Transport of thermospheric NO to the upper stratosphere?

Planer. Spuce So, Vol. 32, No. 4. pp. 3’59409, Printed in Great Bntain.

TRANSPORT

1984

0032-0633/84$3.00+0.00 c] 1984 Pergamon Press Ltd.

OF THERMOSPHERIC STRATOSPHERE?

NO TO THE

UPPER

SUSAN SOLOMON

NOAA

Aeronomy

Laboratory,

Boulder,

CO 80303, U.S.A.

and ROLAND0

NCAR,

Boulder,

R. GARCIA

CO 80307, U.S.A.

(Received injnalform

3 August 1983)

Abstract-The rate of production ofN0 in the thermosphere is expected to vary greatly over the course ofan 1 l-year solar cycle because the fluxes ofboth extreme ultraviolet radiation and aurora1 particles are known to increase substantially from solar minimum to solar maximum. In the stratosphere, NO participates in a catalyticcycle which constitutes thedominant photochemical destruction mechanism for stratosphericozone. If appreciable long range transport of NO from the thermosphere to the upper stratosphere occurs, its effects should therefore be manifested in upper atmospheric ozone density variations over the 1 l-year solar cycle. In this paper, model predictions of the seasonal and latitudinal variations in upper stratospheric 0, associated with NO transport for different levels of solar activity are compared to satellite observations of upper stratospheric ozone abundances.

The production of thermospheric NO, is expected to vary greatly with solar activity, because both the flux of e.u.v. (extreme ultra-violet) radiation and aurora1 particles vary over the solar cycle, by about a factor of two. Therefore, it may be expected that more NO, would be transported to the stratosphere at solar maximum than at solar minimum. In the sunlit upper stratosphere, NO, catalytically destroys atmospheric 0, (Crutzen, 1970; Johnston, 1971)so that theeffectsof such a process might be reflected in the ozone density. Garcia et al. (1983) presented a numerical model study oftheresponse ofthemiddle atmosphere to the 1 lyear solar cycle. Briefly, the model simulates coupling between dynamics and chemistry ; inputs to the solar cycle calculations include changes in e.u.v., ultraviolet (u.v.) and visible radiation and aurora1 particles. Photolysis rates for photochemical processes are thus considered explicitly, as is changing thermospheric NO, production. That study found that large increases occurred in thermospheric, mesospheric and upper stratospheric NO, at solar maximum, as a result of which calculated upper stratospheric ozone decreased markedly in spring, as light returns to the polar cap. These ozone changes are due both to the destruction of ozone by the NO, catalytic cycle and to the interaction between NO, and odd hydrogen (see Garcia et al. for details). Therefore, observations of upper stratospheric ozone spanning at least half of an 1l-year solar cycle might shed some light on the validity of this theoretical prediction.

INTRODUCTION

It has long been suspected that thermospheric NO could be transported to the stratosphere (see, for example, Strobe1 et al., 1970; Brasseur and Nicolet, 1973). These early studies addressed the question by employing onedimensional models. In these models, a barrier to downward transport exists in the mesosphere, where NO photolyzes and recombines with atomic nitrogen. However, since the rate of NO photolysis is a strong function of solar zenith angle, the dependence of the photochemistry on latitude should be considered. In particular, the lifetime of NO becomes very long in the polar night mesosphere, where photolysis does not occur. Thus, two-dimensional models provide a more complete tool with which to study this question. Recent work by Solomon et al. (1982), Brasseur (1982) and Garcia and Solomon (1984) has shown that long range transport of upper atmospheric NO, (defined as N + NO + NO, + NO, + 2xN,O, + HN04) to the upper stratosphere is theoretically predicted to occur, especially in the polar night region. Frederick and Orsini (1982) obtained similar results using a onedimensional model by simulating the seasonal and latitudinal variations in photochemical processes. However, until recently, no data have been available against which these theoretical predictions could be tested. In this paper we present indirect evidence supporting this idea from an analysis of satellite ozone data. 399

400

S. SOLOMONand R. R. GARCIA

Avoiloble

220200 3 IL

180-

5 h 0

160-

% g : ‘0 i

I----

O3 Dota

BUV -1

pJv-(

140 120IOO80604020 0 44

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 46 48 50 52 54 56 58 60

62

64

66

68

70

72

74

76

78

80

82

Year

FIG. 1. OJASERVED VARIATION IN THE OTTAWA 10.7cm FLUX FROM 1944TO 1982(FROM SOLAR GEOPKYSICAL DATA, 1981) AND THE PERIODFOR WHICH SATELLITEOZONE DATA ARE PRESENTLYAVAILABLE.

Such data do exist, although they are not without problems and uncertainties. Figure 1 shows the observed time history of the Ottawa 10.7 cm flux from 1944 to 1982, as well as the years for which satellite ozone data are available. The 10.7 cm flux is a common index of solar activity level ; conventionally 90 defines quiet sun, 150 corresponds to moderate activity and 220 represents active conditions (Timothy, 1977). The Backscatter Ultraviolet (b.u.v.) experiment onboard NIMBUS4provideddatafrom 1970to 1977(seeHeath et al., 1973; Krueger et al., 1973). The coverage is excellent for 1971 and 1972, but is not very good after 1972; we will present those data which are available. Furthermore, instrument degradation produced a gradual change in the derived ozone densities over time. Fortunately, we need not examine absolute ozone values to search for possible effects induced by NO,.; different ways of using relative values will be shown below. The b.u.v. observing period close to solar maximum (1970 and 1971) is somewhat after the maximum of solar cycle 20, and the levels of observed solar activity were rather low. The solar minimum years near 1974 to 1977 were very quiet in terms of solar activity. In order to examine data from a more active period, we will also use some data from 1979 and 1980 taken by the s.b.u.v. instrument onboard NIMBUS 7 (see Heath et al., 1975). We note that good global

coverage is available during 1970 and 1971 from the b.u.v. experiment, and for 1979 and 1980 from the s.b.u.v. experiment. Significant interannual differences can be found in ozone densities even when only these four years are considered (see, for example, Fig. 7). The b.u.v. and s.b.u.v. data may, however, not be comparable at all altitudes, particularly since the inversion methods used were different in the two experiments. Thus the altitude resolution may not be the same. Similar problems can even occur in the b.u.v. instrument over time as a result of degradation. It will be shown, however, that when the absolute ozone values are suitably normalized, the instruments and years which wil be used at a particular altitude provide the same trends where NO, is not expected to influence ozone. This provides a test of the comparability of the datasets under consideration. A recent study by Frederick et al. (1983) also examined these data in detail and concluded that, in general, the two datasets are in good agreement with each other. Finally, we note that the raw s.b.u.v. data have been reanalyzed using the original b.u.v. inversion (D. Heath, private communication, 1983), and the results compared to those obtained with the new s.b.u.v. inversion. Large differences were found between the two inversions below about 7.0mb. In the present study we restrict our attention to pressures below 4.0 mb (the 4,2,1 and 0.75

Transport

of thermospheric

NO to the upper stratosphere?

mb pressure levels). Differences between the two inversions at 2.0 and 4.0 mb were as large as 15%.At 1 and 0.75 mb, however, differences are less than 5%. We show below that many of the observed interannual differences in ozone densities are substantially larger, and thus cannot be attributed to the inversion method. The most important uncertainty facing this analysis is the fact that it is not a controlled experiment. Other factors in addition to NO, certainly influence the upper stratospheric ozone content, notably temperature and the water vapor abundance (see, e.g., Nicolet, 1975 ; and Crutzen, 1971 for a review). The very important role of temperature in determining ozone densities near the stratopause is a well established phenomenon (e.g., Barnett et al., 1975). In particular, large temperature variations are observed to occur in late winter and spring (the seasons under consideration in the present study) during sudden stratospheric warmings, with corresponding changes in the ozone abundance (e.g., Ghazi et al., 1976; Ghazi and Barnett, 1980). To minimize the effects of such temperature variations, we shall concentrate on the Southern Hemisphere, where the zonal mean amplitudes of the temperature perturbations associated with sudden warmings are observed to be smaller than those occurring in the Northern Hemisphere (see, for example, Barnett, 1977). The largest Southern Hemisphere warming observed by the NIMBUS 4 satellite occurred on 26 July 1974 (Barnett, 1975). During this warming, the longitudinal temperature variation at 60s and 2 mb was about 40K, but the zonal mean temperature near 2 mb increased by

-70

JSMB

only 10K. Assuming that the ozone density varies approximately as the exponential of -1000/T (as shown, for example, by Barnett et al., 1975, and Ghazi et al., 1976), then this corresponds to a maximum zonal mean ozone change of 15-20x, which is only slightly smaller than the changes expected to result from NO transport. This number, however, should probably be considered an upper limit in terms of our analysis because the observed maximum temperature change was found on a single day rather than in the monthly mean. Ghazi and Barnett (1980) examined Southern Hemisphere ozone and temperature variations during the October 1970 warming. They found longitudinal temperature variations of about 15 degrees near 60s at 2 mb, accompanied by about 20% changes in ozone at this level. Zonal mean changes in both temperature and ozone were, however, significantly smaller. We can also examine the possible effects of sudden warmings by comparing the ozone data from different years used in the present study. For example, the normalized monthly mean ozone data in August 1970 and August 1971 at 70s are extremely similar at all levels as shown in Figs 2-5, in spite of the fact that the August 1970 warming was significantly larger than that observed in 1971 (Barnett, 1975; Harwood, 1975), suggesting that the effects of these Southern Hemisphere warmings on monthly and zonally averaged ozone densities are probably small. Finally, data is shown below for June (Fig. 7) representing a month in which large Southern Hemisphere warmings are not generally observed (see, e.g., Barnett, 1975). Large interannual ozone changes

IMBOELI

^__ -

401

-70

.7HB

IORTRI

A¶[,,,,,,,,.,,,,, - - N0 SBLRR SBLRR

MINIMUM YRXIMUH

2.3

-

2.2

-

I.,,,,,

- ----

/

-

-

1979. 1975.1976 1970. 1971 1979. 1980

2.1 2.0 m c z =

I

2

3

4

5

6 MBNTti

7

8

9

IO

II

Ii

1

-

1.9 I.8

-

;

1.7

-

5 e -

1.6

-

I.5

-

g

1.4

-

1.3

-

1.2

-

I

2

3

4

5

6

7

a

9

IO

MBNTH

FIG.~.OBSERVEDANDCALCULATED MONTHLY VARIATIONSIN RELAnvE OZONEAMOUNTSAT~OS 0.7 mb LEVELFORDIFFERENTLEVELSOFSOLAR ACTIVITY.

NEARTHE

II

12

S. SOLOMON

402 2.6

,

9.

and

-70 ItILl InmDE~I I I 1 1 1 8 '1' - -- - - N0 RURBRR.S.MlN

0

-

2.4

R. R.

GARCIA -70

1MB (ORTRI

2.6 -_---

S&RR HlNlNllH SBLAR NAXIHUH

-

.-

t

/

1974.19751976 1970.1971 -1979.1960

T

2.2

I.2 I.0

1.0

.,!J I

2

3

4

5

7

6

6

9

IO

II

12

.6’ I

2

3

5

4

9

8

IO

II

I2

HBNTH

FIG. ~.SAME

AS

FIG.LBUTFORlmb.

the NO, perturbation (more than 20%). Also, the ozone data to be presented will be normalized (see below) in such a way that any latitudinally uniform changes in ozone which would be likely to result from systematic global temperature variations over the 11-year cycle will tend to cancel in the normalized data. More detailed analysis would require complete examination of temperature observations for all ofthe periods under consideration in our study. Even if such a study were performed we would not definitively know the

are found to occur during this month as expected from calculated NO variations, and these are probably not due to warmings. It should also be noted that several studies have examined the observed systematic changes in temperature associated with the 1 l-year solar cycle. For example, Quiroz (1979) has estimated the upper stratospheric temperature variation to be of the order of 3-6 degrees during solar cycle 20. The resulting change in ozone near the 1 mb level would be about 48x, which is smaller than the changes expected from

-70 2MB

7

6

HBNTH

IMBDELI

r""'T"r81 1.5

1.0

P -----

.9

-

.e

I

2

3

N0 RlJR0RA.S.HIN 50LRR MINIMUM -SSOLRR MRXiMUM 4

5

6

7

1975. 1971 1980

/ , 8

9

10

ll

I2

.91

I

/,‘,“I 2 3

4

5

’

6

MBNTH

FIF.~.SAMEAS

FIG. 1.BUT FOR

2 mb.

I,’

7

HBNTH

6

1

9

‘,I

IO

II

’

I2

Transport of thermospheric NO to the upper stratosphere?

-_ _--

1.25

p -

NQ RlJRIIRR.S.MIN SBLRR HIHIMUM SOLRR MAXIMUM

' ', ' \ ' \

Y g J

k! Y A e

2

3

4

5

6

7

a

9

ie

LI

I2

1.00 .95 .90

1

2

3

MBNTii

4

s

6

7

8

9

lull

12

M@NTti FIG. 5. SAME

AS Fm.

magnitude of all of the terms capable of affecting the ozone density in the upper stratosphere, since another important factor in determining ozone densities at these altitudes is the water vapor abundance. For H,O, no comprehensive data are available to permit analysis of possible changes over the years considered, particularly at high latitudes. However, the effects of thermospheric NO are expected to be manifested at s~ecificlatitudes(polewardofabout 6O),altitudes{from about 40 to 52 km) and seasons (winter and spring). It seems unlikely that variations in HZ0 or temperature would yield features exactly like those expected to be produced by NO,. Further evidence for this view comes from examining the time constant of the ozone perturbation as a function of altitude; this will be discussed below.

RESULTS

Three different model cases will be presented in this study : (1) no aurora1 input with a modest NO flux at the upper boundary and solar minimum extreme ultraviolet and ultraviolet solar Buxes (this is intended to simulate very low levels of solar activity, see Garcia et ai., 1983) (2) parameterized “solar minimum” conditions for e.u.v., u,v. and particle fluxes as described by Garcia et al. (these include an aurora1 contribution and are probably close to moderate solar activity) and (3) parameterized “solar maximum” conditions also described by Garcia et al. These three cases are roughly comparable to the years 197419751976,1970-1971,

1, BIJTFOR

5

mb.

and 1979, respectively, because of the observed solar activity levels during these periods (Fig. 1). As discussed above, we do not wish to look for the effects of NO, input to the stratosphere by exarn~n~n~ the absolute magnitude of the possible changes, because of the possibilities of instrument degradation and differences between the b.u.v. and s.b.u.v. instruments. One way of examining the effects of thermospheric NO on stratospheric ozone is to consider the monthly variation in high latitude ozone at a particular pressure level for different levels of solar activity normalized to amonth when no change related to the solar cycle is expected to occur. According to our model calculations, large changes are expected in winter and spring, but no changes in summer due to the relatively short lifetime of NO, in the upper stratosphere. Thus the NO, deposited in winter will begin to produce ozone changes as soon as light returns to high latitudes, but is itself destroyed by sunlight in a matter of months (see below). Figures 2--5 present the calculated and observed monthly mean ozone abundances at 70s for four different pressure levels, relative to December (summer) values. Model results are shown only for sunlit months. since data are available only for sunlit conditions. It can be seen that the calculated seasonal trend in ozone abundance is quite different in high latitude spring for differing levels of solar activity. In some cases even the direction of the seasonal trend is affected ; for example at 1 and 2 mb (Figs 3 and 4) less ozone is calculated in August than in September for high solar activity but the trend is reversed for low solar activity. The calculated

S. SOLOMONand R. R.

404

lifetime of NO, is reflected in the time constant of the ozone perturbation (the time required for the three calculated curves to merge) and is of the order of a month near 1 mb (Fig. 3) but about 3 months near 4 mb (Fig. 5). We now turn to the data shown in Figs 2-5. Only months containing at least 8 days of data are shown. Indicated “error bars” correspond to one sigma standard deviations of the observed monthly means, and are intended to provide an indication of likely natural variations. Occasionally, variations occur from year to year which exceed the standard deviation and are not anticipated (e.g., May at 1 mb, Fig. 3). These may be due to unusual temperatures or water vapor abundances in that particular year. However, many of the trends expected from the model studies appear to be present in the data. For example, the shapes of the induced ozone variations in August and September and their general reversal at 2 mb depending on solar activity occur in the satellite data. Further, the time constant of the ozone perturbation appears in general to be significantly shorter at the 0.7 and 1 mb levels than at 2 and 4 mb, a result which is remarkably similar to the model predictions, although we repeat that differences between the b.u.v. and s.b.u.v. inversions are larger for the latter two levels, introducing an additional uncertainty there. However, it seems unlikely that random variations of other variables could produce such features. Figure 6 shows the observed seasonal trends at 40s at the 1 and 2 mb levels. At this latitude no significant solar

GARCIA

cycle variations are predicted by the model calculations. The data at the 1 mb level, and those at the 0.7 and 4 mb levels (not shown) seem consistent with this prediction. While the s.b.u.v. data at 2 mb seem to exhibit large year to year variation compared to the b.u.v. data, they are also mostly consistent with the predicted absence of large systematic variations. Other middle and tropical latitudes have been examined and display similar features to those shown here at 40s. Another way to search for the effects of thermospheric NO, is to examine the trend in ozone as a function of latitude for a particular month. Variations should then occur in the ozone gradient near high latitude in winter and spring. Again we must normalize with respect to some point where no change is expected over the solar cycle. We normalize to the equatorial values, since Garcia et al. (1983) predict a change of 5% or less over the 1 l-year solar cycle in the tropical upper stratosphere. It may be argued that normalizing the ozone data against equatorial values is ill-advised because of the possible presence of a quasi-biennial oscillation (QBO) in tropical ozone, or because of possible variations associated with sudden stratospheric warmings. However, there is little evidence of a QBO in ozone in the range of altitudes (4 mb to 0.7 mb) considered here. Wilcox et al. (1977) have studied the periodic variations in 0, for the years 1962 to 1974 and find that, although a 29-month oscillation can be detected, its amplitude is maximum at about 25 km and falls rapidly with altitude above that level. Similarly, Angel1 (1980) finds little indication of a QBO in ozone

1

1.9,

,

-40 zne lORTAl

I

I

,

,

7

‘3

,

I

,

,

I.8 1.7 m 1.6 r z = 1.5 Y f I.4

-----1974.1975.1976 --1970.1971 --1979.1960 .9’

I

1,1°“,1,’

2

3

4

5

6

’

I

9

i.l.’

IO

HBNTH

FIG.~.~BSERVED

MONTHLY

VARIATION

INRELATIVEOZONE AMOUNTS AT 40s Note change of vertical scale.

FORTHE

1 AND 2 mb LEVELS.

II

I2

Transport JUNE .75 MB

of thermospheric

1

2.4.

2.2 -

.a -

I

.6 L -60

-60

-40

-20

JUNE I

[HEOELI

2.4 --

--

-

-

0

20

1

.7HB

IDRTAI I

- - - 1974.1975.1976

NE RUR0RR.S.HIN

40

60

-

2.2 -

SELRR MINIHUH 50LRR HRXlHUH

J 60

.6-"""""""" -80 -60

-20

-90

0

1970.1971 -1979.1980

20

40

60

80

LATITUDE

LATITLIOE JUNE

Y m z

405

NO to the upper stratosphere?

IHB [DRTRI

I.2 -

I.0 -

.e -

.6 1 -60

-60

-40

-20

0

20

40

60

LRTITUOE

Fs.7.

OBSERVEDANDCALCULATEDRELATIVEOZONEVARIATIONWITH

above 25 km. The effects of sudden warmings on upper stratospheric equatorial ozone abundances are likewise believed to be small. For example, Frederick (1979) finds no changes in the observed zonal mean ozone density near 2 mb during a warming in March-April 1976, and Barnett (1975) indicates a temperature change there of only 2-3K during a Southern Hemisphere warming, which is too small to produce more than a few percent changes in ozone. Figures 7-9 show the observed and calculated latitude trends near 0.75 mb for three months. (Calculated trends at 1 mb are very similar to those shown here at 0.75 mb.) In mid-winter (June) and late

LATITLJDE

NEAR 0.75mb

IN JUNE.

winter (August) large changes in shape are calculated to occur at high latitudes, as sun returns to the region which has accumulated the most NO, as a result of being in the dark for the longest time. Small changes are calculated for September because the photochemical lifetime of NO, at the 0.75 mb level is only about a month. Again, occasional excursions are seen in the data which are not predicted by the model, and the model does not reproduce the details of all the trends with latitude (such as the bulge near 30 degrees in the summer hemisphere which is particularly noticeable in the August data, Fig. 8). But where large variations occur systematically from year to year, these observed

406

S. SOLOMONand R. R. AUGUST

.75 MB

GARCIA

[HBOELI 2.4

I,

I,

.7HB lORTA1 RUGUST I I1 ----

1:: v_*j

-

2.2 -

-

1974.1975.1976 1970.1971 1979.19BO

2.0 -

Y Cs 1.2 2 I.0

:;‘---------_I .B -

.B I .6l"' -80

-60

I' -40

I -20

0

20

40

60

I 80

.6" -80

I’

”

”

-60

-40

”

-20

0

“I’

20

40

' 60

"

BO

LRTITUOE

LATITUDE AUtUST

IHB lOATR1

2.4

! -----1974.1975.1976 1970.1971 1979.1960

2.2

.B -

-80

-60

-40

-20

0

20

40

60

80

LRTITUDE FIG.

8. SAME AS FIG.

trends are very similar to those calculated in the model for the appropriate level of solar activity. A similar curve is presented in Fig. 10 for the observations at the 2 mb level in August. The latitude gradient seen here is obscured due to large year to year mid-latitude variations, which may be due to planetary waves, as suggested by Frederick et al. (1983). This midlatitude variability, however, displays no particular order with respect to solar activity level, year, or observing satellite. The high latitude trends, on the other hand, largely follow the predicted pattern even at this lower level, except for one year of s.b.u.v. data (1980). At this level, however, the observed year to year

7, BUT FOR

AUGUST.

variations at high latitudes are not far outside of the observed variations about the monthly means, unlike the results shown in Figs 7 and 8 at the 0.75 and 1.0 mb levels. DISCUSSION Comparison of observations and model calculations suggests that the trends predicted by the model are present in the data, although the magnitude of the observed variations is generally smaller than predicted. This could be due to a number of causes : for example, the computed vertical velocities which transport NO

407

Transport of thermospheric NO to the upper stratosphere? SEPTEMBER

.7S

86

iM05ELl ,

mm-_ -

-

SfPTf I

.

IDRTRI

0

20

2.4

-55 ,

,

1

1. -40

-60

SEPTEMBER ,

IHB

AS FAG.

from the thermosphere may be too large. The adopted percentage of N(‘D) production assumed for N, dissociation in aurorae atso plays a major role in determining the net amount of NO produced for a given aurora1 ionization rate (see e.g., Rees and Roble, 1979; Solomon, 1983).Thus therearenumerous uncertainties in both the adopted chemistry and dynamics which can contribute to the magnitude of expected changes. Exactly how large an NO, variation is required to produce the ozone changes shown here? Present chemistry implies an increase from about 20 to 40 ppbv during the periodfrom 1974-1975-1976 to 1979-1980 near I mb. However, in the altitude range under

60

IDRTRI

‘.

1’

*,

g,

1.

5

20

40

60

7, BUT

40

LATITUDE

-20 LRTl

Fro. 9. SAME

-20

-40

-60

.6L-80

, .7kiB

N0 R~R0R~.S.M1 SDLRR HlNlMUH SQLRR MAX I MUM

-80

2.4{

MBER

_)

80

TUOE FOR

S~PTEMB~,

consideration, reactions with odd hydrogen and the 0 + 0, reaction contribute to the destruction of ozone along with the NO, catalyzed destruction process. Thus the ozone depletion induced by an increase in NO, depends on how much of the total loss rate is due to NO,. The computed ozone decrease is therefore dependent on how weli known the HO, abundances are at the relatively large solar zenith angles present in high latitude winter and spring. Certainly these depend on some reaction rate constants which are not known with complete certainty (such as the reaction of 0 with OH and OH with HO,, for example) as well as the correct treatment ofsolar U.V.penetration. Therefore, it should

408

S. SOLOMONand R. R.

_---

1.9

---

I.8 t

-

these variations are due to changes in atmospheric composition other than NO, cannot be ruled out. Direct observation of NO, abundances in high latitude spring will be required to provide conclusive proof of upper atmospheric effects.

1974.1975.1976 1970.1971 1979.1980

1.7 2

1 1.6 -

g

1.5 -

=

1.4 -

GARCIA

Acknowledgements-We wish to acknowledge Dr. Julius London, Mr. Xiu-de Ling, Dr. Judith Lean and Ms. Janet Falcon for helpinobtainingandreadingdata tapesusedin this analysis. The National Center for Atmospheric Research is sponsored by the National Science Foundation.

E d 1.3 ^ Y. i:'y. 8.

REFERENCES

1.0 .9 .B .7"""""""'.J -a0 -60

-40

-20

0

20

40

60

60

LATlTUOE

FIG. 10. OESERVFDRELATIVEOZONEVARIATIONWITH LATITUDE AT 2 mb IN AUGUST.

be noted that the absolute magnitude of the required NO, increase could be different from that derived with present chemistry. It has been shown that the changes in high latitude upper stratospheric ozone observed by the b.u.v. and s.b.u.v. instruments onboard NIMBUS 4 and 7 are similar to those inferred from theoretical studies of the effects of thermospheric NO production variations over the 1 l-year solar cycle. Particularly at the 0.75 and 1.0 mb levels, the observed changes in ozone are substantial in comparison to the observed variation about the monthly mean ozone density. The analysis of the lower levels, 2.0 and 4.0 mb, is subject to greater uncertainty both because of differences between the b.u.v. and s.b.u.v. inversions and because the interannual ozone changes there do not appear to be as far outside of the observed standard deviations of the monthly mean ozone abundances. Observed variations with latitude, altitude and season, as well as the apparent time constant for return to “normal” values as a function of altitude are, however, all consistent with theoretical expectations. This provides indirect evidence for transport of NO, from the thermosphere to the upper stratosphere. It should be emphasized that these effects are confined to the region poleward of about 60s. The present analysis applies only to the Southern Hemisphere, and it is not clear whether similar effects would be found in the Northern Hemisphere, where the atmospheric circulation is known to be quite different. The inferences made here are based on indirect evidence, and the possibility that

Angell, J. K. (1980) Temperature and ozone variations in the stratosphere. Pure appl. Geophys. 118,378. Barnett, J. J. (1975) Large sudden warming in the southern hemisphere. Nature, Land. 255, 387. Barnett, J. J. (1977) The antarctic atmosphere as seen by satellites. Phil. Trans. R. Sm. B 279, 247. Barnett, J. J., Houghton J. T. and Pyle, J. A. (1975) The temperature dependence of the ozone concentration near the stratopause. Q. Jl R. met. Sot. 101,245. Brasseur, G. (1982) Physique et chimie de I’atmosphere moyenne. Masson, Paris. Brasseur, G. and Nicolet, M. (1973) Chemospheric processes of nitric oxide in the mesosphere and stratosphere. Planet. Space Sci. 21,939.

Crutzen, P. J. (1970) The influence of nitrogen oxides on the atmospheric ozone content. Q. .I1 R. met. Sot. 96,320. Crutzen, P. J. (1971) Ozone production rates in an oxygen-hydrogen nitrogen oxide atmosphere. J. geophys. Res. 76, 7311. Frederick, J. E. (1979) The behavior of tropical ozone during

the stratospheric warming of March--April, 1976. J. Atmos. Sci. 33, 529. Frederick, J. E. and Orsini, N. (1982) The distribution and variability of mesospheric odd nitrogen: a theoretical investigation. J. atmos. terr. Phys. 44,479. Frederick, J. E., Huang, F. T., Douglass, A. R. and Reber, C. A. (1983) The distribution and annual cycle of ozone in the upper stratosphere. J. geophys. Res. 88,3819. Garcia, R. R. and Solomon, S. (1983)A numerical model ofthe zonally averaged dynamical and chemical structure of the middle atmosphere. J. geophys. Res. 88, 1379. Garcia, R. R., Solomon, S., Roble, R. G. and Rusch, D. W. (1984)A numerical response of themiddle atmosphere to the 1 l-year solar cycle. Planet. Spuce Sci. 32,411. Ghazi, A. and Barnett J. J. (1980) Ozone behavior and stratospheric thermal structure during southern hemisphere spring. Beitr. Phys. Amos. 53, 1. Ghazi, A., Ebel A. and Heath, D. F. (1976) A study of satellite observations of ozone and stratospheric temperatures during 1970-1971. J. geophys. Res. 81,5365. Harwood, R. S. (1975) The temperature structure of the southern hemisphere stratosphere August-October, 1971. Q. J1 R. met. Sot. 101,75. Heath, D. F., Mateer, C. L. and Krueger, A. J. (1973) The Nimbus-4 backscatter ultraviolet atmospheric ozone experiment : Two years operation. Pure appl. Geophys. 108, 1238. Heath, D. F., Krueger, A. J., Roeder, H. A. and Henderson, B. D. (1975) The solar backscatter ultraviolet and total ozone

Transport

of thermospheric

mapping spectrometer (SBUV/TOMS) for Nimbus G. Optic. Enyng. 14, 323. Johnston. H. S. (1971) Reduction of stratospheric ozone by nitrogen oxide catalysts from SST exhaust. Science, 173, 517. Krueger, A. J., Heath, D. F. and Mateer, C. L. (1973) Variations in the stratospheric ozone field inferred from Nimbus 4 satellite observations. Pure appl. Geophys. 108, 1254. Nicolet, M. (1975) Stratospheric ozone : an introduction to its study. Rec. Geophys. Space Phys. 13, 593. Quiroz, R. S. (1979) Stratospheric temperatures during solar cycle 20. J. geophys. Rex 84, 2415. Rees, M. H. and Roble, R. G. (1979)The morphology of N and NO in aurora1 substorms. Planet. Space Sci. 27,453. Solar Geophysical Data (1981) Comprehensive reports, number 443448, part 1. NOAA Environmental Data and Information Service, U.S., Department of Commerce, Boulder, CO.

NO to the upper stratosphere?

409

Solomon, S. (1983) The possible effects of translationally excited nitrogen atoms on lower thermospheric odd nitrogen. Planet. Space Sci. 31, 135. Solomon, S., Crutzen, P. J. and Roble, R. G. (1982) Photochemical coupling between the thermosphereand the lower atmosphere I. Odd nitrogen from 50 to 120 km. J. geophys. Res. 87, 7206. Strobel, D. F., Hunten, D. M. and McElroy. M. B. (1970) Production and diffusion ofnitric oxide. J. geophys. Res. 75, 4307. Timothy, J. G. (1977) The solar spectrum between 300 and 1200 A, in The Solar Output and its Variations (Edited by White, 0. R.), p. 237. Colorado Associated University Press, Boulder. Wilcox, R. W., Nastrom, G. D. and Belmont, A. D. (1977) Periodic variations of total ozone and its vertical distribution. J. appl. Met. 16, 290.