biennal Osicillation and ELNino/Southern Oscillation in the
495
Revised Dobson total Ozone Records. J. Geophys. Res., 97, 10135-10144.
496
497
498
- 23 -
499
Table 1: Variances, spectral peaks, and correlations (Lag = 0) for the first modes of the
500
GEOS CCM sea surface temperature (SST), tropopause pressure (PT), and total column
501
ozone (ΩS). The numbers in parentheses denote significance levels.
502
Mode 1
Variance
Spectral Peaks
Correlations (significance level) with:
of
captured
SST
46.9%
3.5-year; 5-year
PT
33.9%
17 months; 28 months; 0.65 (0.1%)
Lowpass filtered SOI
SST (PC1)
PT (PC1)
0.88 (0.1%)


0.72 (0.1%)

0.57 (0.1%)
0.82 (0.1%)
3.5-year; 5-year
ΩS
63.8%
17 months; 28 months; 0.53 (0.1%)
3.5-year
503
504
- 24 -
505
Table 2. Geographical locations and record lengths of total column ozone time series
506
observed at different stations.
Stations
Latitude
Longitude
Time series
Potsdam
52.36
13.08
1964 - 2003
Belsk
51.84
20.79
1963 – 2003
Hradec
50.18
15.83
1961 – 2005
Hohenpeissenberg
47.8
11.02
1967 – 2005
Arosa
46.7
9.68
1926 – 2005
Toronto
43.78
-79.47
1960 – 2003
Sapporo
43.05
141.33
1958 – 2005
Nashville
36.25
-86.57
1962 - 2004
507
508
- 25 -
509
510
Figure 1: PC1 time series for the first modes of GEOS CCM SST (green dash-dotted
511
line), tropopause pressure (blue dotted line), and total column ozone (red dashed line).
512
Lowpass filtered and detrended SOI index is shown by black solid line.
513
- 26 -
514
515
Figure 2: (a) Power spectrum of PC1 for the first mode of GEOS CCM SST. (b) Power
516
spectrum of PC1 for the first mode of GEOS CCM tropopause pressure. (c) Power
517
spectrum of PC1 for the first mode of GEOS CCM total column ozone.
518
- 27 -
519
520
521
522
523
524
525
526
Figure 3: (a) The spatial pattern of the first mode for GEOS CCM SST in the tropics.
Units are K. The first mode explains 46.9% of the total variance. (b) The spatial pattern
of the first mode for GEOS CCM tropopause pressure in the tropics. Units are hPa. The
first mode explains 33.9% of the total variance. (c) The spatial pattern of the first mode
for GEOS CCM total column ozone in the tropics. Units are DU. The first mode explains
63.8% of the total variance.
- 28 -
527
528
529
530
531
532
533
534
535
Figure 4: (a) First mode of GEOS CCM total column ozone averaged over the Niño-3.4
region (solid line) and first mode of GEOS CCM tropopause pressure at the Niño-3.4
region (dotted line). Correlation coefficient between two time series is 0.82 (0.1%
significance level). (b) Scatter plot for GEOS CCM total column ozone and the GEOS
CCM tropopause pressure. Both data are averaged over the Niño-3.4 region. Solid line is
the linear fit of the data.
- 29 -
536
537
538
539
540
541
542
Figure 5: Power spectra of the deseasonalized and detrended total column ozone observed
at different stations: (a) Potsdam, (b) Belsk, (c) Hradec, (d) Hohenpeissenberg, (e) Arosa,
(f) Toronto, (g) Sapporo, (h) Nashville. Dotted lines are the mean red-noise spectra.
Dash-dot lines and dashed lines are 10% and 5% significance levels, respectively. P
denotes the 3.5-year signal.
- 30 -
543
544
545
546
547
548
Figure 6: Power spectra of the deseasonalized and detrended GEOS CCM total column
ozone data sampled in the same locations as those in Figure 5. Dotted lines are the mean
red-noise spectra. Dash-dot lines and dashed lines are 10% and 5% significance levels,
respectively. P denotes the 3.5-year signal.
- 31 -
549
550
551
552
553
554
555
Figure A1: (a) Spatial pattern of the first mode for tropical total column ozone from
WACCM3. Units are DU. The first mode captures 67.7% of the total variance in
WACCM3 total column ozone. (b) PC1 time series for the first mode of WACCM3 total
column ozone (red dashed line) and lowpass filtered and detrended SOI (black solid line).
(c) Power spectrum of PC1 for the first mode of total column ozone from WACCM3.
- 32 -