Changes in thermospheric composition inferred from twilight O+(2P) emission

0032-0633/79/05014653$02.00/0

Pha spocCSci.,VoL21,~.653-663 QPugamm~Ltd.,l979.RiatedinNortbrm~la,,d

CHANGES

IN THERMOSPHERIC COMPOSITION FROM TWILIGHT 0+(2P) EMISSION 3.F.NOXON

ad

INFERRED

R B. NORTON

Fritz Peak Observatory, Aeronomy Laboratory, NOAA, boulder, CO 80303, U.S.A. (Received 25 September 1978)

Ahatrset-Measurements of the twilight enhancement of airglow emission from O+(‘P) near 7325 A reveal major changes which accompany geomagnetic activity, no significant difference between evening and morning and an increase in brightness paralleling the approach to solar maximum. The principal source for O+(‘P) is direct photoionization from 0(3P) but at low solar activity there appears to be a contribution from another source in early twilight which may be local photoelectron ionization into O+(‘P). The geomagnetic and solar effects appear to reflect changes in the 0 and N, density in the thermosphere; ground based twilight measurements of O+ emission thus provide a simple means for monitoring thermospheric structum from 300 km to -500 km at solar minimum and to -600 km at solar maximum. INI’RODUCI’ION

Emission of the doublet at 7319 and 7330 A from O’(‘P) has been reported at twilight by Carlson and Suzuki (1974), Misawa (1975), and Meriwetber et al. (1978) using ground based measurements; observations of dayglow emission from a satellite are described by Rusch er al. (1977 and earlier references therein). The interest in this emission stems from the fact that O+eP) is produced by direct ionization of 0 atoms into the excited state by e.u.v. sunlight. The emission rate is thus closely related not only to the dominant source of ionixation in the F-region but also to the density of 0, the major neutral constituent above 250 km. The expected intensity of O’(‘P) emission in both the dayglow and at twilight was calculated by Dalgarno and McElroy (1966); the observations which have been reported are in general agreement with the calculations and confirm the value of such measurements in probing the structure of the thermosphere. Meriwether et al. (1978) have discussed both the twilight measurement technique and the intetpretation in some detail. All of the ground based measurements referred to report only one or two observations and cannot easily be compared with one another. The Atmospheric Explorer satellite observations of O+(‘P) emission were accompanied by a variety of measurements of other relevant quantities and there is agreement between the observed emission and that calculated from the measured 0 atom density and solar flux. In addition the Explorer analysis shows clearly the effects of quenching of O’(‘P) by both neutrals and electrons.

This paper presents the results of more than a year’s measurement of the twilight emission from mid-latitude which reveal a major effect upon the emission during geomagnetic storms. The changes appear to be associated with the dramatic alterations in thennospheric composition which occur at such times. We cannot account for all measurements by the photoionixation source for O+(*P) and suggest a second source, possibly photoelectron impact ionization, particularly near solar minimum. The effect of increasing solar e.u.v. emission upon the composition of the thermosphere is evident through an overall rise in twilight O’(2P) emission intensity between 1976 and 1978. OBSJCRVATIONAL

hlETHODS

The instrument nsed to record the twilight O’(*P) emission has been described by Noxon (1975); the procedures followed to extract the wanted emission feature from the background of scattered sunlight are similar, with the additional requirement that a correction is also required for the relatively constant background of OH nightglow. Figure 1 shows a twilight sky spectrum before and after subtraction of OH and after division by the background sunlight spectrum obtained early in twilight. While there can sometimes be a slow change in OH brightnessduring the twilight period, it is immediately recognizable; if necessary the background spectrum obtained in late twilight can then be scaled to ensure complete removal of OH emission from early twilight spectra. In practice the clarity of the sky is such at the Fritz Peak Observatory that at 15” elevation in the Sun’s azimuth the

and R. B. NORTON

during magnetic activity was evident soon after observations were begun in November 1976; magnetic activity leads to a lower intensity early in twilight and enhancement in late twilight. Figure 3 sets out the most complete record covering an entire month and illustrates the effect of geomagnetic activity on the twilight intensity. Figure 4 shows the dependence on geomagnetic activity of early and late twilight brightness together with the ratio of the two. We do not plot results for intermediate x; Fig. 3 shows that there is only a slight enhancement at x = 105”. Most of the points in Fig. 4 come from the period covered by Fig. 3, but we include in the lower plot several intensity ratios from the previous winter; these show that the dependence of the ratio on 4 did not change signScantly with season during the first year of measurement. The drop in the 102”/108” intensity ratio for A,, > 10 was evident until the end of 1977. In early 1978 the solar 10.7 cm radiation (F) underwent a marked increase to values well in excess of 100; as Fig. 5 shows, the geomagnetic dependence then disappeared. Neither the rate of decay nor the absolute intensity then changed much over a range in A from 0 to 50. On only one day in May was a large decay seen at low 4. Figure 6 shows the entire period from the fall of 1976 to the summer of 1978 in monthly averages, a total of 140 twilights in all. An attempt was made to reduce the measurements to 4 - 0 using the solid lines in Fig. 4 for 102” and 108”. For 105” the correction was small and was estimated from Fig. 3. Both corrected and uncorrected measurements were then averaged by month and are shown separately in Fig. 6. The solid lines in the figure are the intensity

FIG. 1. ~TOPSOLIDIJNRISANEARLYTWJLIGHTSPECTRUM AT 15” ELEVATION IN THE SUN’S AZIMJTHANDTHE DASHED LINE Is A LATE TWILIGHT SPRCTRIJMCONTAINING ONLY NIGHTGUJW OH EMISSION.

difference, at the bottom, shows the blend of the O’(q) doublet lines at 7320 and 733OA. The weak Pz lines of OH do not stand out at this resolution but fill in Their

the region between the strong P, lines.

scattered sunlight background is negligible when the solar zenith angle, x > 101” and occasionally for x > 99”. OBSERVATIONS

Figure 2 shows representative measurements on both a magnetically quiet and a disturbed day (4 < 10 and -50) and gives an indication of the measurement precision. That large changes occur

z 200-‘-y

~1~

7%:

‘O

II

IOU

I

102

III1

104

106

III 106

III 110

_ 112

X FIG. 2. -GHT

Each setof

o+ INTRNSITY AT 15” ELEVATION IN TIE SUN’S AZlMUTHFORQUIET(~<~0.x)AND ACTIVE (A, - 50,0) CONDITIONS IN THE FALL OF 1977. points is a single evening’s

measurementand indicatesthe precision of

the

measurements.

Changes in thermospheric

60--

I

I I I I I I I I I I I I I I I I I I I I

655

composition

I I I I I I I

!

I I I I I I i I I I I I

40”

Sept 1977

Aug Pm. 3. Twnmm IN THE FAIL OF

m-marry 1977

r?

0F O+pP)

Is AT

THE

mms.sIo~

BOTIDM;

AT

AT

15”

THE

TOP

-ATION IS THE

mx

,y =

AVERAGE

102”, lOTi”, 108” and 110” BOULDER MAGNETIC

DALY

DISlWRi3ANce,AWHICHMMlLARTO~.

Solid circles are evening, open circles are morning.

L hG. Solid

I IO

4. Top: THE TWILIGHT BRIGHTNESS AT x = 102” and 1080. BoTK)M:

I

I

30 Boulder A

al

OF o+(‘P)

I

40

I

SO

J5MISSION As A FUNCTION

I

60

I

OF MAGNEnC

DLSI’URBAN-

THE RATlO OF THE INTENSITY ATX = 102” and 108”. and open circles are evening and morning in August and September 1977, X are values from the previous late winter pe.riod.

656

J. F. NOXONand R B. NORTON twilight behavior sets in following a sudden increase in magnetic activity. Within 2 h after the Boulder 3 h K value rose above 2, effects were seen in the twilight as a large reduction in the rate at which the brightness decayed between x = 102” and 108”. DlSCUWON

We begin the discussion with the geomagnetic effect which we have found and attempt to see if it can be accounted for in terms of changes in the neutral composition of the thermosphere which are known to accompany geomagnetic storms. We start with a simplified discussion of the twilight geometry and then proceed to detailed modeling of the twilight in which the thermospheric composition is varied. By observing in the Sun’s azimuth at a zenith angle of 75” we greatly improve our sensitivity to the emission from O’eP); as Fig. 8 shows, the wanted emission from high-altitude is associated with a smaller value of x than is the unwanted scattered sunlight background arising in the troposphere. The calculations of Dalgamo and McElroy (1966) show that the effective “screening height”, H,, for 0’ at twilight is -300 km; this is the altitude at which the ionizing radiation is reduced to l/e as it traverses the atmosphere. Knowing H, we can estimate the altitude at which the observing line of sight crosses the shadow line and so obtain an altitude scale. Since N, and 0 have

AP FIG.

5.

THEDISAFT

DENCE

MR

o+(‘P)

EARANCE

OF TliE GE!OMAGNEl.IC

TWILJGHT

EMISSION

DEPEN-

F> 100.

WHEN

The heavy solid line is that shown in Fig. 4; the light lines are the MSIS prediction from Fig. 12 for F = 85 at the top and F=200 at the bottom. The numbers in parenthesis

are the mean values for F on the dates when measurements were made. calculated from a thennospheric model (MSIS) and are discussed in detail later. The outstanding feature in the measurements shown in Fig. 6 is the large increase in twilight brightness from early 1977 to early 1978; it obscures any purely seasonal dependence. Figure 7 illustrates how rapidly the change in

10

I

I I I I NDJFNANJJASONDJFNAN 1976

FIG.

I

I

I

I

I

I

I

I

I

I

1977

6. MONI-HLYMEAN lWlLIGHT INTENSITYOF o+(v) EhfISION x = 102”, 105” and 108”.

II

I

II

I

1976

I AT

15”

ELEVATTON

ANGLE

FOR

Dots are the measured values, circles give values corrected to A = 0 using the curves in Fig. 4. In 1978 no geomagnetic correction was needed (see text). The dashed line connects the corrected values; the solid lmes are calculated from the MSIS model as described in the text.

Changes in thermospheric composition

9

lo

II

n

I3

I4

15

December 1977

6_ 6-

,

In the ensuing discussion we assume that the excited 0+(2P) ions radiate at essentially the same location as they are produced and that any electric field which may be present does not move the ions more than a few km during their radiative lifetime of -5 s. This appears a reasonable assumption at mid-latitude and consistent with our measurements, At high-latitude, near aurora, this need not be the case. A later paper will discuss such high-latitude measurements wherein there is evidence for a significant motion of the ions during their lifetime as a result of a local electric field. Changes

$:1

27

20 July

29

30 1977

31

I

2

August

FIG. 7. SHOWS THE RAPIDITY WHICXi THE TwlL.IGHT DECAY IN o+(2P) Is REDUCED AFIER ONsET OF MAGNETIC AClWlTY; THE ARROws INDICATE THE TlhiE OF ON-.

nearly the same absorption coefficient (-lo-l7 cm’) over most of the wavelength range <700A at which O+(‘P) production is important it follows that under normal conditions H, is mainly determined by the 0 atom distribution, since at 300 km [Ol_6lSJ2l.

657

in N2 and 0 during magnetic storms

Over the past few years it has become clear that during geomagnetic storms there is sticient heating of the lower thermosphere at high-latitude to induce an upward and outward mass flow of such magnitude as to affect the thermospheric composition over the entire world. Both satellite composition measurements and their interpretation are discussed by Mayr and Volland (1974) and by Hedin et al. (1977). At mid-latitude the N, density is observed to rise more than a factor of 5 at all altitudes above 200 km; the [0] density, on the other hand, changes little at lower altitudes (<300 km) but increases at altitudes above 300 km. The 6rst evidence that such changes occur came indirectly even earlier from study of the decrease in electron density in the F-region during storms; the

FIG. 8. -h/LIGHT GEOMETRY. H, is the “screening height” and the arrows show the direction of observation during twilight as the observer moves away from the. sunset point. The Sun is at the left.

658

J. F. Noxon and R. B. NORTON

most plausible hypothesis was that the O/N2 ratio decreased since the production of ions depends upon [0] whereas the loss depends upon the reaction of 0’ with N2 and Oz. Even optical and incoherent scatter observation of low-latitude aurora only made sense when it was realized that such composition changes did occur (Noxon and Evans, 1976). The change in O’(*P) twilight emission associated with magnetic activity is in qualitative agreement with these changes in composition. With more N, near H, two effects come into play which both tend to reduce the volume emission rate immediately above the shadow line near the sunset point. If, for example, [N2] increases by 5 and [0] remains the same then 0 no longer dominates in the determination of H. and If, will increase. However, the absolute 0 density at the shadow will then be less and so the emission intensity will be reduced. There can also be a further reduction resulting from quenching of O+(*P) by Nz. In our observations A = 10’2” corresponds to the time when the line of sight intersects the shadow not far above H. and so will correspond to the situation described above. When x- 108” the intersection is above 400km where quenching is not important and where [0] is known to increase during storms; as a result the late twilight intensity will be enhanced. It follows that both a reduction in early twilight intensity and an increase in late twilight are qualitatively expected to occur during magnetic storms. The curious behavior of [0] reflects a competition between a tendency towards reduction as a result of composition change and a tendency towards an increase as a result of the associated rise in thermospheric temperature brought about by the auroral heating. The two cancel near 300 km but the latter dominates at 400 km. A model for the twilight decay We now investigate how well our measurements fit the recent MSIS thermospheric model (Hedin et al., 1977) which attempts to express the composition as a function of latitude, season, time of day, 4 index and solar decimetric radiation, F. Calculation of O+(*D-‘P) emission for comparison with the twilight measurements can be made by integrating the volume emission along the 15” elevation ray shown in Fig. 8. As indicated earlier the main source of O’(2P) is photoionization by solar U.V. radiation short of about 670A. The importance of photoelectrons will be estimated later. These metastable ions can relax to either the *D state by emitting at 7319-7330 A or the % state by

emitting at 2470A with transition probabilities of A1 = 0.16 s-l and A2= 0.046 s-l, respectively (Seaton and Osterbrock, 1957). In addition the *P state can be depopulated by collisions with electrons, atomic oxygen and molecular nitrogen with rate constants K1, K,, and K3 respectively. The density of molecular oxygen is too small to be important. The volume emission rate, q, can then be written: A,u~(A)E(AXOII(A)

dA

(1)

A1+A2+K1N.+K,CO]+K&l’ I(A)

=

b(~)e-J(u~o*~&”

(2)

N,, CO] and [N2] are the densities of electrons, atomic oxygen and molecular nitrogen, respectively; t is measured along the solar ray path. I,,(A) is the unattenuated solar flux just outside the Earth’s atmosphere. e(A) is the fraction of ionixations for A < 670 A yielding the *P state; it is approximately constant at 0.2. u*(A) and as(A) are the absorption cross-sections for 0 and N, respectively. Electron quenching can be estimated using the rate coefficient given by Hemy et al. (1969). An electron temperature of 1000 K leads to K,10m7cm3 s-l. The electron density is less than about 5 X 10’ cmm3 which leads to a quenching rate of at most 0.05 s-l. Rusch et al. (1977) have analyzed satellite observations of the O’(*P) emission and have estimated the neutral quenching coefficients to be: K2=5.2*2.5x10-“cm3s-’ and K3= 4.8 f 1.4 x lo-lo cm3 s-l. We have tried several combinations of K2 and K3 and find the dependence of the emission upon x not to be especially sensitive to the assumed values. We present results only for K2 = 0 and K3 = 10m9 cm3 s-‘. The values of solar flux used in the calculation are listed in Table 1 and are due to Hinteregger as reported by Rusch et al. (1976). The U.V. flux measurements correspond to a mean 10.7 cm flux of 92.2 which is in the range reported during our

TABLE 1. INCIDENT SOLARFLLX, IO, ANJJ~nsorwrrou CCXlFFICIENTSOFoANDN2FORTHESMWA IONS DISCUSSED

169-206 220-284 304 310-435 43.5-510 529-625

3.6 3.3 3.0 4.0 3.0 5.4

VELENGTIi

REG-

PI THE TEXT.

5.3 7.4 8.2 8.9 9.7 10.0

2.2 3.7 5.0 10.0 22.0 23.0

Changesin thennosphericcomposition observing period. Fluxes in the 300-5OOA region are included also and were obtained by scaling the values reported by Hinteregger and Hall (1965). The column emission calculations are insensitive to cross-section grouping so only six groups were used. Both atomic oxygen and molecular nitrogen contribute to the flux attenuation and their absorption cross-sections are listed also in Table 1. Again molecular oxygen is too tenuous to contribute to the flux attenuation at these altitudes. The atomic oxygen cross-sections were obtained from a compilation by G. Victor for the Atmospheric Explorer (private communication) and were adapted from the work of Henry (1970). Molecular nitrogen cross-sections were obtained from the compilation due to Hinteregger and Hall (1965). Figure 9 shows the thennospheric composition and exospheric temperatures given by the MSIS model for three values of A, and F= 100 for mid-latitude at sunset; the resulting predictions for the twilight emission appear Fig. 10. With increasing 4 the enhancement in [0] at high altitude leads to an increase in late twilight intensity whereas in early twilight the intensity is reduced owing primarily to the rise in I& brought about by enhanced N, density. We have compared our calculations using the MSIS model at low A, with the results of Dalgamo and McElroy (1966). The agreement with their calculation for the zenith intensity is within 10% in absolute intensity.

659

While Fig. 10 shows that there is a clear tendency for the twilight decay to become slower at high A,, there are serious discrepancies between our measurements and the predictions from the MSIS model. The predicted initial response (4 = 0 to 4 = 60) is an increase in early twilight brightness, contrary to observation. For an 4 intermediate between 60 and 400 Fig. 11 shows that a good fit can be obtained to the disturbed curve but there is a difference between the predicted and observed curves at low A,. There is also a problem, evident in Fig. 4, with the dependence of the intensity ratio on A,. The ratio appears constant for A,, < 10, drops swiftly between 4 = 10 and -30 and then assumes a much slower decrease with 4. No corresponding behavior is to be found in the MSIS model nor is it expected. Since there are several points of disagreement, we are led to question our assumptions concerning the production of o+(*P).

Photoelectron production of O+(*P) We have assumed the only important mechanism to be direct production of O’(*P) in the initial photoionization of 0, as have earlier workers. Additional production of photoelectrons has hitherto been neglected; it was mentioned by Rusch et al. (1977) but stated to be only a few per cent of the total with no details presented. If photoelectron

500-

‘\ _ -

Ap=O ‘. ‘. *.

‘xx -

Density

FIG. 9. [Cl, [NJ ANDEXOSPHERIC

TEMPElRATuRE

(cms3) FROM

THE

hrIsIs

MODEL

AT

4Q’=NFOR

F=

100

AND

SRVERALVAL~OF~ASDISCUSSED~N~TRXT. The

[0] curves are keyed by As and the [N,] curves by the cerrespendingexospheric temperature.

J. F. Noxou and R B. NORTON

660

FIG.

10.

PREDIClED

TWILIGHT

o+(‘P)

EMISSION

AT

15” ELEVATION

FROM

MSIS

THE

MODEL

FOR SRV-

eRALVALUESOF%;FOR%=OSEPARATECURVESARESHOWNH)RF=70AND100.

production were important then some of the discrepancies between predicted and observed behavior might vanish. In undisturbed conditions this source would rapidly decrease with time during twilight relative to photoionization. The reason is that when ionization occurs well above 300 km the photoelectrons do not lose their energy locally but either escape upwards or go down to be lost near 3OOkm (see Lejeune and Dalgamo, 1973). The geometry of the twilight observations is such that where the line of sight meets the shadow at 400 km the photoelectrons created there produce O+(‘P) emission well below the line of sight which does not contribute to the. observed glow. This effect will then lead to higher intensity in early twilight (when the shadow is near 3OOkm and electron energy degradation is local) but a steadily reduced enhancement from this source in later twilight. Under disturbed conditions the increase in Nz could lead to a strong competition for photoelectrons which would reduce the effectiveness of photoelectrons as a source for O’(*P). We might then identity the drop in the 102” intensity above 4 = 10 with the point where enhancement in N2 reduces photoelectron production of O+(*P). The gradual further reduction in the 102” intensity would then be due only to the effect of changing composition (decrease in [OI/m2D on the direct photoionization process. Figure 11 shows that photoelectrons must contribute about 40% of the total brightness in early

twilight when 4 is small if they are to account for the observations. To estimate whether they do contribute this much we turn to the discussion by Dalgamo and McElroy (1966). In their paper they calculate both the direct photoionization source for O’(2P) and also give an estimate for the effect of

I

I

I

I

I

I

I

1

X 11. mIs PREDICTIONS FOR 15” RLEVATION APPEAR ASDASHEDLINESFORF=100AND%=0,100AND200. The upper two solid lines are typical measurements for quiet and disturbed condition (4 - 0. SO) while the ldwer solid line is the ditference between the observed and calculated curves for 4 =0 and is what we assign to photoehtron production of O+(zP). FIG.

Changes

in thermospheric composition

conjugate photoelectrons. The latter is stated to be 103a Rayleighs (R) where a is the energy deposition rate in the high atmosphere by conjugate electrons in ergs cm-* SC’. At x = 1 lo” they calculate a zenith brightness of 9 R and also show the total ionization rate (centered at -400km) to be -10’ cm-* s-r. This permits us to calculate the intensity which would result if all photoelectrons deposited their energy locally as 3 R, assuming a mean energy of 20 eV for the photoelectrons. This estimate then suggests that perhaps 25% of the total production of O’(‘P) is due to photoelectrons. As an alternative estimate one can use the results of Dalgarno et al. (1969) who show that about 40% of the photoelectrons at 300 km have energy >20 eV, sticient to produce O+(‘P). The work of Lejeune and Dalgarno (1973) and Dalgamo and Lejeune (1971) indicates that perhaps 10% of these electrons will produce O’(*P) during their degradation in energy, ignoring any loss to NZ. Thus each ionization event leads to 0.04 O’(‘P) ions via photoelectrons. Direct photoionization leads to 0.1 O+(‘P) per ionization, and so about 30% of the total production is due to photoelectrons. These are very crude estimates but at least do not appear to rule out the photoelectron source completely. Owing to a similarity in the crosssections for energy loss of electrons in 0 and NZ the latter will not be important under quiet conditions near 300 km where [O]- 6j?lZ]. But during storm conditions &] approaches [0] so N2 should compete with 0 for photoelectrons and thereby reduce the importance of the photoelectron contribution, as required by our observations. We conclude that the abrupt drop in twilight intensity at 102” for 4 > 10 is probably due to the suppression of the photoelectron source and does not reflect a corresponding abrupt change in thermospheric composition. At 108” the onset of an enhancement for A, > 10 presumably represents the threshhold at which the mid-latitude thermosphere responds to storms. The early twilight dependence on A,, thus appears to reflect the increase in [Nz] near 300 km while the behavior in late twilight depends mainly upon the increase in [0] above 400 km. At present we can do no more than conjecture that the photoelectron source may contribute at early twilight under quiet conditions. The observations from the Atmospheric Explorer cannot be used directly to decide the question. Although one expects the source to be less important at lower altitude, compared with direct production of O’CP), the effect would be masked by quenching

661

production of of O+(‘P) by N> Photoelectron O+(zP) may account for the twilight behavior when 4 is small and when F< 100, and suppression of this source may explain the drop in brightness at 102“ when &> 10. This still leaves unexplained the actual twilight decay observed at 4 - 30 since the MSIS model requires a very much larger 4 in order to predict what is observed (see Figs. 10 and 11). We suggest that the model may in fact underestimate the degree to which the mid-latitude thermospheric composition is altered by only moderate magnetic activity. Recent discussion by Hedin et al. (1977~) tends to support this conclusion. The very rapid response of the twilight decay rate to geomagnetic activity seen in Fig. 7 implies a rapid movement of thermospheric air from auroral regions to mid-latitude. To travel 25’ of latitude in less than 2 h the southward velocity must exceed 300 m s-r. Just such speeds are observed (Hays and Roble, 1971; Hemandez and Roble, 1976) and are the result of upward and equatorward mass flow in the thermosphere induced by electrojet heating in the aurora1 oval. Seasonal

and diurnal changes

Figure 6 does not show any clear seasonal pattern, althought the tirst year’s measurements, up to November 1977, might suggest a summer minimum. For constant A,, = 0 and F= 100 the MSIS model does not predict a sign&ant seasonal change in twilight brightness but rather small maxima in spring and fall with a total range of a factor 1.6. Figure 3 shows there to be no discernible systematic differences between evening and moming brightness; this is not in agreement with the MSIS model which predicts an evening intensity at 108” 1.6 times larger than in the morning. Any seasonal or diurnal effects mainly reflect corresponding changes in the composition of the thermosphere; a true seasonal dependence is probably masked in Fig. 6 by the effect of changing solar e.u.v. output on the composition, as discussed below. It is possible that after several years of measurements a systematic seasonal effect may become evident. Eflect of changing solar activity Figure 6 shows that a steady increase in twilight intensity, and a decrease in the twilight decay between 102“ and 108”, have both occurred over the 18 months of observations. Moreover, Fig. 5 shows that while the dependence on 4 seen in Fig. 4 persisted through 1977 it changed completely in the spring of 1978 when F rose well above 100.

J. F. NOXON and R B. NORTON

662

SUMMARY

FIG. 12. h&IS FOR SFXRRAL IN0

THAT

MODEL VALUES

FOR LARGE

CALCULATIONS OF A,

SHOWN

EDR

F = 85 AND 200

ON THE

F THE PRJ3DICTRD

RIGHT

SI-IOW-

CURVELS ARJZ BOTH

LESSSENSITIVETO~ANDEXHLRITASLO~DRCAY.

The later period is one in which there is almost no dependence of the decay on 4 ; even at geomagnetically quiet conditions the decay rate is small. To see how welI the MSIS model can account for the effect of a rise in F we computed the solid lines in Fig. 6. We assumed 4 = 0 and used the mean value of F for the particular days in each month when measurements were taken. The e.u.v. flux was increased by 1.5 over the value discussed earlier and was assumed constant in time. The fit at 102” is fairly good but at 108” the model underestimates the actual range in brightness. Nevertheless it seems evident that the overall change is probably associated with changes in composition brought about by a rise in solar e.u.v. The tendency for the observed intensity to rise more than predicted after the summer of 1977 may be due to our use of a constant e.u.v. flux in calculating the intensity. If the e.u.v. flux were assumed to be proportional to F then the fit would be better. The greatest discrepancy is actually in the spring of 1977 in late twilight; presumably the 0 atom scale height was less than the model predicted. Figure 5 shows that when F>lOO the twilight decay no longer exhibits the dependence on A, seen in Fig. 4. Figure 12 shows a model calculation for F= 85 and 200 for di#erent A, values in which there is just a tendency; the 102°/108” intensity ratio at F = 200 is -3 for low 4, not too different from which is observed in the spring of 1978. There is no need to invoke a second photoelectron source although it is not clear whether it is rendered completely unimportant by the associated increase in @IJ/[O] in the thermosphere.

We conclude that the behavior of the twilight emission from O+(*P) is generally consistent with the main source being photoionixation into the excited state; at least at low solar activity there appears to be evidence for a second source in early twilight which may be local photoelectron impact ionization. It would be desirable to have better calculations for the photoelectron mechanism; those in turn may require better cross-sections for electron impact production of O+(‘P). Our measurements suggest that the MSIS thermospheric model may underestimate the change in thermospheric composition at mid-latitude brought about by moderate geomagnetic activity; the model also implies a systematic difIerence between evening and morning twilight intensity for O+(*P) emission which is not observed. At the same time it does predict rather well the change in twilight emission arising from an increase in solar e.u.v. emission through an effect on thermospheric composition. Although the model may not reproduce fully the response of the thermosphere at midlatitude to localized heating at high-latitude the rapidity with which the response occurs is very clear in our measurements. Although the interpretation of O’(*P) emission is not yet definitive, particularly in early twilight, the method seems to permit a fairly direct measurement of the atmospheric density prolile between 300 and -600 km on a routine basis from the ground at modest expense. Ground based monitoring of the emission is clearly a powerful tool for studying changes in thermospheric composition both in its short term response to geomagnetic activity and in the longer term response to change in the solar e.u.v. output.

Acknowledgements-We appreciate tbe help of E. Marovich and J. Smith in obtaining some of tbe twilight observations. Our interest in this work was stimulated by conversation with J. Meriwetber; we appreciate diiion on tbe topic both with bim and with D. Rnsch. C. Love has been of great assistance in the computer modelbng of tbe twiligbt.

Carlson, R. W. and Suzuki, K. (1974). Observation of O’(‘P-‘D) 7319A emission in the twilight and nigbt airglow. Nature 248, 400. Dalgamo, A. and McElroy, M. B. (1966). Twilight effects of solar ionizing radiation. Planet Space Sci. 14, 1321. Dalgamo, A. and Lejeune, G. (1971). The absorption of electrons in atomic oxygen. Planet. Space Sci. 19,1653.

Changesin thermospheric composition

Dalgarno, A., McElroy, M. B. and Stewart, A. I. (1969). Electron impact excitation of the dayglow. J. atmos. Sci. 26,753. Hays, P. B. and Roble, R. G. (1971). Direct observations of thermospheric winds during geomagnetic storms, J. geophys. Res. 76, 5316. Hedin, A. E., Saiah, J. E., Evans, J. V., Reber, C. A., Newton, G. P., Spencer, N. W., Kayser, D. C., Aicaydt, D.. Bauer, P., Cogger;L. and McClure, J. P. (1977a). A global thermospheric model based on mass spectrometer and incoherent scatter data-I. N, density and temperature. J. geophvs. Res. 82.2139. Hedin; A. E., Redkr,. C: A., Newton, G. P., Spencer, N. W.. Brinton. H. C.. Mavr. H. G. and Potter. W. E, (1977b). A global thermospheric model based on-mass spectrometer and incoherent scatter data MSIS. J. gwphys. Res. 82,2148. Hedin. A. E., Bauer, P., Mayr, H. G., Carignan, G. R., Brace, L. H., Brinton, H. C., Parks, A. D. and Pelz, D. T. (1977c). Observations of neutral composition and related ionosuheric variations during a maanetic storm in February 1974. J. geophys. Res. 52, 3183. Henrv, R. J. W. (1970). Photoionixation cross-sections for atoms and ions of carbon, nitrogen, oxygen and neon. Asnuphys. J. 165 1153. Henry, J. J. W., Burke, P. G. and Sinfaiiam, A. L. (1969). Scattering of electrons by C, N, 0, N+, O+ and O*+. Phys. Rev. 178, 218. Hemandez, G. and Roble, R. G. (1976). Direct measurements of night-time thermospheric winds and temperatures-2. Geomagnetic storms. J. gwphys. Res. 81, 5173.

663

Hinteregger, H. E. and Hail, L. A. (1965). Solar e.u.v. radiation and neutral particle distribution in the July 1963 thermosphere. Space Rex V, 1175. Lejeune, G. and Dalgamo, A. (1973). The red line of atomic oxygen at twilight. Planet. Space Sci. 21, 1937. Mayr, H. G. and Volland, H. (1974). Magnetic storm dynamics of the thermosphere. J. annos. terr. Phys. 36, 2025. Meriwether, J. W., Totr, D. G., Walker, J. C. G. and Nier, A. 0. (1978). The 0+(2P) emission at 7320 A in twilight. J. geophys. Res. 83. 3311. Misawa, K. (1975). Test observation of [OIII 7330A twilight glow. Rep. lotwsph. Space Res. Japan. 29, 199. Noxon, J. F. (1975). Twilight enhancement in O,(b’P) airgiow emission. J. geophys. Res. 80, 1370. Noxon, J. F. and Evans, J. V. (1976). Simultaneous optical and incoherent scatter observations of two lowlatitude auroras. Planet. Space Sci. 24,425. Rusch, D. W., Torr, D. G., Hays, P. B.-and Torr, M. R. (1976). Determination of the O’(2P) ionization fre&ency using satellite airglow and neutral composition data and its implications on the e.u.v. solar flux Geophys. Res. Len. 3, 537. Rusch, D. W., Torr, D. G., Hays, P. B. and Walker, J. C. G. (1977). The 011(7319-7330 A) dayglow. J. geophys. Res. 82,719. Seaton, M. J. and Osterbrock, D. E. (1957). Relative (OH) intensities in gaseous nebulae. Astrophys. J. Us, ‘66.