Further evidence of solar cycle variability in
middle atmospheric ozone and the
importance of incorporating solar spectral
irradiance in atmospheric modeling
Aimee Merkel, Jerald Harder, Thomas Woods, Anne Smith, David Rusch, Mar<n Mlynczak Laboratory for Atmospheric and Space Physics (LASP) University of Colorado Na<onal Center for Atmospheric Research (NCAR ACD ) NASA Langley Research Center, Hampton, VA Jan. 28, 2014 SORCE Science Mee5ng 1 a Lean model
60
0.3
It all started with Haigh et al. Nature 2010 Pressure (hPa)
1
0.3
0.4
0.5
0.6
3
0.3
0.4
0.5
0.6
0.7
0.7
SSI influence study using a 2-­‐D (la5tude-­‐height) radia5ve-­‐chemical-­‐transport model of the atmosphere (IC2-­‐D). LETTERS
|
10
0.6
Modeled ozone response (max-­‐min) 0.5
60 0.3
0.3
0.2
1
0.3
0.4
0.5
0
Latitude (°N)
0.7
30
0.3
0.4
0.5
0.6
50
1
–1.4
20
2
0
–2
–4
b 10–6.8 hPa
4
2
0
–2
40
4
60
2
0
–1.2
–1.2
–1.0
–2
–0.8
–0.6
–4
–0.4
–0.2
0.0
0.2
0.4
0.6
0.8
1.0 10–6.8
1.2
1.4
1.6
1.8 4
4
NATURE Vol 0.6
000 | 00 Month 2010
–4
50
0.2
0.4
0.8
20060.61.0
1.2
1.4
1.6
0.0
Approximate altitude (km)
10
–20
40
SORCE model run O3 anomalies (%)
4
0.3
–40
0.8
a 0.68–0.32data
hPa
b SIM/SOLSTICE
Pressure (hPa)
Approximate altitude (km)
200-­‐300nm important for mesospheric ozone NRLSSI model run a Lean model
Pressure (hPa)
15
Difference in spectral irradiance at >242 nm (mW m–2 nm–1)
30
6
50
Approximate altitude (km)
0.2
O3 anomalies (%)
a 0.68–0.32 hP
0.3
O3 anomalies (%)
Supplementary Information for further details.) This type of model
produces realistic simulations of the upper stratosphere (above about
25 km) but is less reliable at lower altitudes where photochemical
time constants are longer and a more accurate representation of
transport processes is required. The results below come from four
model runs using solar spectra derived from the SIM measurements
(with SOLSTICE data for wavelengths less than 200 nm) and those
produced by the Lean model, each for both 2004 and 2007.
In Fig. 2 we present latitude–height maps of the difference between
2004 and 2007 in December ozone concentrations. The Lean spectral
data produce a broad structure of ozone concentrations greater in
2004 than in 2007, with maximum values of around 0.8% near
40 km, whereas the SIM data produce a peak enhancement of over
2% in low latitudes around 35 km, along with significant reductions
above 45 km. The predicted temperature differences (Supplementary
Difference in spectral irradiance at >242 nm (mW m–2 nm–1)
onse of the
n the speceen 0.2 mm
M) instrut (SORCE)
s declining
ger decline
n our preensated in
ible waveear to have
atospheric
above this
ochemical
ements of
ime period
also show,
e climate is
nsufficient
s observed
ur findings
on temperto current
Figure 3 | Tim
(solid black lin
deseasonalized
values reconst
dashed lines. T
components a
Fig. 2. Data we
6.8 hPa (b).
Over the p
at altitudes 3
increasing co
1.8
40
40
NRLSSI 2000, howev
0.8
Lean model
2.0
stopped decl
2
0
0
SOLSTICE
probably also
10
10
0.7
2.0
1.6
0
SIM
0.6
the chemical
–2
1.8
30
the present s
30
0.6
1.6
0.5
1.4 –4
factors from
–5
–2
30
200
300
400
500
600
700
30
–40
–20
20 2006
40
2005 0
2007seems likely
–40
–20
0
20
40
Latitude (°N)
Wavelength (nm)
ozone below
Latitude (°N)
Figure 3 | Time series of AURA-MLS v2.2 ozone concentrations. The
is data
also more
Figure 2(solid
| Modelled
difference
in
ozone
between
December
2004
and
Figure 1 | Difference in solar spectrum between April 2004 and November
b SIM/SOLSTICE data
black lines) are percentage anomalies of tropical (22.5 uS–22.5 tral
uN) variation
Estimates
of
the
percentage
difference
(2004–2007)
in
December
2007.
2007. The difference (2004–2007) in solar spectral irradiance
0.3
deseasonalized
monthly
means
from
August
2004
to
November
2007.
The
60zonal mean ozone concentration (labels on contour lines in per cent)
–1.4
solar signal i
(W m22 nm21) derived from SIM data4 (in blue), SOLSTICE data8 (in red)
values reconstructed from the 4-component regression model are shown as
produced by the model using solar spectra from the Lean model (a) and SIM/
(1979 to 200
and from the Lean model5 (in black). Different scales are used for values at
dashed lines. The solar component of the regression is shown in red. Other
SOLSTICE data (b).
that the decl
–1.2
wavelengths less and more than 242 nm (see left and
right
axes
respectively).
Mesosphere – out-­‐of-­‐phase –1.2
Stratosphere – in-­‐phase components
are shown, along with the solar component, in Supplementary
previous
and sola
Fig. 2. Data were averaged between 0.68 hPa and 0.32 hPa (a) and 10 hPa
1
–1.0
aboratory for Atmospheric and Space Physics, University of Colorado, 1234 Innovation Drive, Boulder,
behaviours d
also (very
50Fig. 1) are
b). different, with the Lean data set showing tempera6.8 hPa
–0.8
–0.6
To unders
tures 0.3–0.4 K greater in 2004 than in 2007 at the top of the model
–0.4
0.2
0.0
–0.2
0.4
Over thethe
period
from
late 1970s
to the
late 1990s
the modelled
1
domain, whereas
SIM data
setthe
produces
a peak
warming
of 1.8tropical
K at ozone
0.6 0.8
0.2 0.0
0.4
1.0
0.6
0.8
1.2
at altitudes
km decreased
by about 9%differences
(ref. 13) inare
response
Thetosharp d
the summer
polar35–50
stratopause.
These temperature
3
1.0
1.4
1.2
1.6
(Fig. 2b) is c
increasing
of active
species.
Since about
qualitatively
similarconcentrations
to, but about 50%
larger chlorine
than, those
estimated
1.4
1.8
1.6
40
1.8
with
by ref. 2000,
12 withhowever,
an idealized
a full climate
model, possibly
the forcing
trend inin chlorine
has reversed
and ozone
hasthe dom
2.0
These
owing stopped
to the broader
spectral
resolution cooling
imposed by
andgreenhouse
the lack ofgases
declining.
Stratospheric
has losses
through pho
ozone–temperature
feedback
in thatto
model
version.trend reversal by slowing
probably
also
contributed
the
ozone
10
2.0
1.6
14
nates
The the
verychemical
different scenarios
by theittwo
spectral
setsperiod
. Over
thedata
short
ofthe loss
reactionsproduced
that destroy
athese
self-healing
1.8
they
mightstudy
be distinguishable
in observational
A
30suggestthe
present
it is not possible
statistically torecords.
differentiate
1.6
to lower
multiple
regression
has or
been
carried
out of
deseasonalized
1.4
factors
from analysis
each other,
from
any solar
influence.
Nevertheless,
it leve
30
(SeeinSupplem
monthly
meanlikely
ozonethat
datathe
from
theisMicrowave
Sounder
(MLS)
seems
Sun
importantLimb
in the
apparent
decrease
–40
–20
0
20
40
instrument
onbelow
the Earth
Observing
System
(EOS)
Aura
satellite.
Four
Latitude (°N)
ozone
45 km
from 2004
to 2007.
The
change
in sign
near 45 To
kmassess t
sured
irradia
regression
indices
were
used: with
a constant,
two orthogonal
indices
is
also
more
consistent
the
modelled
response
to
the
SIM
specFigure 2 | Modelled difference in ozone between December 2004 and representing the quasi-biennial oscillation (which dominates ozone
13,15 another set o
of
the
tral
variations
than
to
the
Lean
spectra.
Previous
analyses
December 2007. Estimates of the percentage difference (2004–2007) in
SOLSTICE t
variability in the tropical stratosphere)13 and a solar index consolar signal in ozone, averaged over approximately 2.5 solar cycles
zonal mean ozone concentration (labels on contour lines in per cent)
There are dif
structed from SIM data integrated over 200–400 nm. Motivated by
produced by the model using solar spectra from the Lean model (a) and SIM/
(1979 to 2005 or 2003), have not shown this structure. This suggests
fields but the
the model results (Fig. 2), we chose two spatial regions, both spanSOLSTICE data (b).
the declining phase of solar cycle 23 is behaving differently
to
stratosphere
tropics, one at altitude 10–6.8 hPa, where the model predicts
2
Jan. 28, 2014 SORCE Science Mee5ng ning thethat
previous
solar2004–2007,
cycles or possibly
the solarhPa,
cyclewhere
exhibits
tosphere. Th
the largest
difference
and onethat
at 0.68–0.32
the different
itsvalues.
ascending
and
descending
Fig. 1) are also very different, with the Lean data set showing temperaresponse but
model behaviours
shows largestduring
negative
Figure
3 shows
the rawphases.
data and
To
understand
the
different
spatial
structures,
and itmagnitudes,
tures 0.3–0.4 K greater in 2004 than in 2007 at the top of the model
the of
impact o
the fits reconstructed from the four regression components;
also
the
ozone responses
we consider
photochemical
domain, whereas the SIM data set produces a peak warming of 1.8 Kshows
at
the radiative
(in modelled
red) the derived
solar component,
which
is statisticallyprocesses.
The atsharp
in ozone
above
45 km
with
thelevels
SIM spectra
the summer polar stratopause. These temperature differences are
significant
.95%decrease
at the upper
levels and
.99%
at the
lower
The respon
0.6
3
0.7
0.7
3
Decreased ozone at solar max. Approximate altitude (km)
MLS ozone regression b
O3 anomalies (%)
5
Pressure (hPa)
2
242
ed as numDecember
ce between
IM. This is
cal models,
nd area, as
e SIM data
a factor of
e radiation,
r empirical
aviolet than
rent in the
m, are indeance Com. The data
hPa
2005
2007
Merkel et al. 2011 GEOPHYSICAL RESEARCH LETTERS, VOL. 38, L13802, doi:10.1029/2011GL047561, 2011
Simulate Irradiance SORCE Difference (%)
NRLSSI Difference
Changes in W(%)
ACCM WACCM 10-2
Active 2004
Quiet 2007
Active 2004
Quiet 2007
Active 2002/2003
Quiet 2008/2009
Difference (%)
o
N-15
S
Avg.
15o–
25S 25N spectral irradiance variability
10
on middle atmospheric ozone
Pressure (hPa)
-1
10
TIMED/SABER Day
Aimee
W. Merkel,1 Jerald
W. Harder,1 Daniel R. Marsh,2 Anne K. Smith,2
O3 emission measurements 1
1
Juan
M. Fontenla,1 and Thomas N. 10Woods
from the 9.6μm channel – 80
0
Received Version 24 March 12011;
18 May 2011; published 2 July 2011.
2
.07 revised 15 May 2011; accepted
10
60
SABER
02/03-08/09
04-07
Model w/NRLSSI
Model w/SORCE
40
20
Approximate Altitude (km)
3-­‐D fully coupled chemistry, radia5on and The
impact
ofmsolar
dynamics global odel. Compare Simula<ons with SABER SABER
Difference
(%) Observa<ons Pressure (hPa)
Approximate Altitude (km)
[ 1 ] This study presents the impact of solar spectral 2009]. The SIM irradiance measurements have an accuracy
3
irradiance (SSI) variability on middle 10atmospheric
of -50better0 than
-50
0 ozone
50
50 2%
-10 -8 -6 Based
-4 -2 0 on
2 4 the
-60 [Harder
-40 -20 0 et
20 al.,
40 602010].
over the declining phase of solar cycle 23.
differentNRLSSI comparison
independent SIM
long‐
Δ O3 (%)
a) Two
Latitude
d) channels,
b)
Latitude of thec)two Latitude
model types of spectral forcing are applied
to the Whole term uncertainty in the 200–300 nm region is ∼0.5–0.1%,
80
10-2
Atmosphere Community Climate Model (WACCM) to from 310–400 nm it is ∼0.2–0.05%, and in the 400–1600 nm
SABER O3 average vmr between periods
SABER ozone: 2-­‐yitr isrunning annual mean trends simulate
the ozone
response
of
quiet and
range
better than
0.05%.zonal Currently
there
is no direct
10-1
60
25°S –
2
5°N high solar activity. One scenario uses the solar proxy validation through independent irradiance observations that
0 a
0.2 .08 r e c o n s t r u c t i o n s m o d e l f r o m t h e 10N
v ato l 0R
e smeb a r c h have a physically
et al.,
Day based degradation correction [Harder Night Night
40
60-­‐66km Laboratory (NRLSSI), and the other is based on SSI 2009, auxiliary material]. We present a modeling study on
1
10
observations from the Solar Radiation
and Climate the impact of SORCE SSI variability on middle atmospheric
Experiment (SORCE). The SORCE observations show 3– ozone.
20
2
5 times more variability in ultraviolet 10
(UV)
radiation than
[3] Up to the present time, most climate and solar cycle
predicted by the proxy model. The NRLSSI
forcing had related modeling studies have relied on solar input spectra
2 to 4 mb 3
10
long‐term
of -10
solar
minimal impact on ozone, however, the
UV that
-50are scaled
0
50 by -60
-40 -20 0proxies
20 40 60
-50 higher
0
50
-8 -6activity
-4 -2 0 such
2 4 as
38-­‐44km the
F10.7
radio
flux
[Marsh
et
al.,
2007;
Randel
and
Wu,
variability from SORCE induced a 4% reduction
in
ozone
e)
f)
g) Latitude
h) Δ O3 (%)
Latitude
Latitude
concentration above 40 km at solar active conditions. The 1999] or an equivalent Mg II core‐to‐wing [Soukharev
model result is supported by 8 years (2002–2010) of and Hood, 2006]. A more dependable solar model emDescending phase of SC23 based on combined 3 ploys
solarMirradiance
reconstructions
ozone observations from the Sounding of the JAtmosphere
an. 28, 2014 SORCE Science ee5ng usin g B roa dba nd Emiss ion Rad iometr y (SABER) sunspot and facular proxy indicators (predominately the
instrument. The SABER ozone variations have greater photometric sunspot index and the Mg II index) such as the
similarity with the SORCE SSI model simulations. The Naval Research Laboratory SSI model (NRLSSI) [Lean,
SSI Solar Forcing and Earth Atmospheric Response Understandably, this discovery revitalized the irradiance and atmospheric modeling communi5es. Lots of recent modeling ac5vity! Author Reference Model/Topic Haigh et al. Nature, 2010 IC2D model/SC Ozone Cahalan et al. GRL, 2010 GISS ModelE/Trop. Temp. Merkel et al. GRL, 2011 WACCM/SC ozone & TIMED SABER Ineson et al. Nature Geosci., 2011 HadGEM3/NAO Oberländer et al. GRL, 2012 EMAC-­‐FUB/Strat. temp   Radia5ve response Swartz et al. ACP, 2012 GEOS CCM/ Strat. Ozone & temp   Circula5on -­‐ NAO Wang et al. PNAS, 2013 WACCM/MLS & grnd based hydroxyl Shapiro et al. JGR, 2013 SOCOL/SC response Wen et al. JGR, 2013 GISS ModelE/Temp. response Ineson et al. (in prepara5on) HadGEM3/NAO/CMIP5 study, Maunder Minimum response Jan. 28, 2014 SORCE Science Mee5ng Modeling studies focusing on SSI implica5ons:   Photochemistry   “Top down” vs “Bojom up” 4 (a)
Compiled Modeling Results (b)
Fig. 10. Differences for January between 2004 and 2007 averaged over the tropics from 25◦ S to
NRLSSI
(solidKlines)
and
the SORCE/SIM
andGSORCE/SOLSTICE
data (dashed lin
In the Ermolli using
et al. the
2013 ACP data
paper, atja M
ajhes (GEOMAR, ermany) compiled −1
Kelvin per day (K day ) and (b) the temperature in Kelvin (K). For details please see text.
the results of these modeling studies. Ozone response (max-­‐min) to SORCE and NRLSSI 1
GEOSCCM NRLSSI
GEOSCCM SORCE
IC2D NRLSSI
IC2D SORCE
SOCOL NRLSSI
SOCOL SIM
SOCOL SOLSTICE
WACCM NRLSSI
WACCM SORCE
WACCM* NRLSSI
MMM NRLSSI
MMM SORCE
60
GEOS CCM 55
IC2D 50
SOCOL WACCM 45
MMM 40
35
10
Height (km)
Pressure (hPa)
0.1
30
25
20
100
−10 −8 −6 −4 −2
0
2
4
6
8
10
Percent zone ariability due into solar Fig. 11. As Fig.
10 butofor
the vozone
response
percent.
Jan. 28, 2014 SORCE Science Mee5ng the tropospheric AO and NAO already respond to lower UV
Even though the solar
the 11 yr solar cycle is b
should note here again t
ergy source of the clim
activity in the past few
solar cycles) and its poss
evolution has attracted th
public (e.g. Lockwood e
Jones et al., 2012; Rozan
4.3
Discussion of CCM
We described the impact
represent the lower and
cle estimated variations,
depicted in CCM simula
provide the standard dat
past and future (e.g. SPA
2012). The atmospheric
5 dard data set is compared
estimate to understand n
Ques5ons and Debate Now there is sugges5ve evidence of this surprising SC signal in mesospheric ozone measurements that you can only get if more UV variability is incorporated into atmospheric models. Lots of ques5ons and debate. Solar ques5ons: -­‐ Integrity of the SORCE dataset (Degrada5on correc5ons?) -­‐ Why do previous measurements at these wavelengths disagree with SORCE? -­‐ Can we use a solar spectrum scaled by a variability proxy (TSI, MgII, Lyman α, F10.7) as a standard to characterize the sun in atmospheric models? -­‐ Do all wavelengths vary the same way as TSI? -­‐ How good do the solar measurements need to be? Gaps in 5meseries? -­‐ Does this variability only pertain to SC23-­‐24 or has it been there all along? Atmospheric ques5ons: -­‐ Integrity of the ozone measurements. Why haven’t we seen this signal before? -­‐ Why is the mesospheric signal out of phase with solar cycle and different than stratospheric ozone? -­‐ Is this a special solar cycle? Is it in previous measurements? -­‐ What does this mean for the modeling community? Is a SSI proxy good enough for atmospheric modeling studies? Are we missing important SC variability? Jan. 28, 2014 SORCE Science Mee5ng 6 Interes5ng Conundrum   The modelers -­‐ want the solar physicist to give them something to put in their models. Most models have been upgraded to included SSI on a daily cadence.   The solar physicist -­‐ want the modelers to tell them how good they need to measure the Sun. •  Good enough is different depending on the model, atmospheric region studied, type of model (photochemistry, radia5ve).   Work in progress: We are having a workshop next month with the NCAR folks to discuss this very thing. Jan. 28, 2014 SORCE Science Mee5ng 7 Understanding Mesospheric Ozone Variability Top of the atmosphere ozone~ 1% of total column –  Dominated by photochemistry. –  Photochemical life5me is hours. –  Strong diurnal component. Local 5me is important for solar cycle analysis. –  More UV causes more loss of ozone at solar maximum. Loss due to photolysis and cataly5c cycles with OH and H. Mesospheric ozone influenced by the solar radia5on in the 200-­‐300nm band. (produc5on and loss) Herzberg Con5nuum Hartley bands Herzberg Con5nuum (200-­‐242nm) J2 rate: O2 Photolysis  O3 produc5on Hartley Bands (200-­‐300nm)– J3 rate -­‐ O3 Photolysis  O3 loss Jan. 28, 2014 SORCE Science Mee5ng 8 TIMED/SABER Ozone – 12 years of ozone data We can now look at the ascending phase of solar cycle 24. Did the mesosphere respond? •  Data now spans from 2002 – 2013 (12 years of data) •  Recently updated to Version 2 – Reprocess all results •  9.6μm channel – O3 emission measurements •  Version-­‐2 data validated by A. Smith (NCAR) 2012. Known systema5c bias compared to other ozone measurements. Bias is constant over 5me, so does not influence differences. Jan. 28, 2014 SORCE Science Mee5ng 9 SABER Ozone Time Series – Yearly average Mesosphere – Out-­‐of-­‐phase 9.6 day +/-25deg Meso. yearly average .2-.08mb
Equatorial average 5me series O3 VMR
0.87
0.86
~4% 0.85
0.84
0.83
0.82
2002
2004
2006
2008
2010
2012
2014
Stratopause – Licle varia5on +/-25deg Stratpause yearly average 1-2mb
1 mb 48 km O3 VMR
5.6
5.5
5.4
5.3
5.2
2002
2004
2006
2008
2010
2012
Stratosphere – in-­‐phase 2014
9.6 day +/-25deg Strat. yearly average 2-4mb
10.0
9.8
O3 VMR
SABER O3 Average vmr 0.2 to 0.08 mb 60-­‐66km 0.88
2 to 4 mb 38-­‐44km 9.6
9.4
9.2
9.0
2002
~5% 2004
2006
Jan. 28, 2014 SORCE Science Mee5ng 2008
2010
2012
2014
10 SABER Ozone Time Series Mesosphere – Oyearly
ut of average
phase .2-.08mb
9.6 day
+/-25deg Meso.
0.2 to 0.08 mb 60-­‐66km 0.88
0.10
0.84
0.08
0.83
0.82
2002
0.06
2004
2006
2008
2010
2012
2014
0.04
0.02
0.00
2002
Stratosphere in phase 9.6 day
+/-25deg Strat.– yearly
average 2-4mb
10.0
2004
2006
2008
Year
2010
2012
2014
2 to 4 mb 38-­‐44km 9.8
Courtesy of Jerry Harder O3 VMR
Fractional Irradiance Change
O3 VMR
Irradiance Variability 0.87
0.86
Ac5ve Area Network -­‐PSPT 0.85
240-­‐260nm SSI Proxy basedSIM on the
Active Network Area, SIM 240-260nm
9.6
9.4
9.2
9.0
2002
2004
Jan. 28, 2014 SORCE Science Mee5ng 2006
2008
2010
2012
2014
11 20
4-­‐component regression, deseasonalized 5me series at each pressure level. Solar, Annual, 2-­‐QBO Solar Coefficient Mesosphere out-­‐of-­‐phase Mesosphere
ozone–variation,
0.08 mb 66 km
Percent ozone variation due to solar
-10
Residual ozone variation
regression fit
solar component
Polynomial fit
10
-5
0
5
10
70
0
-10
-20
2002
2004
2006
2008
Year
2010
2012
2014
Stratosphere in-­‐phase Stratosphere
ozone–variation,
5.5 mb 36 km
15
Residual ozone variation
regression fit
solar component
Polynomial fit
10
60
50
1.0
approximate altitude (km)
0.1
Pressure levels (mb)
Residual ozone variation in percent
Residual ozone variation in percent
SABER Regression Analysis 40
5
0
-5
10.0
30
-10
-15
2002
-1.5
2004
2006
2008
Year
2010
2012
2014
Jan. 28, 2014 SORCE Science Mee5ng -1.0
-0.5 0.0
0.5
Solar coefficient
1.0
1.5
12 Ozone measurements in previous solar cycles JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 111, D20314, doi:10.1029/2006JD007107, 2006
Ques5on: Why have we not seen this mesospheric ozone behavior in Figure 7. Same format as Figure 5 but as calculated from the UARS HALOE zonal mean time series
Full
(1992
–2003).
previous solar cycles? Article
Click
Here
for
for each of the three satellite data sets were averaged over
the 25!S to 25!N latitude band and the averaged time series
were then analyzed using the same multiple regression
model (1). Results are plotted in Figure 8 as vertical profiles
of the solar minimum to maximum regression coefficient
with 95% confidence error bars for each data set. It is seen
that the basic features of the SBUV(/2) tropical ozone
Reason response (Figure 8a) are also present in the SAGE II
response (Figure 8b) and in the HALOE response
(Figure 8c). In the lower stratosphere, a statistically significant positive response is obtained with a percent amplitude
that increases with decreasing altitude (with the exception of
the lowest level for the SAGE II data, which may be less
reliable). In the middle stratosphere, the 2-s error bars
Solar cycle variation of stratospheric ozone: Multiple
regression analysis of long-term satellite data sets
and comparisons with models
Regression analysis results B. E. Soukharev1 and L. L. Hood1
Remsberg et al., 2SOLAR
008 CYCLE IN OZONE AND TEMP
REMSBERG:
Historical ozone data comprised of occulta5on [ ] Previous
multiple
of the solar cycle
stratospheric
data or regression
only analyses
measure up variation
to 5of0km. ozone are improved by (1) analyzing three independent satellite ozone data sets with
D22304
Received 22
January 2006; revised 28 June 2006;
accepted 24 July 2006; published 31
October 2006.
SAGE SBUV HALOE 1
1991-­‐2005 lengths extending up to 25 years and (2) comparing column ozone measurements with
ozone profile data during the 1992–2003 period when no major volcanic eruptions
occurred. Results show that the vertical structure of the tropical ozone solar cycle response
has been consistently characterized by statistically significant positive responses in the
upper and lower stratosphere and by statistically insignificant responses in the middle
stratosphere (!28–38 km altitude). This vertical structure differs from that predicted by
most models. The similar vertical structure in the tropics obtained for separate time
intervals (with minimum response invariably near 10 hPa) is difficult to explain by
random interference from the QBO and volcanic eruptions in the statistical analysis. The
observed increase in tropical total column ozone approaching the cycle 23 maximum
during the late 1990s occurred primarily in the lower stratosphere below the 30 hPa level.
A mainly dynamical origin for the solar cycle total ozone variation at low latitudes is
therefore likely. The amplitude of the solar cycle ozone variation in the tropical upper
stratosphere derived here is somewhat reduced in comparison to earlier results. Additional
data are needed to determine whether this upper stratospheric response is or is not larger
than8.model
estimates.
Figure 12. Profiles of the 11-year and the SC-like,
Figure
Solar cycle
ozone regression coefficients (expressed as percent change from solar minimum
The solar signal is “washed out” due to the mixing of source and loss mechanisms in a photochemical dominate region of the atmosphere. (sunrise and sunset) to maximum) for (a) the SBUV(/2) data set, (b) the SAGE II data set, and (c) the HALOE data set. The
maximum
minus
minimum
responses (in percent) for the
Citation: Soukharev, B. E., and L. L. Hood (2006), Solar cycle variation of stratospheric ozone: Multiple
regression
analysis
of longregression coefficients were obtained by applying equation (1) to 3-month time series averaged over the
tropical
to
subtropical
ozone
from HALOE. The solid curve
termto satellite
data sets
25!S
25!N latitude
band.and comparisons with models, J. Geophys. Res., 111, D20314, doi:10.1029/2006JD007107.
is the model result for 5!N from Brasseur [1993].
9 of 18
formation in the stratosphere [Heath and Schlesinger,
1986; Viereck and Puga, 1999].
3.3. Solar Cycle (or SC-Like) Responses of HALOE
[2] The observed solar cycle variation of stratospheric
Ozone
and11-year
Temperature
of the
ozone re[3] Observational estimates
ozone is a key constraint on climate models that include sponse as a function of altitude and latitude have been
40] In order to obtain estimates of the actual response of
[
solar variability as a forcing mechanism and that account for reported by a number of analysts based mainly on Solar
atmospheric ozone and temperature to a proxy for the direct
the existence of the stratosphere [Haigh, 1994, Jan. 1996;
28, Backscattered
2014 SORCE Ultraviolet
Science M
ee5ng (SBUV)
forcing data
fromand
the Stratospheric
UV flux, their sets of responses from
Shindell et al., 1999; Rind, 2002]. It is also important to Aerosol and Gas Experiment (SAGE) I and II data using
Figures 9a and 11a have been adjusted by the amount that
determine and understand the solar cycle variation of ozone multiple regression methods they
were not
exactly
[Chandra,
1991;
Hoodinetphase
al., from Figures 9b and 11b.
so that anthropogenic trends, including possible evidence 1993; Chandra and McPeters,
Those1994;
adjustments
were made
McCormack
andby multiplying the responses
for an ozone ‘‘recovery,’’ can be more accurately evaluated Hood, 1996; Wang et al., 1996].
by the
cos analyses
[2p p/11],ofwhere p is the deviation of
Forfactor,
example,
1. Introduction
are model/HALOE
ozone responses wit
in part, to not accou
in the present HAL
McCormack et al.,
[42] Lee and Smit
SBUV data for a s
SAGE II are simila
Figure 9a, at leas
latitudes; their resu
though, especially
results from SAGE
[2006]). Lee and Sm
the SC response, b
phases of the QBO
with the QBO in
responses in the s
hemispheres, but t
QBO indexes that th
for the present HAL
QBO and its asso
accounted for to fir
those terms in the H
were able to simulat
20 to 30 hPa, but
response at about 5
lations reported by M
positive ozone respo
period of 1979 to 2
1950 to 2003. It is
spheric ozone respon
13 to solar cycl
related
[43] The SC-like,
for ozone (in perc
summarized in Tab
and for three distinc
Solar Mesosphere Explorer – 1982 -­‐ 1989 •  UV ozone channel (Rusch et al. 1984) Solar irradiance measurements. (Rojman et al. 1982) •  SME covers parts of solar cycle 21-­‐22 -­‐ Good ozone measurements between 1982-­‐1986 •  Day5me (3pm) limb profiles of ozone with good global coverage. •  SME ozone measurements are analyzed consistent with SABER analysis. (Woods and Rojman 1997) Jan. 28, 2014 SORCE Science Mee5ng 14 Solar Irradiance variability 2 decades apart LISIRD site Courtesy of Jerry Harder Jan. 28, 2014 SORCE Science Mee5ng 15 SME Ozone Time Series SME yearly average vmr SME yearly ean 5me series SME m
yearly
average
70
0.5
2.0
1.0
0.2
2.0
50
0.77
1982
0.5
0.2
45
-50
0
latitude
0.79
0.78
1983
1984
1985
1986
1987
50
SME irradiance variability 240-260nm
0.07
Delta Irradiance (W/m^2)
0.5
1.0
55
0.2 to 0.08 mb 60-­‐66km 0.80
O3 VMR
60
1.0
approximate altitude (km)
65
0.5
0.2
Pressure level (mb)
10-1
100
0.81
0.2
0.2
0.06
0.05
0.04
0.03
0.02
0.01
0.00
1982
Jan. 28, 2014 SORCE Science Mee5ng 1983
1984
1985
Year
1986
1987
1988
16 Solar Coefficient Mesosphere 60 km 0.2 mb 60km
Mesosphere
ozone–variation,
Percent ozone variation due to solar
-5
0
5
4
70
2
0
Residual ozone variation
regression fit
solar component
Polynomial fit
-4
1982
1983
1984
Year
1985
1986
1987
Stratopause– 48 km 1 mb 48km
Mesosphere
ozone variation,
4
60
55
2
approximate altitude (km)
65
0.1
-2
Pressure levels (mb)
Residual ozone variation in percent
Residual ozone variation in percent
SME Regression 50
1.0
0
-2
Residual ozone variation
regression fit
solar component
Polynomial fit
-4
1982
1983
1984
Year
1985
1986
45
-1.5
-1.0
1987
Jan. 28, 2014 SORCE Science Mee5ng -0.5
0.0
Solar coefficient
0.5
1.0
17 SME compared to SABER Regression – solar component 0.10
Day Solar Variation
0.08
SME UV - SC 21
SORCE - SC 23-34
SABER
0.04
SME UV
0.1
0.00
-0.02
-8
-6
-4
-2
0
2
Years from solar minimum
4
6
Ozone yearly 5me series 1980
1982
1984
1986
1988
1990
Pressure levels (mb)
0.02
70
60
50
1.0
1.04
approximate altitude (km)
0.06
Normalized ozone variability
Fractional Irradiance Change
Irradiance Variability 240-­‐260nm 40
1.03
10.0
30
1.02
-15
1.01
1.00
2002
-10
-5
0
5
10
Percent Solar variation
15
SME UV - SC 21
SABER 9.6 - SC 23-34
2004
2006
2008
2010
2012
2014
Jan. 28, 2014 SORCE Science Mee5ng 18 Summary •
Sugges5ve evidence that UV variability in the 240-­‐260nm range and mesospheric ozone from SC 21 are consistent with SC 23-­‐24. Has it been there all along? Further evidence that this signal is real. •
Mul5ple modeling studies show that increased UV variability as observed by SORCE (both SIM and SOLSTICE) helps to resolve differences between modeled ozone and observa5ons in the mesosphere. •
The UV variability is probably somewhere in between NRLSSI and SORCE, however it is apparent that the atmosphere is sensi5ve to this difference. When compiling SSI proxy model please consider that this type of variability in the UV majers in the mesospheric photochemistry. •
Need to approach the issue from both direc5ons. Atmospheric modelers and solar physicists/modelers need to work together to constrain this variability . The atmospheric modelers can perform case studies to fine tune the response to different solar variability but this can’t be used to validate the solar, can only be used as a guideline. Wavelength dependent. •
Importance of the con5nua5on of mesospheric ozone measurements in the future. Jan. 28, 2014 SORCE Science Mee5ng 19 Thank You!
Jan. 28, 2014 SORCE Science Mee5ng 20 9.6 day +/-25deg Meso. yearly average .3-.08mb
1.08
O3 VMR
1.06
Version 1.07
Version 2
1.04
1.02
1.00
2002
2004
2006
2008
2010
Jan. 28, 2014 SORCE Science Mee5ng 2012
21 Analyze SABER as if Occulta5on Experiment Day Night Par5al Day-­‐Night Annual Mean Difference 25S-­‐25N -­‐ When SABER is analyzed with only measurements taken at “occulta5on” local 5mes: Solar signal is washed out and response is more similar to night results Jan. 28, 2014 SORCE Science Mee5ng 22 Analyze WACCM at Occulta5on local 5mes Modeled ozone results Day Night Par5al Day-­‐Night When WACCM is analyzed with only measurements taken at “occulta5on” local 5mes: Solar signal is washed out and response is more similar to night results. Confirms results from SABER. Jan. 28, 2014 SORCE Science Mee5ng 23