Dec 20, 1993 - Earlier theoretical studies [Frederick, 1977; Callis and. Nealy, 1978' Penner ...... stratosphere, Ann. Geophys., 6, 417-424, 1988. Donnelly, R. F. ...
JOURNAL
OF GEOPHYSICAL
RESEARCH,
VOL. 98, NO. D12, PAGES 23,079-23,090, DECEMBER
20, 1993
The Response of theMiddleAtmosphere to Long-TermandShort-Term SolarVariability' A Two-DimensionalModel GuY BRASSEUR
NationalCenterfor Atmospheric Research, Boulder,Colorado
A two-dimensional chemical-dynamical-radiative modelof the middleatmosphere is usedto investigate
thepotential changes of temperature, ozone,andotherchemical constituents in response to variations in the solarultravioletflux, associated with the solarrotation(27 days)andthe solarcycle(11 years). The model
reproduces satisfactorily theresponse (amplitude andphase)to the27-dayforcingof ozoneandtemperature in the stratosphere but doesnotproperlyexplainthe ozoneandtemperature responses of opposite sign observed near70 km altitude.The changein the ozonecolumnabundance associated with the 27-daysolar forcingis estimated to belessthan0.5%. Variations in middleandupperatmospheric ozoneconcentrations andtemperatures inducedby solarvariations on the ll-year time scaleare not negligiblecompared to changes produced asa resultof humanactivities overthesameperiodof time. Thecalculated change in the ozone column abundancefrom solar minimum to solar maximum conditionsis of the order of 1.1-1.3% in the
tropicsandincreases withlatitude,especially in winter,to reachupto 1.5-1.7%in thepolarregions.There are largeunexplained differences betweenthe calculatedand observedstratospheric responses.For example, in thephotochemically controlled regionof theupperstratosphere, themodelseems to underestimate the ozoneresponse by morethana factor2. In addition,the negativetemperature andozoneresponses observed in thelowerstratosphere cannotbereproduced by themodel. Potential dynamical feedbacks exist, but reliabledatasetscoveringa periodlongerthanonesolarcyclehaveto becomeavailablebeforethis problemcan adequatelybe addressed.
200 mn and 6-7% at 205 nm. The variability over the. full wave.length interval ranging between 120 and 300 nm. as It has been recognized for a long time that the estimated by Rottman [1988], is shown in Figure 2a. These electromagnetic flux emitted by the Sun is not perfectly values remain somewhatuncertainbecauseof potential drift in constant in time, but that it slightly varies, especially at the solar detectors over the 7-year period. Note also that the shortest wavelengths, on different time scales (see, e.g., Lean SME values cover only a fraction of an l 1-year solar cycle. [1987], Donnelly [1988, 1991], Simon [1989], Simon and Shown in Figure 2b is the 27-day solar variability derived for Tobiska [1991] for reviews of the subject). Generally, the two distinctperiodsof the solar cycle (active Sun in 1982, and most pronounced variability is associated with the l 1-year quiet Sun in 1985, respectively)and coveringthe 120-300 nm solar cycle. In addition, an observed short-term variation in spectral region. As suggestedby Figures2a and 2b, the the solar flux results from the uneven distribution of active variability decreasessubstantiallywith increasingwavelength. regions on the Sun which rotates with an apparentperiod of 27 In addition, a sharpdrop in the solar variation is visible at 208 days. Figure 1 shows the temporal evolution of the solar flux nm (aluminum edge in the solar spectrum)and a relatively large at Lyman-ct (121.6 nm), in the region of the 0 2 Schumannamplitude in the solar signal is noticeable at 280 nm Runge bands (180-200 nm) and in the region of the 0 2 (magnesiumII). Herzberg continuum (205 nm), deduced from the observations Earlier theoretical studies [Frederick, 1977; Callis and by the Solar Mesosphere Explorer (SME) between 1982 and Nealy, 1978' Penner and Chang, 1978; Brasseurand Simon, 1988 (P. C. Simon and G. Rottman, personal communication, 1981' Garcia et al., 1984; Callis et al., 1985; Eckman, 1986; 1989). The solar flux at Lyman-ct affects mainly the Brasseur et al., 1987; Wuebbles et al., 1991] have suggested atmosphereabove 75 km altitude, while the flux at 180-200 that periodicchangesin the solarflux associatedwith the solar nm is mainly absorbed by molecular oxygen in the cycle and the rotation of the Sun should affect the chemical mesosphere. The solar radiation between 200 and 240 nm is compositionand thermal structureof the stratosphereand the primarily responsible for the formation of ozone in the mesosphere. Because reliable measurementrecords for the stratosphereand hence the flux at the particular wavelength of middle and upper stratospherespan only slightly more than 205 nm is often used as an index to representthe intensity of one solar cycle, statistical evidence for the response of the the solar forcing on the stratosphere. As suggested by middle atmosphereto the 1i-year solar variation cannot yet be Figure 1, the variation between the early 1980s (solar 1. INTRODUCTION
maximum condition) and the mid 1980s (solar minimum condition) is of the order of 50-60% at Lyman-ct, 5% at 180-
established with full confidence. Nevertheless, analyses of total ozone column abundancesmeasuredby the Nimbus 7 total
ozone mapping spectrometer (TOMS) [Chandra,
1991'
Paper number93JDO2406.
Stolarski et al., 1991' Hood and McCormack, 1992], of upper stratosphericozone densitiesobservedby the Nimbus 7 solar backscatteredultraviolet (SBUV) instrument (L. L. Hood et al., Quasi-decadal variability of the stratosphere' Influence of
0148-0227/93/93JD-02406 $05.00
long-termsolar ultravioletvariations,submittedto Journal of
Copyright1993 by the AmericanGeophysicalUnion.
23,079
23,080
BRASSEUR: SOLARVARIABILITYANDMIDDLEATMOSPHERE COMPOSITION SME SOLAR
FLUX 1.97
-,.
, •
1.0•• m
0.83
4
•
z
o
o•
3 I 1982
I 83
I 84
I 85
I 86
performedearlier by Brasseuret al. [1987] usinga simpleonedimensionalmodel. In the presentpaper a two-dimensional(2D) model with a formulationof coupledchemical,radiative and
dynamical processesaccountsfor potential feedbackswhich could not be simulated by earlier one-dimensional (l-D) models. In a secondpart of the presentstudy, the responseof the atmosphereto long-term changesin the solar forcing will be discussedonly briefly, since a recent paper by Huang and Brasseur [1993] addressesthis question. The paper is organizedin the following way: In section2 the model used in this study will be describedand the adopted variation in the solar actinic flux will be presented. In section
I 87
be examined. First, the response(amplitude and phase) of the middle atmosphere to the 27-day solar forcing will be considered. This part of the work is an extensionof the study
88
Fig. 1. Solar actinic flux measuredby the Solar Mesosphere 3 the responseof the middle atmosphereto the 27-day solar Explorer at different wavelengths (Lyman-tz, 180-200 nm, variability will be discussed. Section4 will presenta similar 205 nm) between 1982 and 1988. (From G. Rottrnan and P. C. analysis, but for the l 1-year forcing. Differences between Simon, personal communication, 1990.) theoretical predictions and observational estimates will be stressed. Finally, the key conclusionswill be summarizedin section
AtmosphericSciences,1993) (hereinafterreferredto as (Hood et al., submittedmanuscript))and of temperaturesarchived by the NOAA National Meteorological Center (NMC) (Hood et al., submittedmanuscript)have revealed a likely correlation with solar cycle. Data from the StratosphericAerosol and Gas Experiment II (SAGE II), obtainedover a period of increasing solar activity (after 1985), also suggest a positive relation between ozone in the upper atmosphereand solar ultraviolet flux [Keating et al., 1993a]. Observationsof stratospheric temperaturesmeasuredby radiosondes[Labitzkeet al., 1986; Angell, 1988, 1991], rocketsondes[Angell, 1991] and lidar [Chanin et al., 1987; Hauchecorne et al., 1991], as well as measurements of total ozone by ground-based Dobson instruments [Angell, 1989; World Meteorological Organization, 1988] also suggesta possibleresponseto the radiative forcing associatedwith the solarcycle. The effectson the dynamics of the middle atmosphereremain a matter of discussion.
The
observational
evidence
of
reliable
statistical
evidence
has
5O
I
been
I
I
I
I
I
SME 1982-1986
--
205 nm lO
.•_ n,.., Mgll •
(a)
A.2
an association
between the l 1-year solar cycle, the quasi-biennialoscillation (QBO) and the interannualatmosphericvariability reported by Labit•e and van Loon [1988] was questionedby Baldwin and Dunkerton [1989] and by Satby and Shea [1991] on statistical groundsandhasnot yet beenproperlyexplained. Kodera et al. [1991] showed that the observed temperature anomalies reportedin winter by Labitzke and van Loon [1988], including their dependence on the phase of the QBO, could be qualitatively reproducedin a general circulation model only if an unrealistic change in the stratosphericheating rate of at least 30% was usedto simulatethe solar forcing. More
5.
100 120.
I
I
I
140
160
180
-
I I I I It .
200
220
240
260
280
300
SME
(b) lO
obtained
concerning the potential responseof the middle atmosphereto
short-term solar variations. The analyses by Gille et al. [1984], Hood [1984, 1986], Keating et al. [1985], Chandra
o
12o
14o
16o
18o
200
220
240
260
280
300
[1986] and Hood et al. [1991], based on satellite observations WAVELENGTH (nm) in the 1980s, have documented ozone and temperature Fig. 2. Variation (percent) in solar ultraviolet flux as a perturbationsat the 27-day period. The purposeof this paper is to provide a new theoretical function of wavelength derived from the measurementsby the estimate of the potential changesin the chemical composition Solar Mesosphere Explorer. (a) Variation between solar of the middle atmosphere in response to natural solar maximum and solar minimum conditions(11-year cycle); (b) variability. The focus will be on ozone in the stratosphereand variation over a solar rotation period (27 days) for active and mesosphere, but changes in temperature which affect the quiet Sun,respectively.(From Rottman[1988] and G. Rottman chemical reaction rates will also be considered. Two cases will and P. C. Simon (personal communication), 1990).
B RASSEUR: SOLAR VARIABILITY AND MIDDLE ATMOSPHERE COMPOSITION
23,081
solar maximum conditions, the flux is increased between 120
2. MODEL DESCRIPTION
The model which is usedin the presentstudy [seeBrasseur et al., 1990] is two-dimensional (latitude, altitude) and extends from pole to pole and from the surfaceto the mesopause. Its spatial resolutionis 5 degreesin latitude and 1 km in altitude. Chemical, dynamical, and radiative processes are treated
and 300 nm by the relative values shown as a function of wavelength in Table 1. Because the data deducedfrom satellite
observations remain approximate over the entire spectral domain, and particularly between 160 to 200 nm, where the sensitivityof the two spectrometerson board SME was low (G. Rottman, personal communication, 1993; see also Figure 2a), interaciively above 15 km altitude. Below this level, the we have adopted a somewhat idealized solar variability and distributions of the chemical species are calculated with have, for example, ignoredrapid fluctuationsof this variability prescribedtemperatureand winds. as a function of wavelength. In our study the peak-to-peak Dynamical equations are expressedusing the transformed variability associatedwith the l 1-year solar cycle is 68% at Eulerian mean formulation (TEM, also called the residual Lyman-• (P. C. Simon, private communication, 1990) and circulation) proposed by Andrews and Mcintyre [1976]. 6.6% at 205 nm. Between 208 nm (A1 edge) and 265 nm, the Momentum depositionand eddy diffusivities are parameterized variation adopted in standardcalculationsis 3%, although the as a functionof the zonal mean wind and temperaturefields. possibility for a variation as high as 5% is consideredin one accordingto Lindzen [1981] and Holton [1982] in the case of sensitivity test. For calculations dealing with the 27-day gravity waves. and accordingto Hitchman and Brasseur[1988] variability, the relative variation at all wavelengths is assumed in the caseof the Rossbywaves. The solar forcing is calculated to be half of that shown in Table 1 (standardcase). For the sake through an explicit spectral integration of the energy of simplicity, we have adopted a wavelengthdependenceof the deposited.while. below 65 km altitude, the infrared forcing is variability similar to that of the l 1-year cycle. The specific obtained from the radiative transfer code used in the NCAR caseof Lyman-• will be discussedlater in the paper. Because community climate model (CCM1) [Kiehl et ttl.. 1987]. Above the amplitude of the 27-day solar signal evolves as a function this level, where local thermodynamical equilibrium (LTE) of time, the atmosphericresponse(e.g., ozone and temperature conditions do not apply, rather than using a complex and changes)will be expressedrelative to a 1% change in peak-tocomputationallyexpensivenon-LTE code, a simple Newtonian trough solar variation at 205 nm. The corresponding cooling approachis adoptedinstead. amplitudeswill, in this case, be referred to as "sensitivities." The chemical scheme includes approximately 60 species belonging to the oxygen, hydrogen, nitrogen, chlorine and 3. OZONE AND TEMPERATURE RESPONSES TO THE 27-DAY bromine families, and 120 chemical and photochemical SOLAR VARIABILITY reactions. The solar actinic flux at the top of the atmosphere used for the calculation of the photolysis and shortwave The responseof ozone and temperature to short-term solar heatingrates is taken from Brasseur and Simon [1981] and is variability is calculated by applying to the solar flux a assumedto be representativeof solarminimum conditions. For sinusoidal perturbation with a peak-to-trough amplitude as
TABLE 1. AdoptedVariability in the SolarFlux (11-Year Cycle) as a Functionof Wavelength(120-300 nm) Peak-to-Peak
Wavelengths,nm Lyman-lx(121.6)
Variation, % 68.0
Variation
Relative
to 205 nm 10.3
Schumann-RungeContinuum 116-141
24.0
3.6
141-149
14.0
2.1
149-159
20.0
3.0
159-170
14.0
2.!
10.0
1.5
170-175
Schumann-Run ge Bands 175-180
10.0
!.5
180-182
14.0
2.1
182-185
10.0
1.5
185-191
9.0
!.4
191-200
7.6
1.2
200-202
6.6
1.0
Herzberg Continuum 202-208
6.6
1.0
208-243
3.0(5.0)*
0.45 (0.76)*
243-267
Hartley Band 3.0(5.0)*
0.45 (0.76)*
267-270
0.6
0.09
270-278 278-282 282-303
2.0
0.30
*High value adoptedwhen specifiedin the text.
6.0
0.91
0.8
0.12
23,082
BRASSEUR:SOLARVARIABILITYANDMIDDLEATMOSPHERE COMPOSITION
specified insection 2. Thesolution of a linearized equation for an ozone perturbationsuggeststhat if the forced perturbation actsonly on the 02 photolysisrate (ozoneproduction)and if the temperaturefeedbackis ignored,the amplitudeof the ozone signal is proportionalto
AO3
&/o2•:
03
1986]. In the upper stratosphere, where an increase in temperature tends to enhance the ozone destruction rate, the ozone amplitudecalculatedwith temperaturefeedbackincluded is smaller than if this feedback is ignored. The phase lag is also modified and can even become negative near the stratopause (ozone response leads the solar signal). This behavior of the ozone/temperaturesystem has been discussed by Hood [1986], Brasseur et al. [1987], and Keating et al.
[1987]. Becausethe phaselag of the temperaturecalculatedby
model of Brasseuret al. [1987] was and the phase (time lag of the responserelative to the solar the one-dimensional approximatelya factor 2 smaller than the observedvalue, the calculatedozone responsecould not be accuratelycalculated. The two-dimensionalmodel, which is usedin the presentstudy,
signal) is given by
(i)=tan -1(car)
includes
a more
elaborate
radiative
scheme
and allows
for a
numberof potentialdynamicalfeedbacks. It is thus useful to revisit this question. Figure 3 shows as a function of altitude the calculated response of the equatorial temperature to a solar forcing correspondingto a peak-to-trough amplitude of 3.3% at 205 nm. The largest response(approximately 0.37 K) is found at responseis not straightforward(see section5). Furthermore, the stratopause,where the phase lag is equal to 4 days. The the time lag of the ozone signalrelative to the solar signalis amplitudeof the temperature responsedecreases towardslower necessarily short in the upper stratosphere, where the altitudesand becomesinsignificantbelow 30 km altitude. At photochemicallifetime of ozone is small comparedto the the sametime, the phaselag increasesand reaches6 days at 40 period of the solar forcing but is expectedto increasetowards km and 14 days at 30 km. lower altitudes. These conclusions need to be modified The responseof equatorial ozone, shown in Figure 4, is somewhatwhen other importantprocessesare included in the relatively complex. Below 70 km, a maximum in the ozone analysis. For example, in the mesosphereabove 65 km responseis predicted at approximately40 km altitude with a altitude, the most importantforcing is provided by changesin value of 1.2% for a solar variation of 3.3% at 205 nm. At this the photolysisof water vapor by Lyman-et, which produces altitude, the ozone responseis exactly in phasewith the solar hydroxylradicalsand effectivelyreducesthe concentration of forcing. Lower down, the ozone amplitudedecreasesand at 30 ozone at these levels. In addition, when the effect of km, it is only 0.4% with a phaselag of 5.5 days. In the upper temperaturefeedbackis taken into account,the amplitudeand stratosphereand lower mesosphere,the amplitudeof the ozone the phaseof the ozoneresponsedependon the amplitudeand response decreases and the ozone response leads the solar where co= 2 n/T (T being the period of the solar forcing) and '[ is the chemicallifetime of ozone [e.g., Hood, 1986; Brasseur et al., 1987]. Thus under these assumptionsthe amplitude of the ozoneresponseincreaseswith the periodof the solarsignal and an extrapolation of a 27-day response to an l 1-year
phase of the temperatureresponseand vice-versa[Hood,
km, the amplitude of the ozone variation increasessuddenly and reaches4% at 75 km. It also becomesexactly out of phase with the solar signal. This is the region of the atmosphere
TEMPERATURE
DIFFERENCE (K)
forcingby 3 daysat 50 km and 4.5 daysat 60 kin. Above 70
EQUATOR
6O
8O
03 DIFFERENCE (%)
EQUATOR
5O
g':'
',',i',•;14 aay;{ .._-:..., ....
;,'
uJ
4O
-'-• "•' :'
'0 '"" :
6O
,
3O
4O
2O
5/7
IIIII
Solar Min/
/
I I I I I I'• I I I I I I t 5/17
5/27
TIME
Solar Max
,,,,,
,,
III
6/6
Min Sol
I I I I i I I •1 I I I I I I I I I I I I [ I 20 Fig. 3. Calculated responsein the atmospherictemperature 5/7 5/17 5/27 6/6 (Kelvin) to the 27-day variation in solar ultraviolet flux. TIME Values are shown at the equatorbetween20 and 60 km altitude and cover the period May 7 to June6. The time lag (in days) Fig. 4. Sameas Figure3, but for the ozonedensity(expressed of thetemperature response •o thesolarsignal is indicated. in percent).
B RASSEUR: SOLAR VARIABILITY AND MIDDLE ATMOSPHERE COMPOSITION
where the photochemicaleffect of the solar flux at Lyman-rt becomes important. In Figures 5a-5d, the amplitudes(sensitivities)and phases of the ozone and temperatureresponsescalculatedat the equator are comparedwith an analysisof the data averagedover + 20ø latitude, provided by the Nimbus 7 limb infrared monitor of the stratosphere(LIMS), SBUV and stratosphericand mesospheric sounder (SAMS) instruments as well as by the Solar MesosphereExplorer [Keatinget al., 1987; Hood, 1986]. The calculatedozone sensitivityis of the order of 0.06 (percent03 by percentsolarflux at 205 nm) at 30 km, 0.25 at 35 km, 0.38 at 40 km and 45 km, 0.25 at 50 km, 0.20 at 55 km, and 0.15 at
23,083
opposedto 205 nm), sincethe role played by the solar flux in the 0 2 Herzberg continuumdiminishesabove the stratopause, while the influence of Lyman-rt becomesdominant. In this case, the ozone sensitivity (relative to Lyman-rt)is 0.01 (percent per percent) at 60 km, -0.14 at 70 km and approximately zero at 80 km. Again, the agreementbetween
model and observationis satisfactory. If, however,the ozone response above 65 km is expressedrelative to the 205-nm solar radiation,the sensitivityis too large by approximatelya factor of 1.5, suggestingthat the ratio between the relative solar variation at Lyman-rt and at 205 nm is approximately56 rather than 10, as adopted in the present calculation (see Table 1). A ratio of 5 is consistentwith the preliminary 60 km, while the calculatedtemperature sensitivityis 0.01 K/% at 30 km, 0.06 K/% at 40 km, and 0.12 K/% at 60 km. The analysis of solar measurements made by the SOLSTICE overallagreementwith the datais good,with a few differences instrumenton board the Upper AtmosphereResearchSatellite such as in the calculatedtemperaturesensitivitybelow 40 km (G. J. Rottman, personal communication, 1993). The phaselags for the temperature(Figure 5c) (14 days at 30 (Figure 5a). In Figure5b, the ozonesensitivityabove60 km is expressedrelative to the solar variability at Lyman-rt (as km, 7 days at 40 km, 4 days at 50 km, 5 days at 60 km) and ozone density (Figure 5d) (4.5 days at 30 km, 0 days at 40 km, -3 daysat 50 km) calculatedby the 2-D model are in much better
AT/AI20____5 / I205
0
0.2
0.1
T
1.0
agreement with observationsthan those derived earlier by 1-D models [Eckman, 1986; Brasseur et al., 1987]. The difference
0.3
--
•
50 6O
(c)
[]Observations 5O
1.0 30
10.0
-0.6
-0.4
0
-0.2
0.2
0.4
0.6
AO3/AI205 O3/ I-•o 5
30
4
0
8O m
0.01
40
,
10.0
8
12
16
T LAG RELATIVETO 205 nm (Days)
o x
• m
I
o SME 1.27 pm m
1.0
50
_(d)
--Model
7O •
m
40
0.1
60
\
10.0 -5
-4
-3
-2
-1
0
I
I
Ixb•
1
2
3
4
0 3 LAG RELATIVETO 205 nm (Days) -0.1
0
0.1
AO3/AILyo• 0 3/ ILyo• Fig. 5. (a) Ozone and temperaturesensitivitiesrelative to a 1% variation(27 days) in solar flux at 205 nm (expressed in % 03/% UV flux, and in Kelvin/% UV flux, respectively).Calculatedvaluesrepresentative of equatorialregionsarecompared as a functionof altitudewith LIMS, SBUV, andSAMS dataanalyzedby Keating et al. [1987] for the+ 20ø latitudeband. (b) Ozonesensitivityin the mesosphere relativeto a 1% variation(27 days)in solarflux at Lyman-rt(121.6 nm; expressed in % 03/% Lyman-ct). Calculatedvaluesare comparedto the analysisby Keatinget al. [1987], basedon the SME data. (c) Calculatedtime lag (days)of the temperature signal relative to the 27-day solar forcing, comparedwith observationsdata [Hood, 1986; Keating et al., 1987]. (d) Calculatedtime lag (days) of ozone signalrelative to the 27-day solar forcing, comparedwith observationaldata (SBUV) analyzedby Keatinget al. [1987]. Solid curvesrepresentthe model values,dashed curves the values deduced from observations.
•
[] SBUV: 27 Days
30
v
23,084
BRASSEUR:SOLARVARIABILITY AND MIDDLE ATMOSPHERE COMPOSITION
could be due to differencesin the adoptedradiative codesor to the fact that our 2-D model simulatesmore accurately than earlier
models
some of the feedbacks
between
radiation
and
zonal mean dynamics.
Finally, the variability predicted foi' the ozone column abundanceis presentedin Figure 6. Assuming again that the solar variation at 205 nm is 3.3%, the maximum responsein the vertically integrated ozone concentration,which is located near the subsolar point, is 0.25% (maximum sensitivity of 0.076 percent by percent). Thus even during the periods of highest solar activity when the 27-day solar signal is largest, the variation in the ozone column abundancein responseto the 27-day solar variability should be less than 0.5%. The response of the ozone column lags the solar forcing by 2.5 days. The overall pattern does not differ significantly with season in responseto changesin the intensity of meridional transport; the maximum ozone responsefollow, however, the seasonalvariation of the subsolarpoint. 4. MIDDLE ATMOSPHERE RESPONSE TO 11-YEAR SOLAR CYCLE
This is a direct consequence of the strongdownward circulation associatedwith planetary wave breaking during this period of the year, which transportsto relatively low levels and high latitudesozone-enhancedair massesfrom higher altitudesand lower
latitudes.
The difference
in the two results is noticeable
and puts constraintson the possible impact on total ozone of uncertaintiesin the solar variability in the spectral region of the ozone Harfiey bands. Note that the difference between the results shown in Figures7a and 7b is much smaller than the changein the forcing between208 and 265 nm (from 3 to 5%), because a significant fraction of stratospheric ozone is produced-by radiation at wavelengths shorter than 208 nm. Uncertaintiesin the flux variationsover the spectralregion of
the 0 2 Schumann-Runge bands(•, < 200 nm) affect mostlythe layers in and above the upper stratosphereand hence also have a limited
effect
on the total
ozone
column
abundance.
The
variability in the Herzberg continuum, especially in the atmosphericwindow between 190 and 210 nm has the largest impact on total ozone, and the long-term solar variability in this wavelengthregion seemsto be fairly well established. The
variation
from
solar
minimum
to
solar
maximum
Because the l 1-year period is long compared to the time conditions of temperature, ozone and other chemical neededfor the chemicalsystemto reacha quasisteadystate,the constituentsis shownin Figures8a and 8b to 11a and 11b as a responseof the atmosphere over a solarcycle can be derivedby function of altitude and seasonat the equator and at 60øN, comparingtwo model simulationsperformedat steadystatefor solar minimum and solar maximum conditions,respectively. As indicated earlier, the variation in the solar actinic flux between
these two model
calculations
is shown
in Table
SOLAR MAX- SOLAR MIN (11 yr)
1.
90N
Figures 7a and7b representas a functionof latitude and season the calculatedchangein the ozone column abundancefor an increase
in
the
solar
flux
from
minimum
to
60N
maximum
conditions. In Figure 7a the lower solar variability of 3% between208 and 265 nm is used,while in Figure 7b the higher value of 5% is adopted.In both cases,the changein the ozone abundance,which is of the order of 1.1-1.3% at the equator, increaseswith latitude, especiallyin winter and early spring.
TOTALOZONE - (DIF.%)
m 30N F--
EQ
ßJ
30S
60S
90N
90S
(a'•O3(' ._• 3%(20.8-265 nm) •_•_
L_L_J_J__t_
90N [_, , , ,,
I • ! • • I.-IJ •-•-• • I • •. • I •_
60N
60N ß
•
• 30N
• 308 6ON
90S 12/23
( 3/23
6/21
-' ..... •
9/20
12/18
TIME OF THE YEAR 90N
5•
5/17
5/27
6/6
Fig. 7. Responseas a functionof latitudeand •e of the ye• (percent)of the ozonecolran abundanceto the 11-year sol• cycle. (a) Assumedv•ation in the sol• flux be•een 208 and Fig. 6. Sameas Figure4, but for the response of the ozone 265 nm, 3%; (b) sine asFig•e 7a but wi• a v•iation of 5% TIME
colurn as a •nction of la•tude.
for the sol• flux between 208 and 265 nm.
B RASSEUR: SOLAR VARIABIIITY AND MIDDLE ATMOSPHERECOMPOSITION
TEMPERATURE DIFFERENCE (K) 60
0.96 •Xv• 5O
40
30
•""---'-'"••
0.32
• 20Za)Equato-I I I I • •1•
• • • I•
•
•,
I•
• • • • •
23,085
2-2.5% at 35-40 km altitude. Again, a strong seasonal variation in the responseis seen at high latitudes,whereasit is entirely lacking at the equator. Moreover, the amplitudeof the responsedecreaseswith altitude in the upper stratosphereand even becomes negative in the mesosphere. This feature in the model is a manifestation of the important temperature/ozone feedback in this region and of the role played by hydrogen compoundsin the ozone budget. It is, however, different from the l 1-year ozone variations deduced from the SBUV observations (Hood et al., submitted manuscript) which suggestthat the maximum ozone responsetakes place near the stratopause. This discrepancy between models and observations, which needs to be further investigated, is perhaps explained by the short time period over which data have been collected (poor statistics)but could also be related to the inability of the models to accurately reproduce the ozone concentration at and above 40 km altitude [Eluszkiewicz and
Allen, 1993]. Under this hypothesis,the "missing" process
2 60
0 3 DIFFERENCE (%)
50 60 40
50
30
20
40
-(b) 60øN I I ! I I I I
12/23
3/23
I I I I I I • • • • • I•
6/21
9/20
• I ! I •
12/18
30 -
TIME Fig. 8. Responseto the l 1-year solar cycle of temperature (Kelvin) as a functionof altitude and time of the year: (a) at the equator.and(b) at 60øN.
' J1.6•
Equator
Z
• 20 ! • [ •/ I I I I • • •'•!'qIII • • • II L •
tIff•••/'• •
••v
••
.-_•
