- Email: [email protected]
Thermospheric winds in the cusp: Dependence of the latitude of the cusp
P/mm. Printed
space sa.. Vol. 33, in Great
No
3, pp 305-313,
1985
00324633:85 $3.00 + 0.00 Pergamon Press Ltd.
Britain.
THERMOSPHERIC WINDS IN THE CUSP: DEPENDENCE THE LATITUDE OF THE CUSP
OF
R W. SMITH,*1 K. HENRIKSEN,? C. S. DEEHRJ
D. REES,$ F. G. McCORMAC* and G. G. SIVJEES Polytechnic, Jordanstown, Co. Antrim, U.K.; t University of Tromss, Tromsa, Norway; $ University of Alaska, Fairbanks, Alaska, U.S.A.; and 9 University College London, Gower Street, London, U.K.
*Ulster
(Received infinalform
I August 1984)
Abstract-The dayside thermospheric wind pattern as observed from Spitsbergen generally shows moderately strong westward winds with a small poleward component. The flow is almost zonal, frequently with sufficient westward velocity that parcels of air cross the noon meridian travelling towards the morning before turning antisunward towards the nightside across the polar cap. There have been some exceptions which arecharacterized by much weaker winds having been increased in thepoleward direction but with a very much reduced westward component. Making use of the meridian scanning photometer data obtained simultaneously on the same site, it is clearly shown that the normal behaviour occurs when the cusp, as indicated by the region of high 630/428 nm and 630/558 nm photometric intensity ratios, is to the North of the station. Just below the latitude of the cusp, the strong thermospheric flow generated by neutral coupling to the strong westward convection in the dusk sector continues across the dayside. It is maintained in the zonal direction because of the balance between the poleward Coriolis force and the equatorward pressure force caused by cusp heating. Poleward of the high pressure region at the cusp the flow is diverted northward and initially makes much slower progress across the Polar Cap. When the aurora1 oval has expanded such that the cusp is well to the South of our Spitsbergen station, the thermosphere in the sampled region has been found to be within this slow flow zone. On such occasions, the nightside speeds are well in excess of those on the dayside, in contrast to the normal behavior of comparable dayside and nightside wind speeds. INTRODLXZTION
Recent results with 3-dimensional thermospheric timedependent models (Fuller-Rowe11 and Rees, 1980; Roble et al., 1982) have shown the need for a high latitude, high altitude heat source which will give both the mean equatorward flow and the high cross polar velocities which are well-known features of the observational data. This paper reports an investigation ofthe cusp area of the dayside polar cap which is a prime candidate for this heat source. Fuller-Rowe11 and Rees (1980) have shown that even during periods of little magnetic activity, the averaged equatorward wind found from their thermospheric model is too small compared with the best expectations obtained by averaging long series of data from incoherent scatter radars and mid-latitude FPI stations. A means of providing a selective polar heat input of about 10’ W is required to resolve this difficulty. No particle heating is included in their model parametrically, since the heating effects at high latitudes have been presumed to be dominated by the Joule heating of aurora1 electrojet currents. Since these l_-.“l .I_ .l-* C R. W. Smith is now at the University of Alaska, Fairbanks, Alaska, U.S.A.
must be small at times of weak magnetic disturbance, they cannot be invoked for the missing heat source. Additional evidence for the heat source from DE 2 observations (Spencer et nl., 1982 ; Rees et al., 1983) is that an extra pressure gradient force is required in the region of cross polar flow in order to account for the high ratio of neutral-to-ion velocities. Such a pressure gradient is required at the noon entrance to the polar cap for neutral gas which is at the polar cusp. A pressure bulge in the cusp region may occur at high altitude because of the particle heating of the large flux of subkev particles which has been observed to precipitate there. If such a heat source exists, then the resultant high pressure region will be expected to have a noticeable effect on winds and temperatures in the upper regions of the thermosphere. A search was made in the data recorded by the Ulster Polytechnic Fabry-Perot Interferometer at Longyearbyen, Spitsbergen (LYR), whose geographic coordinates are 78.2N, 15.6E. In Fig. 1 the relationship between LYR and the aurora1 oval can be seen, showing that for typical Q = 4 conditions, the cusp passes near the zenith. This paper shows the results of an investigation of the effects in the observed winds, using the meridian scanning photometer data recorded simultaneously at the same site to locate the cusp. 305
306
R. W. SMITH et al.
FIG. 1. RELATIONSHIPOF SPITSBERGEN STATIONSTO THEAURORAL OVAL(Q = ~)AND THE SUNLIT EARTH IN DECEMBER ATMAGNETIC NOON(O~.OO U.T.). From Deehr et al. (1980).
THE OBSERVATIONS
The optical data discussed in this work was obtained during the Multi-National Aurora1 Expedition to Spitsbergen which has been in operation each winter season since 1978. During these 5 years, investigators from Norway, Alaska, U.K., Canada and Eire have collaborated in campaigns using Ebert-Fastie spectrometers, meridian scanning photometers (MSP), all-sky TV and Fabry-Perot Interferometers (FPI). Previous papers arising from this work have described the instrumentation (see for example, Deehr et al., 1980). The following brief summary applies only to the MSP and FPI instruments involved in the present work. The FPI was a 12.7-cm aperture piezoelectricallyscanned instrument controlled by a microprocessor which also provided coherent summing of successive interferograms. Each interferogram was obtained by counting photon pulses produced by a photomultiplier behind a pinhole in the center oftheinterference pattern in the focal plane of the associated telescope. Each scan, covering approximately one free spectral range, was 16 s in period, during which 100 samples were taken, equally spaced in wavelength. Since 20 or more scans were summed to produce an averaged interferogram, it
was expected that the temporal fluctuations of the 630nm 01 source intensity had a negligible effect on the profile shape. The free spectral range of the instrument was in the region of 23 pm for all the data reported and Doppler shifts in the 630-nm line were measureable to a precision of l/1000 of this wavelength interval, or better, depending on the intensity of the source. Hence the random error bar of the Doppler shift measurements along the line of sight was 11.0 m s-i. As discussed by many authors (Hernandez, 1982; Smith, 1980, etc.) there is a potential source of systematic error in finding the wavelength for zero source velocity since most observers do not have a suitable laboratory source of the 630-nm 01 emission. In this work, the zero velocity wavelength was determined on the assumption that a long-term average of wavelengths observed in directions symmetrically related in azimuth and of equal elevation is a good approximation to the zero wavelength and will not be biassed by the effects of local divergences of the wind field. Checks showed that the values obtained in this way were in good agreement with long-term averages of the zenith results. Systematic observations were made at an elevation of 30” using the sequence North, East, South, West, in geographic coordinates. Zenith observations were
307
Thermosphedc winds in the cusp inserted once per cycle of the azimuth. The Doppler shift data have been reduced by correction to the horizontal assuming negligible vertical component, and for vector diagrams, by a combination of the horizontal components measured assuming spatial uniformity. It is fully recognized that vertical velocities areoftennot negligible(Hernandez, 1982;Smith, 1980; Rees et al., 1984) but corrections are difficult since there is little justification for the assumption that an overhead measurement could be applied to the region at a sampled volume about 430 km away. Also, the assumed spatial uniformity used in the vector determinations is likely to be a crude approximation, hence vector plots should be treated with caution and used as a guide only. The MSP technique is well-known, and the instrument in use at the LYR site is ofa design similar to many instruments found at other observatories. A cluster of parallel narrow band filter photometer channels received light reflected from a chosen direction in the magnetic meridian plane by a plane mirror mounted on a shaft at 45” to the axis. Different directions of view were selected sequentialIy by steady rotation ofthe shaft. An on-line computer summed the photon counts from the photometers gated by the pulses from a shaft encoder for every degree of rotation. The channels of interest had a 1.O”conical field-of-view and were centered on 630,558 and 428 nm, recording the emission from 01 and N:. The 630-nm emission on the dark dayside was enhanced mainly by the soft particle precipitation in the cusp and cleft regions (Shepherd, 1980). The 428- and 558~nm emissions were sensitive to the harder precipitation which is typical of a structured aurora1 form. Also in the dayside region, 42%nm emission is enhanced by resonance scattering of sunlight. This gives a major brightening of the higher parts of aurora1 arcs [see for example, Stormer (1955) and Deehr et a[. (1980)]. Hence, the ratio of intensities 630/428 which is indicative of the nature of precipitation in nightside auroras is unreliable for dayside events. It is found that cusp precipitation selectively excites the 630-nm emission of atomic oxygen rather than the 558-nm line by a ratio of up to 100 in the mid-day gap (Henriksen et al., 1984). Therefore, a high intensity ratio 630/558 nm is indicative of cusp precipitation. However, it may normally be presumed (with less rigour) that the cusp or cleft is in the vicinity if there is a substantial increase in the red line intensity.
COMBINATION
OF’ DATA
Typical examples of two basic types of midday thermospheric flow as seen from LYR are displayed in
Fig. 2. The plots are in geographic coordinates and show the averaged wind over an 800-km circle centered on the observing site and at the height of emission of 630 nm, approxjmately 250 km. Figure 2a illustrates the most common type (A) in which the flow is about 300 m s-* and almost entirely westward. The other type(B) is quite distinct, illustrated by Fig. 2b, being of generally lower velocity, near 200 m s- * and with a significant poleward component superimposed on the much reduced westward wind. A survey of 13 such dayside wind plots indicates nine of type A and four of type B. MSPdata for the cases shown in Fig. 2 is displayed in Fig. 3. The stack plots of 630-nm intensity vs scan angle show that for the A case the diffuse stationary red arc maximizes overhead the LYR site in the pre-noon period. Also the 630/428 and 630/558 intensity ratios are high indicating that the cusp is on the LYR field line during the day. Conversely, the B case has the peak 630nm intensity to the South of LYR also with a high 630/428 ratio which is evidence that the cusp was South of LYR that day. Using this principle on all I3 cases, the following results were obtained : TABLE 1. CATEGORIESOF WIND PATTERNS Number
of cases 5 3 3 1 1
Wind type A B A B A
Cusp position Poleward Equatorward Overhead Undecided Slightly equatorward
The results of Table 1 show a clear trend in which winds on the poleward side of the cusp have a distinct poleward component while those at the same latitude or on the equatorward side of it are dominated by the westward component and stronger. Figure 4 shows the line-of-sight Doppler shift data for the same 2 days as in Fig. 3, corrected to the horizontal assuming no vertical winds. This shows that the smooth variations which appear on the vector plots are averages of a more variable wind. If the wind was uniform there should be no difference between velocity components measured in diametrically opposed directions (i.e., North and South, or East and West). Considering the days according to A or B type, it becomes clear that the reduced zonal and increased poleward trend is confirmed for the B typeday when the cusp passes equatorward of LYR. To investigate the wind gradients, Fig. 5 was prepared in which the southward (N-S) and eastward (E-W) gradients are plotted for all days and for each
308
R. W. SM~IH et al.
I -250ms
‘TRUE NORTH TIMES
IN
UT
(a)
I
250m.s
i’b)
TRUE
NORTH
TIMES
IN
UT 24
F1o.2. P~LARVECT~RPL~TSOFTHE~RMOSPHERICWINDVELOCITIESATABOUT~~~~~OBSERVEDFROM LYR BY ~E~PPLERSHI~OFOI 630-nm EMI~ION;(~)FOR ~~DECEMBER 1981, (~)FOR 30 DECEMBER 1981. data interval, grouping type A and B days separately. Type B days show a poleward wind gradient in which the wind observed to the North was rather stronger than that to the South and the westward wind rather weaker to the West than to the East. The type A days at LYR show no particular trends in the wind gradients. Type B days are thus characterized by a systematic divergence of the wind in the prenoon period of the dayside, in which the cross-polar wind increases poleward of the cusp region.
DISCUSSION
When geomagnetic activity increases, the aurora1 oval expands and the cusp passes on the equatorward side of the LYR site. Although observations have been made both polcward andequatorward ofthecusp those made of the poieward side were always for periods of greater geomagnetic activity. Hence it is necessary to estimate the effects that change of activity may have on the wind field in order to assess the significance of the
Thermospheric winds in the cusp
MAGNETIC
0830,&////&d-CUSP -
0800+=2
NOON
EMISSION
08OOe
v
N
309
2
S
FIG. 3. STACKPLOTSOF MSP SCANSINTHEMAGNETICMERIDIANPLANEAT
LYR SHOWINGTHEINTENSITYOFOI ~~O-~~EMI~~IONASAFUN~TIONOFELEVATIONANGLEANDTIME;(~)FOR~~DE~EMBER~~~~,(~)FOR~ODECEMBER 1981.
relationship which has been found concerning the difference between the winds on either side. The expansion of the amoral oval can change the relative importance of the driving terms in the momentum equation of the neutral thermosphere since, for example, the conductivities and electron densities have generally increased. This will increase the influence of the ion drag term making the wind field more immediately subject to the prevailing ion convection pattern. Also the thermospheric polar circulation has expanded to lower latitudes following the boundary of the aurora1 oval. Neutral winds on the
nightside at LYR generally tend to increase with geomagnetic activity. According to DE 2 results (Killeen et al., 1982), these winds are frequently a large fraction ofthe ion velocity and may even exceed it in the cross-polar jet. Hence one may expect generally increased winds at most points in the enlarged circulation pattern. The wind plot in Fig. 6 shows that in the UCL model (Rees et al., 1983) (which has no high latitude sub-kev particle heat source but includes ion-neutral coupling and Joule heating), the winds poleward of the cusp are not greatly different from those underneath or
310
R. W. SMITHet
equatorward. On that basis, simply an increase in wind velocity might have been expected at LYR when sampling on the poleward side of the cusp since the geomagnetic conditions would be more active. In the model, the zonal dayside winds occur because the turning effect of the poleward Coriolis force on the flow is resisted by an equatorward force due to a local pressure gradient. The implication of the data presented is that there is a feature in the thermospheric wind pattern which can be attributed to the presence of the cusp. The magnetospheric cusp is the location of a source of subkev particle precipitation and is often associated at the confluence of the two major ion convection cells which exist in the polar cap when IMF B, is negative (Heelis, 1984). Either heating due to electron bombardment, or ion drag, or Joule heating, or any combination of these, must be considered as candidates for the observed effect. Although each of these phenomena is present in all regions of the polar cap, there are good reasons to expect that all of them will be enhanced in the cusp region. The soft electron flux is locally intense in the cusp region (Heikkila and Winningham, 1971 ;
ms-’
al.
Potemra et al., 1978) which enhances both the particle heating and the Pedersen conductivity in the F-region. With increasing Pedersen conductivity, both ion drag and Joule heating are boosted. The wind field must adjust to these effects and hence show a signature of the cusp. Considering first the heating, one may attempt to estimate its effect by using photometric data on the Nl (0,l) first negative band at 428 nm and the 630 nm emission of0. The curves of Rees and Luckey (1974) can be used to provide estimates of the particle heating from this data if the observed intensities are due to particle impact only. Unfortunately, the LYR 428 nm photometer data on the cusp is substantially contaminated by emission due to resonance scattering. Stamnes et al. (1984) estimate that only 1% of the observed 428-nm emission observed in the zenith at magnetic noon at LYR originated from direct electron impact. Hence the curves of Rees and Luckey (1974) are not very useful in this instance. M. H. Rees (private communication, 1984) has estimated that the heating rate due to electron bombardment at 250 km in the cusp is2 x 10-9Wm-3ifoneassumes thatthecharacteristic
500 400 1 300 2OOJ
1
loo-
/-\, ,I I y,/,,,,,,,,,,.,,
\ -loo-
OfOE~:&jO_ll ,
.
-2oo-
ms-’
12% \
.1
/
14 15q6 __/'
17
18 19 20
21 22
23 24
01 02
03 04
05
06
"1
: : 01 02
! 03
: 05
4 06
UT
500400. 300.
100. \ -lOO-200
07'08-09 10 \ \ -
-3OO-
(a)
'\,
.I
fRST J';--_
"EST -
13
14
V\\.
__' STRRT TIME - 06 9D
: 24
04
/
,--,' ,
'\
40 _500
\
11 12
I'' ,-d_!IJ\~_:\‘---X :,:-F. 15 16'6 18 19 20 21 2223
I'
_'
I' I , ,I WdR5
Wil
1s xc
,961 NEUTRRL WIND
FIG. 4. THERMOSPHERK WIND VELOCITY COMPONENTS AT ABOUT 250 km OBSERVED FROM LYR IN THE GEOGRAPH~CCARD~NALDIRECTIONSDERIVEDFROMOPTICALDOPPLERMEASUREMENTSOF 630nmOI EMISSION; (a)FOR 18 DECEMBER 1981,(b)ro~~~DECEMBEK 1981.
311
Thermospheric winds in the cusp
i12
energy and energy flux of the precipitating electrons are 100 eV and 0.5 mW m-* respectively. Using a reasonable particle density of 2 x 1015 rnm3, a heating rate of 0.06 K s ’ is obtained which is comparable with the heating rate of 0.14 K s- ’ for parcels of the thermosphere passing through the cusp as computed by Killeen and Roble (1984). It is therefore concl-uded that such heating must be considered to be a competitive effect. The confluence of the ion convection cells will couple into the neutral gas causing a disruption of the uniform flow pattern. The precise details of this effect depend upon the local Pedersen conductivity and the configuration of the convection pattern. Conceivably, a different result would be expected for different signs of IMF B, (Heel& 1984). It has been shown (McCormac and Smith, 1984) that systematic variations in the pattern of thermospheric flow do occur for different signs of IMF B,. The present study points to another
13
14
15 16
17
18 19
20 21 22
23 24
UT
systematic effect due to the location of the cusp. At the time ofwriting, it remains an open question whether the changes between types A and B pattern reported above are caused mainly by heating or by ion drag.
CONCLUSIONS
The combination of FPI and MSP data from LYR shows that the thermospheric wind pattern at about 250 km observed on the dayside is dependent on the path taken by the polar cusp as it passes over the site. For cusp passes in the zenith or poleward of LYR, the winds are comparatively strong and dominated by the westward (geographic) component. When the cusp passes on the equatorward side, the winds are generally weaker and have rotated poleward. The behavior could be.due to the effects of a high pressure island caused by particle heating in the cusp or to the retardation which occurs at the confluence of two streams of air, each
312
R. W. SMITH et al.
(a)
(b)
500 A 300
”
i.
300-
No TREND .,
NO TREND
-i mloD~::..._.
: ,: ;
!.
..” : : :
:
: :
:
:
d.i: ”
E
JT
: :
Ii g-100”
,. ;
: :
.; ;
,i;5,.
:
B
s
8
(..
;.:.
g
.
G 5 E
.
.j:: :
I.,,‘. :
I.
~-,oo~Y~5
g_,,,-’
,.;‘.‘.I:,
I_
I.1
”“’
09.
‘07..;
,’
’
II
; : ; :’
;,,-
: ”
T
.
500B MAINLY
-
POLEWARD
-
ACCELERATION 100:‘:
g-100-
MAINY EASTWARD ACdELERATlON
“’
,.::
..’
6 .
.,
:: i :(
:
Y $300-
‘. loo-
:
‘,,i~:::Iii~~.::i’:‘;~r,:,:
k! w,-300k cl g-500g A
-
“.
;.,:_, ..05
. .
9
.. . .
‘y.
c::
09.
:.
”
..,:I.
it,.::
:
“,-300t
-5oo-
$500-
FIG. 5. THERMOSPHERIC WIND GRADIENTSAT ABOUT 250km ALTITUDEOBSERVED FROMLYR FOR 13 DAYS SEPARATED INTOA AND B TYPESAS IN TABLE 1; (a) EASTWARD,(b) SOUTHWARD.
UCL THERMOSPHERIC MODEL CONSTANT LATITUDE EVOLVING LOCAL TIME
82--c-ao-‘-
I I
-
‘f -+
NORTHERN HIGH LATITUDES WINTER SOLSTICE
.
1600
ival
z&al
cm
cm
1200
TIME UT FIG. 6. THEORETICAL VECTORPLOTOF THERMOSPHERIC WINDS AT 240 km ALTITUDEFROMTHE UCL MODEL (FULLER-R•WELLAND RED, 1980) FORA VARIETYOFLOCALTIMESAND LATITUDES ASSEENAT 15”E LONGITUDE.
Thermospheric
driven by ion drag, meeting probability both
effects
thermospheric
at the cusp locality. The
is that there are significant to
the
winds in the cusp
observed
cusp
contributions signature
by in the
flow.
Acknowledgements-This work has been supported by the British Science and Engineering Council under grant SC/D/04299 and the National Science Foundation under Contract ATM77-24837. The data analysis has been supported under NATO research grant 0513/82.
REFERENCES
Deehr, C. S., Sivjee, G. G., Egeland, A., Henriksen, K., Sandholt, P. E., Smith, R., Sweeney, P., Duncan, C. and Gilmer, J. (1980) Ground-based observations of F-region auroraassociated with themagnet0sphericcusp.J. geophys. Res. 85, 2185. Fuller-Rowell, T. J. and Rees, D. (1980) A three-dimensional time dependent global model ofthe thermosphere. J. Atmos. Sci. 37, 2545. Heelis, R. A. (1984) The effects of interplanetary magnetic field orientation on dayside high latitude ionospheric convection. J. geophys. Rex 89,2873. Heikkila, W. J. and Winningham, J. D. (1971) Penetration of magnetosheath plasma to low altitudes through the dayside magnetospheric cusps. J. geophys. Rex 76,883. Henriksen, K., Deehr, C. S., Romick, G. J., Sivjee, G. G., Fedorova, N. J. and Tatunova, G. F. (1984) Low energy enhancement of the 016300 A line and enhancement due to resonance of the N; first negative bands. Ann. Geophys. 2, 191. Hernandez, G. (1982) Thermospheric vertical motions. Geophys. Res. Lett. 9, 555. Killeen, T. L., Hays, P. B., Spencer, N. W. and Wharton, L. E. (1982) Neutral winds in the polar thermosphere as measured from Dynamics Explorer. Geophys. Res. Lett. 9, 957. Killeen, T. L. and Roble, R. G. (1984) Neutral parcel transport in the high latitude F-region, in Proceedings qfthe NATO
313
Advanced Workshop on the MorphologyandDynamicsofthe Polar Cusp (Edited by Holtet, J. A.). D. Reidel, Dordrecht. McCormac, F. G. and Smith, R. W. (1984) The influence of the Interplanetary Magnetic Field Y component on the ion and neutral notions in the polar thermosphere. Geophys. Res. Lett. 11,935. Potemra, T. A., Bostrom, C. O., McEntire, R. W., Petersen, W. K., Doering, J. P. and Hoffman, R. A. (1978) Low energy particle observations in the quiet dayside cusp from AE-C and AE-D. J. geophys. Rex 82,476s. Rees, D., Fuller-Rowell, T. J., Gordon, R., Killeen, T. L., Hays, P. B., Wharton, L. and Spencer, N. W. (1983) A comparison of wind observations of the upper atmosphere from the Dynamics Explorer satellite with the predictions of a global time dependent model. Planet. Space Sci. 31, 1299. Rees, D., Smith, R. W., Charleton, P. J., McCormac, F. G., Lloyd, N. and Steen, A. (1984) The generation of vertical thermospheric winds and gravity waves at aurora1 latitudes I ; Observations of vertical winds. Planet. Space Sci. 32,667. Rees, M. H. and Luckey, D. (1974) Aurora1 electron energy derived from the ratio of spectroscopic emissions. 1. Model computations. J. geophys. Res. 79,5181. Roble, R. G., Dickinson, R. E. and Ridley, E. C. (1982) Global circulation and temperature structure of the thermosphere with high latitude plasma convection. J. geophys. Res. 87, 1599. Shepherd, G. (1980) The dayside cleft and its ionospheric effects. Rev. Geophys. Space Phys. 17,2017. Smith, R. W. (1980) Neutral winds in the polar cap, in Exploration of the Polar Upper Atmosphere (Edited by Deehr, C. S. and Holtet, J. A.), pp. 189-198. D. Reidel, Dordrecht. Spencer, N. W., Wharton, L. E., Carignan, G. R. and Maurer, J. C. (1982) Thermospheric zonal winds, vertical motions and temperature as measured from Dynamics Explorer. Geophys. Res. Lett. 9, 953. Stamnes, K., Rees, M. H., Emery, B. A. and Roble, R. G.(1984) Modelling of cusp auroras: The relative impact of solar EUV radiation and soft electron precipitation, in Proceedings ofthe NATO Advanced Workshop on the Morphology and Dynamics of the Polar Cusp (Edited by Holtet, J. A.). D. Reidel, Dordrecht. Stormer C. (1955) The Polar Aurora. The Clarendon Press, Oxford.
space sa.. Vol. 33, in Great
No
3, pp 305-313,
1985
00324633:85 $3.00 + 0.00 Pergamon Press Ltd.
Britain.
THERMOSPHERIC WINDS IN THE CUSP: DEPENDENCE THE LATITUDE OF THE CUSP
OF
R W. SMITH,*1 K. HENRIKSEN,? C. S. DEEHRJ
D. REES,$ F. G. McCORMAC* and G. G. SIVJEES Polytechnic, Jordanstown, Co. Antrim, U.K.; t University of Tromss, Tromsa, Norway; $ University of Alaska, Fairbanks, Alaska, U.S.A.; and 9 University College London, Gower Street, London, U.K.
*Ulster
(Received infinalform
I August 1984)
Abstract-The dayside thermospheric wind pattern as observed from Spitsbergen generally shows moderately strong westward winds with a small poleward component. The flow is almost zonal, frequently with sufficient westward velocity that parcels of air cross the noon meridian travelling towards the morning before turning antisunward towards the nightside across the polar cap. There have been some exceptions which arecharacterized by much weaker winds having been increased in thepoleward direction but with a very much reduced westward component. Making use of the meridian scanning photometer data obtained simultaneously on the same site, it is clearly shown that the normal behaviour occurs when the cusp, as indicated by the region of high 630/428 nm and 630/558 nm photometric intensity ratios, is to the North of the station. Just below the latitude of the cusp, the strong thermospheric flow generated by neutral coupling to the strong westward convection in the dusk sector continues across the dayside. It is maintained in the zonal direction because of the balance between the poleward Coriolis force and the equatorward pressure force caused by cusp heating. Poleward of the high pressure region at the cusp the flow is diverted northward and initially makes much slower progress across the Polar Cap. When the aurora1 oval has expanded such that the cusp is well to the South of our Spitsbergen station, the thermosphere in the sampled region has been found to be within this slow flow zone. On such occasions, the nightside speeds are well in excess of those on the dayside, in contrast to the normal behavior of comparable dayside and nightside wind speeds. INTRODLXZTION
Recent results with 3-dimensional thermospheric timedependent models (Fuller-Rowe11 and Rees, 1980; Roble et al., 1982) have shown the need for a high latitude, high altitude heat source which will give both the mean equatorward flow and the high cross polar velocities which are well-known features of the observational data. This paper reports an investigation ofthe cusp area of the dayside polar cap which is a prime candidate for this heat source. Fuller-Rowe11 and Rees (1980) have shown that even during periods of little magnetic activity, the averaged equatorward wind found from their thermospheric model is too small compared with the best expectations obtained by averaging long series of data from incoherent scatter radars and mid-latitude FPI stations. A means of providing a selective polar heat input of about 10’ W is required to resolve this difficulty. No particle heating is included in their model parametrically, since the heating effects at high latitudes have been presumed to be dominated by the Joule heating of aurora1 electrojet currents. Since these l_-.“l .I_ .l-* C R. W. Smith is now at the University of Alaska, Fairbanks, Alaska, U.S.A.
must be small at times of weak magnetic disturbance, they cannot be invoked for the missing heat source. Additional evidence for the heat source from DE 2 observations (Spencer et nl., 1982 ; Rees et al., 1983) is that an extra pressure gradient force is required in the region of cross polar flow in order to account for the high ratio of neutral-to-ion velocities. Such a pressure gradient is required at the noon entrance to the polar cap for neutral gas which is at the polar cusp. A pressure bulge in the cusp region may occur at high altitude because of the particle heating of the large flux of subkev particles which has been observed to precipitate there. If such a heat source exists, then the resultant high pressure region will be expected to have a noticeable effect on winds and temperatures in the upper regions of the thermosphere. A search was made in the data recorded by the Ulster Polytechnic Fabry-Perot Interferometer at Longyearbyen, Spitsbergen (LYR), whose geographic coordinates are 78.2N, 15.6E. In Fig. 1 the relationship between LYR and the aurora1 oval can be seen, showing that for typical Q = 4 conditions, the cusp passes near the zenith. This paper shows the results of an investigation of the effects in the observed winds, using the meridian scanning photometer data recorded simultaneously at the same site to locate the cusp. 305
306
R. W. SMITH et al.
FIG. 1. RELATIONSHIPOF SPITSBERGEN STATIONSTO THEAURORAL OVAL(Q = ~)AND THE SUNLIT EARTH IN DECEMBER ATMAGNETIC NOON(O~.OO U.T.). From Deehr et al. (1980).
THE OBSERVATIONS
The optical data discussed in this work was obtained during the Multi-National Aurora1 Expedition to Spitsbergen which has been in operation each winter season since 1978. During these 5 years, investigators from Norway, Alaska, U.K., Canada and Eire have collaborated in campaigns using Ebert-Fastie spectrometers, meridian scanning photometers (MSP), all-sky TV and Fabry-Perot Interferometers (FPI). Previous papers arising from this work have described the instrumentation (see for example, Deehr et al., 1980). The following brief summary applies only to the MSP and FPI instruments involved in the present work. The FPI was a 12.7-cm aperture piezoelectricallyscanned instrument controlled by a microprocessor which also provided coherent summing of successive interferograms. Each interferogram was obtained by counting photon pulses produced by a photomultiplier behind a pinhole in the center oftheinterference pattern in the focal plane of the associated telescope. Each scan, covering approximately one free spectral range, was 16 s in period, during which 100 samples were taken, equally spaced in wavelength. Since 20 or more scans were summed to produce an averaged interferogram, it
was expected that the temporal fluctuations of the 630nm 01 source intensity had a negligible effect on the profile shape. The free spectral range of the instrument was in the region of 23 pm for all the data reported and Doppler shifts in the 630-nm line were measureable to a precision of l/1000 of this wavelength interval, or better, depending on the intensity of the source. Hence the random error bar of the Doppler shift measurements along the line of sight was 11.0 m s-i. As discussed by many authors (Hernandez, 1982; Smith, 1980, etc.) there is a potential source of systematic error in finding the wavelength for zero source velocity since most observers do not have a suitable laboratory source of the 630-nm 01 emission. In this work, the zero velocity wavelength was determined on the assumption that a long-term average of wavelengths observed in directions symmetrically related in azimuth and of equal elevation is a good approximation to the zero wavelength and will not be biassed by the effects of local divergences of the wind field. Checks showed that the values obtained in this way were in good agreement with long-term averages of the zenith results. Systematic observations were made at an elevation of 30” using the sequence North, East, South, West, in geographic coordinates. Zenith observations were
307
Thermosphedc winds in the cusp inserted once per cycle of the azimuth. The Doppler shift data have been reduced by correction to the horizontal assuming negligible vertical component, and for vector diagrams, by a combination of the horizontal components measured assuming spatial uniformity. It is fully recognized that vertical velocities areoftennot negligible(Hernandez, 1982;Smith, 1980; Rees et al., 1984) but corrections are difficult since there is little justification for the assumption that an overhead measurement could be applied to the region at a sampled volume about 430 km away. Also, the assumed spatial uniformity used in the vector determinations is likely to be a crude approximation, hence vector plots should be treated with caution and used as a guide only. The MSP technique is well-known, and the instrument in use at the LYR site is ofa design similar to many instruments found at other observatories. A cluster of parallel narrow band filter photometer channels received light reflected from a chosen direction in the magnetic meridian plane by a plane mirror mounted on a shaft at 45” to the axis. Different directions of view were selected sequentialIy by steady rotation ofthe shaft. An on-line computer summed the photon counts from the photometers gated by the pulses from a shaft encoder for every degree of rotation. The channels of interest had a 1.O”conical field-of-view and were centered on 630,558 and 428 nm, recording the emission from 01 and N:. The 630-nm emission on the dark dayside was enhanced mainly by the soft particle precipitation in the cusp and cleft regions (Shepherd, 1980). The 428- and 558~nm emissions were sensitive to the harder precipitation which is typical of a structured aurora1 form. Also in the dayside region, 42%nm emission is enhanced by resonance scattering of sunlight. This gives a major brightening of the higher parts of aurora1 arcs [see for example, Stormer (1955) and Deehr et a[. (1980)]. Hence, the ratio of intensities 630/428 which is indicative of the nature of precipitation in nightside auroras is unreliable for dayside events. It is found that cusp precipitation selectively excites the 630-nm emission of atomic oxygen rather than the 558-nm line by a ratio of up to 100 in the mid-day gap (Henriksen et al., 1984). Therefore, a high intensity ratio 630/558 nm is indicative of cusp precipitation. However, it may normally be presumed (with less rigour) that the cusp or cleft is in the vicinity if there is a substantial increase in the red line intensity.
COMBINATION
OF’ DATA
Typical examples of two basic types of midday thermospheric flow as seen from LYR are displayed in
Fig. 2. The plots are in geographic coordinates and show the averaged wind over an 800-km circle centered on the observing site and at the height of emission of 630 nm, approxjmately 250 km. Figure 2a illustrates the most common type (A) in which the flow is about 300 m s-* and almost entirely westward. The other type(B) is quite distinct, illustrated by Fig. 2b, being of generally lower velocity, near 200 m s- * and with a significant poleward component superimposed on the much reduced westward wind. A survey of 13 such dayside wind plots indicates nine of type A and four of type B. MSPdata for the cases shown in Fig. 2 is displayed in Fig. 3. The stack plots of 630-nm intensity vs scan angle show that for the A case the diffuse stationary red arc maximizes overhead the LYR site in the pre-noon period. Also the 630/428 and 630/558 intensity ratios are high indicating that the cusp is on the LYR field line during the day. Conversely, the B case has the peak 630nm intensity to the South of LYR also with a high 630/428 ratio which is evidence that the cusp was South of LYR that day. Using this principle on all I3 cases, the following results were obtained : TABLE 1. CATEGORIESOF WIND PATTERNS Number
of cases 5 3 3 1 1
Wind type A B A B A
Cusp position Poleward Equatorward Overhead Undecided Slightly equatorward
The results of Table 1 show a clear trend in which winds on the poleward side of the cusp have a distinct poleward component while those at the same latitude or on the equatorward side of it are dominated by the westward component and stronger. Figure 4 shows the line-of-sight Doppler shift data for the same 2 days as in Fig. 3, corrected to the horizontal assuming no vertical winds. This shows that the smooth variations which appear on the vector plots are averages of a more variable wind. If the wind was uniform there should be no difference between velocity components measured in diametrically opposed directions (i.e., North and South, or East and West). Considering the days according to A or B type, it becomes clear that the reduced zonal and increased poleward trend is confirmed for the B typeday when the cusp passes equatorward of LYR. To investigate the wind gradients, Fig. 5 was prepared in which the southward (N-S) and eastward (E-W) gradients are plotted for all days and for each
308
R. W. SM~IH et al.
I -250ms
‘TRUE NORTH TIMES
IN
UT
(a)
I
250m.s
i’b)
TRUE
NORTH
TIMES
IN
UT 24
F1o.2. P~LARVECT~RPL~TSOFTHE~RMOSPHERICWINDVELOCITIESATABOUT~~~~~OBSERVEDFROM LYR BY ~E~PPLERSHI~OFOI 630-nm EMI~ION;(~)FOR ~~DECEMBER 1981, (~)FOR 30 DECEMBER 1981. data interval, grouping type A and B days separately. Type B days show a poleward wind gradient in which the wind observed to the North was rather stronger than that to the South and the westward wind rather weaker to the West than to the East. The type A days at LYR show no particular trends in the wind gradients. Type B days are thus characterized by a systematic divergence of the wind in the prenoon period of the dayside, in which the cross-polar wind increases poleward of the cusp region.
DISCUSSION
When geomagnetic activity increases, the aurora1 oval expands and the cusp passes on the equatorward side of the LYR site. Although observations have been made both polcward andequatorward ofthecusp those made of the poieward side were always for periods of greater geomagnetic activity. Hence it is necessary to estimate the effects that change of activity may have on the wind field in order to assess the significance of the
Thermospheric winds in the cusp
MAGNETIC
0830,&////&d-CUSP -
0800+=2
NOON
EMISSION
08OOe
v
N
309
2
S
FIG. 3. STACKPLOTSOF MSP SCANSINTHEMAGNETICMERIDIANPLANEAT
LYR SHOWINGTHEINTENSITYOFOI ~~O-~~EMI~~IONASAFUN~TIONOFELEVATIONANGLEANDTIME;(~)FOR~~DE~EMBER~~~~,(~)FOR~ODECEMBER 1981.
relationship which has been found concerning the difference between the winds on either side. The expansion of the amoral oval can change the relative importance of the driving terms in the momentum equation of the neutral thermosphere since, for example, the conductivities and electron densities have generally increased. This will increase the influence of the ion drag term making the wind field more immediately subject to the prevailing ion convection pattern. Also the thermospheric polar circulation has expanded to lower latitudes following the boundary of the aurora1 oval. Neutral winds on the
nightside at LYR generally tend to increase with geomagnetic activity. According to DE 2 results (Killeen et al., 1982), these winds are frequently a large fraction ofthe ion velocity and may even exceed it in the cross-polar jet. Hence one may expect generally increased winds at most points in the enlarged circulation pattern. The wind plot in Fig. 6 shows that in the UCL model (Rees et al., 1983) (which has no high latitude sub-kev particle heat source but includes ion-neutral coupling and Joule heating), the winds poleward of the cusp are not greatly different from those underneath or
310
R. W. SMITHet
equatorward. On that basis, simply an increase in wind velocity might have been expected at LYR when sampling on the poleward side of the cusp since the geomagnetic conditions would be more active. In the model, the zonal dayside winds occur because the turning effect of the poleward Coriolis force on the flow is resisted by an equatorward force due to a local pressure gradient. The implication of the data presented is that there is a feature in the thermospheric wind pattern which can be attributed to the presence of the cusp. The magnetospheric cusp is the location of a source of subkev particle precipitation and is often associated at the confluence of the two major ion convection cells which exist in the polar cap when IMF B, is negative (Heelis, 1984). Either heating due to electron bombardment, or ion drag, or Joule heating, or any combination of these, must be considered as candidates for the observed effect. Although each of these phenomena is present in all regions of the polar cap, there are good reasons to expect that all of them will be enhanced in the cusp region. The soft electron flux is locally intense in the cusp region (Heikkila and Winningham, 1971 ;
ms-’
al.
Potemra et al., 1978) which enhances both the particle heating and the Pedersen conductivity in the F-region. With increasing Pedersen conductivity, both ion drag and Joule heating are boosted. The wind field must adjust to these effects and hence show a signature of the cusp. Considering first the heating, one may attempt to estimate its effect by using photometric data on the Nl (0,l) first negative band at 428 nm and the 630 nm emission of0. The curves of Rees and Luckey (1974) can be used to provide estimates of the particle heating from this data if the observed intensities are due to particle impact only. Unfortunately, the LYR 428 nm photometer data on the cusp is substantially contaminated by emission due to resonance scattering. Stamnes et al. (1984) estimate that only 1% of the observed 428-nm emission observed in the zenith at magnetic noon at LYR originated from direct electron impact. Hence the curves of Rees and Luckey (1974) are not very useful in this instance. M. H. Rees (private communication, 1984) has estimated that the heating rate due to electron bombardment at 250 km in the cusp is2 x 10-9Wm-3ifoneassumes thatthecharacteristic
500 400 1 300 2OOJ
1
loo-
/-\, ,I I y,/,,,,,,,,,,.,,
\ -loo-
OfOE~:&jO_ll ,
.
-2oo-
ms-’
12% \
.1
/
14 15q6 __/'
17
18 19 20
21 22
23 24
01 02
03 04
05
06
"1
: : 01 02
! 03
: 05
4 06
UT
500400. 300.
100. \ -lOO-200
07'08-09 10 \ \ -
-3OO-
(a)
'\,
.I
fRST J';--_
"EST -
13
14
V\\.
__' STRRT TIME - 06 9D
: 24
04
/
,--,' ,
'\
40 _500
\
11 12
I'' ,-d_!IJ\~_:\‘---X :,:-F. 15 16'6 18 19 20 21 2223
I'
_'
I' I , ,I WdR5
Wil
1s xc
,961 NEUTRRL WIND
FIG. 4. THERMOSPHERK WIND VELOCITY COMPONENTS AT ABOUT 250 km OBSERVED FROM LYR IN THE GEOGRAPH~CCARD~NALDIRECTIONSDERIVEDFROMOPTICALDOPPLERMEASUREMENTSOF 630nmOI EMISSION; (a)FOR 18 DECEMBER 1981,(b)ro~~~DECEMBEK 1981.
311
Thermospheric winds in the cusp
i12
energy and energy flux of the precipitating electrons are 100 eV and 0.5 mW m-* respectively. Using a reasonable particle density of 2 x 1015 rnm3, a heating rate of 0.06 K s ’ is obtained which is comparable with the heating rate of 0.14 K s- ’ for parcels of the thermosphere passing through the cusp as computed by Killeen and Roble (1984). It is therefore concl-uded that such heating must be considered to be a competitive effect. The confluence of the ion convection cells will couple into the neutral gas causing a disruption of the uniform flow pattern. The precise details of this effect depend upon the local Pedersen conductivity and the configuration of the convection pattern. Conceivably, a different result would be expected for different signs of IMF B, (Heel& 1984). It has been shown (McCormac and Smith, 1984) that systematic variations in the pattern of thermospheric flow do occur for different signs of IMF B,. The present study points to another
13
14
15 16
17
18 19
20 21 22
23 24
UT
systematic effect due to the location of the cusp. At the time ofwriting, it remains an open question whether the changes between types A and B pattern reported above are caused mainly by heating or by ion drag.
CONCLUSIONS
The combination of FPI and MSP data from LYR shows that the thermospheric wind pattern at about 250 km observed on the dayside is dependent on the path taken by the polar cusp as it passes over the site. For cusp passes in the zenith or poleward of LYR, the winds are comparatively strong and dominated by the westward (geographic) component. When the cusp passes on the equatorward side, the winds are generally weaker and have rotated poleward. The behavior could be.due to the effects of a high pressure island caused by particle heating in the cusp or to the retardation which occurs at the confluence of two streams of air, each
312
R. W. SMITH et al.
(a)
(b)
500 A 300
”
i.
300-
No TREND .,
NO TREND
-i mloD~::..._.
: ,: ;
!.
..” : : :
:
: :
:
:
d.i: ”
E
JT
: :
Ii g-100”
,. ;
: :
.; ;
,i;5,.
:
B
s
8
(..
;.:.
g
.
G 5 E
.
.j:: :
I.,,‘. :
I.
~-,oo~Y~5
g_,,,-’
,.;‘.‘.I:,
I_
I.1
”“’
09.
‘07..;
,’
’
II
; : ; :’
;,,-
: ”
T
.
500B MAINLY
-
POLEWARD
-
ACCELERATION 100:‘:
g-100-
MAINY EASTWARD ACdELERATlON
“’
,.::
..’
6 .
.,
:: i :(
:
Y $300-
‘. loo-
:
‘,,i~:::Iii~~.::i’:‘;~r,:,:
k! w,-300k cl g-500g A
-
“.
;.,:_, ..05
. .
9
.. . .
‘y.
c::
09.
:.
”
..,:I.
it,.::
:
“,-300t
-5oo-
$500-
FIG. 5. THERMOSPHERIC WIND GRADIENTSAT ABOUT 250km ALTITUDEOBSERVED FROMLYR FOR 13 DAYS SEPARATED INTOA AND B TYPESAS IN TABLE 1; (a) EASTWARD,(b) SOUTHWARD.
UCL THERMOSPHERIC MODEL CONSTANT LATITUDE EVOLVING LOCAL TIME
82--c-ao-‘-
I I
-
‘f -+
NORTHERN HIGH LATITUDES WINTER SOLSTICE
.
1600
ival
z&al
cm
cm
1200
TIME UT FIG. 6. THEORETICAL VECTORPLOTOF THERMOSPHERIC WINDS AT 240 km ALTITUDEFROMTHE UCL MODEL (FULLER-R•WELLAND RED, 1980) FORA VARIETYOFLOCALTIMESAND LATITUDES ASSEENAT 15”E LONGITUDE.
Thermospheric
driven by ion drag, meeting probability both
effects
thermospheric
at the cusp locality. The
is that there are significant to
the
winds in the cusp
observed
cusp
contributions signature
by in the
flow.
Acknowledgements-This work has been supported by the British Science and Engineering Council under grant SC/D/04299 and the National Science Foundation under Contract ATM77-24837. The data analysis has been supported under NATO research grant 0513/82.
REFERENCES
Deehr, C. S., Sivjee, G. G., Egeland, A., Henriksen, K., Sandholt, P. E., Smith, R., Sweeney, P., Duncan, C. and Gilmer, J. (1980) Ground-based observations of F-region auroraassociated with themagnet0sphericcusp.J. geophys. Res. 85, 2185. Fuller-Rowell, T. J. and Rees, D. (1980) A three-dimensional time dependent global model ofthe thermosphere. J. Atmos. Sci. 37, 2545. Heelis, R. A. (1984) The effects of interplanetary magnetic field orientation on dayside high latitude ionospheric convection. J. geophys. Rex 89,2873. Heikkila, W. J. and Winningham, J. D. (1971) Penetration of magnetosheath plasma to low altitudes through the dayside magnetospheric cusps. J. geophys. Rex 76,883. Henriksen, K., Deehr, C. S., Romick, G. J., Sivjee, G. G., Fedorova, N. J. and Tatunova, G. F. (1984) Low energy enhancement of the 016300 A line and enhancement due to resonance of the N; first negative bands. Ann. Geophys. 2, 191. Hernandez, G. (1982) Thermospheric vertical motions. Geophys. Res. Lett. 9, 555. Killeen, T. L., Hays, P. B., Spencer, N. W. and Wharton, L. E. (1982) Neutral winds in the polar thermosphere as measured from Dynamics Explorer. Geophys. Res. Lett. 9, 957. Killeen, T. L. and Roble, R. G. (1984) Neutral parcel transport in the high latitude F-region, in Proceedings qfthe NATO
313
Advanced Workshop on the MorphologyandDynamicsofthe Polar Cusp (Edited by Holtet, J. A.). D. Reidel, Dordrecht. McCormac, F. G. and Smith, R. W. (1984) The influence of the Interplanetary Magnetic Field Y component on the ion and neutral notions in the polar thermosphere. Geophys. Res. Lett. 11,935. Potemra, T. A., Bostrom, C. O., McEntire, R. W., Petersen, W. K., Doering, J. P. and Hoffman, R. A. (1978) Low energy particle observations in the quiet dayside cusp from AE-C and AE-D. J. geophys. Rex 82,476s. Rees, D., Fuller-Rowell, T. J., Gordon, R., Killeen, T. L., Hays, P. B., Wharton, L. and Spencer, N. W. (1983) A comparison of wind observations of the upper atmosphere from the Dynamics Explorer satellite with the predictions of a global time dependent model. Planet. Space Sci. 31, 1299. Rees, D., Smith, R. W., Charleton, P. J., McCormac, F. G., Lloyd, N. and Steen, A. (1984) The generation of vertical thermospheric winds and gravity waves at aurora1 latitudes I ; Observations of vertical winds. Planet. Space Sci. 32,667. Rees, M. H. and Luckey, D. (1974) Aurora1 electron energy derived from the ratio of spectroscopic emissions. 1. Model computations. J. geophys. Res. 79,5181. Roble, R. G., Dickinson, R. E. and Ridley, E. C. (1982) Global circulation and temperature structure of the thermosphere with high latitude plasma convection. J. geophys. Res. 87, 1599. Shepherd, G. (1980) The dayside cleft and its ionospheric effects. Rev. Geophys. Space Phys. 17,2017. Smith, R. W. (1980) Neutral winds in the polar cap, in Exploration of the Polar Upper Atmosphere (Edited by Deehr, C. S. and Holtet, J. A.), pp. 189-198. D. Reidel, Dordrecht. Spencer, N. W., Wharton, L. E., Carignan, G. R. and Maurer, J. C. (1982) Thermospheric zonal winds, vertical motions and temperature as measured from Dynamics Explorer. Geophys. Res. Lett. 9, 953. Stamnes, K., Rees, M. H., Emery, B. A. and Roble, R. G.(1984) Modelling of cusp auroras: The relative impact of solar EUV radiation and soft electron precipitation, in Proceedings ofthe NATO Advanced Workshop on the Morphology and Dynamics of the Polar Cusp (Edited by Holtet, J. A.). D. Reidel, Dordrecht. Stormer C. (1955) The Polar Aurora. The Clarendon Press, Oxford.