Adv. Space Res. Vol. 14, No. 4, pp. (4)29-(4)39, 1994 0273-1177/94 $6.00 + 0.00 Printed in Great Britain. All rights reserved. Copyright © 1993 COSPAR

CORONAL PHYSICS FROM ECLIPSE OBSERVATIONS

S. K o u t c h m y

Institut d'Astrophysique de Paris*, CNRS 98 bis Boulevard Arago, F-75014 Paris, France

ABSTRACT

Solar total eclipses are rare occasions offering the opportunity to make a snapshot of the solar corona. Thanks to the availability of a large radiative flux in the optical region, sophisticated methods can be used from the ground to analyse all parts of the highly structured whlte-light corona. Coronal absolute intensities and line emissions, including their polarization, are also studied to analyse density and temperature inhomogeneities, velocities and magnetic fields. Detailed density distribution is directly extracted from fine coronal structures. During the July I I , 1991 eclipse, the large 3.6m aperture CFH optical telescope was used to analyse time sequences over small coronal fields and to image the finest structures ; results from this experiment are presented with emphasis on small-scale dynamical plasma processes with possible inclusion of wave-phenomena. Finally, to prepare the SOHO mission on coronal physics, we give an overview of what is known on coronal structures from eclipse observations of the past solar cycles: temperatures, densities, velocities, occurence and structure of streamers, coronal holes, threads, overall variability.

INTRODUCTION

The corona is normally too faint to be seen from ground except at an eclipse of the Sun, because it is fainter than the normal blue sky, ///1//////24//. On the average, a solar total eclipse occurs somewhere in the world every year and a half, but often the band of totali ty crosses oceans. Its duration can be up to 7 min. The recent July 11, 1991 total eclipse was observed from populated areas including the big island of Hawaii, with the largest and best optical telescopes in the world.

During total solar eclipses, when first the photosphere and then the chromosphere are completely hidden from view, a faint white polarized halo produced by the scattering of the solar light from electrons (the K- corona) becomes visible: the white-light (W.L.) corona. Around the moving disc of the Moon slightly larger than the Sun, the corona is accessible from the transition region (T.R.) up to the most extended structures, like streamers, during a few minutes. At low spatial resolution "snapshot" observations are made, including spectra, to analyse the temperature and velocities. No shadowing or vignetting of the W.L. corona is seen during a natural eclipse and in good observing conditions, the limitations in the field of view is essentially due to the dominance of the brightness of the dust F-corona, //2//10//. The approximately constant sky brightness can also be considerably reduced, by using the IR or airborne experiments.

Absolute photometric measurements can be performed with a high accuracy, thanks to images of calibration stats ///1//, simultaneously observed on the same picture as the corona. Suitably at tenuated solar light can also be used immediately before or after the totality. Progress made in the last decades includes the use of an accurately calibrated radial neutral density filter to compensate the large average radial g rad ien t /1 / /2 / / of the W.L. corona and to precisely remove parasitic effects produced in the instrument and the sky (the aureola effect) ; the use of such filters considerably improves the spatial resolution by avoiding overexposures. At the 1991 eclipse, strictly linear and highly sensitive detector systems were used, especially for making time sequences with narrow passband filters. However, because of its simplicity, reliability and large storage capability, film is still largely in use at eclipses. In IR, new 2D arrays were used for the first time in 1991 to study the far corona (outside a radial distance of r : 5 solar radii from the center of the Sun). For this purpose the radial filter is replaced by an externally occulting system ("double" occultation !).

* associated with the Pierre et Marie Curie University of Paris VI, France

(4)29

(4)30 S. Koutchmy

Fig. 1. Processed WL coronal eclipse picture obtained on Ju- ly 31, 1990 to show small-scale gradients of intensities and im- prove ~he visibilit~ or faint de- tails ; notice the large noise produced by the graininess of the filra.

mimics the observed density s t ructures /5 / , so it is tempting to believe in the obtained amplitudes. At very large scales, the approximation is probably correct /6/ . Unfortunately, the corona is filled with very small- scale structures like jets and threads, and even the beta (/3 : 81rP/B 2) of the plasma is not known and the magnetic field can only be conjectured. This is possible in the case of coronal discontinuities, for example. Therefore, a rather detailed knowledge of the plasma density distribution is required to calculate the magnetic field needed to balance the lateral gas pressure. Loops are a special case as a dynamical equilibrium should be considered and the field is likely to be non-potential ; there, temperature inhomogeneities play a hidden role ; the so-called scaling law relating the temperatures, the pressure and the loop length, prevails.

At first sight, it is tempting to consider the temperature responsible for the variations of the heights of arches observed in the intermediate corona, outside active regions or close but not above active region. There are noticable variations during the solar cycle. The radial variation of electron densities is presumably related to the temperature ; however, arches are clearly larger of size at sunspot minimum when the coronal temperature, presumably related to the strength of the solar magnetic field, is lower, so we suspect that the morphology of the corona is probably more dependent on the evolution of the large-scale solar magnetic field ; it is not directly related to thermodynamical parameters like the temperature or, even, to the quasi-static magnetic field.

Although the occurence of many loops above and nearby active regions strongly suggests the coronal plasma is frozen in magnetic lines anchored in the Sun, there are many locations in the intermediate and outer corona where densities are largely enhanced above neutral lines and sheets which are extended outside active regions ; the most extreme cases are the large equatorial streamers at sunspot minimum. This equatorial "belt" of streamers is believed to be the source of the heliospheric sheet /6 / ; however, streamers are often seen at high latitudes and even at poles when the magnetic field is reversing its sign of dominant polarity. Accordingly, it is important to consider the morphology of the corona and the solar cycle /3/ . In /3/ , /9 / and / 2 4 / w e give the weighted deviation of the coronal streamers main extension, the stalk, from the radial direction, as a function of the phase of the cycle. The deviation is positive when the deviation is equatorward, a situation which is observed almost all the time, and negative when it is poleward ; this last situation is typically observed at time of polar reversal, after the maximum of sunspot activity ; the same behaviour is obtained from an analysis taking the sunspot number as a parameter /3/ . Using the average values and assuming that the coronal plasma is confined in streamers above the neutral l i n e s / 7 / / 8 / o f the chromosphere, a ha'ire representation of the flow lines in the corona can easily be drawn, see Figure 2, which quite well matches the real distribution of densities in the outer corona, as evidenced on an eclipse picture. This is in contrast with the map deduced from the computation of "magnetic field lines" using the current free approximation (potential field) and the observed magnetic fields at the photospheric level ; a detailed analysis of such maps / 5 / shows that they poorly match the observed individual structures of the outer corona. The occurence of very large azimutal and/or tangential gradients of pressure, far overpassing the radial gradients, is another difficulty when potential fields are considered.

Coronal Physics from Eclipse Observations (4)31

Optical spectroscopy is usually a t tempted at total eclipses. Many emission lines are easily observed with a slit-less spectrograph. However, it is today more useful to look with modern detectors at more "extended" diagnostics which cannot be performed with other methods: full line profile analysis of faint parts, like coronal holes and of streamers at r ~ 2, study of Ftannhofer-lines scattered by the interplanetaxy dust or the neutral component. Because of the lack of photons, X-EUV spectroscopy can hardly provide information coming from the analysis of an emission line at r ~ 1.5 ; Doppler shifts, line-width (to measure turbulent velocities) and even the radial gradients are there poorly known ; the Lyot-coronograph is also, for obvious reasons, improper for that , so total eclipses offer now the best opportunity for using the most advanced diagnostics for testing the theory of heating and acceleration of the coronal plasma. Unfortunately, total eclipses are rare and rather short. Because the coronal activity changes from one eclipse to an other one, it is difficult to check the results of a "snapshot", so results of the most sophisticated experiments are sometimes disappointing and even, not reported ! Finally, no information is provided over the solar disc during an eclipse.

There axe well established and classical results from eclipses: WL-photometry and polaximetry, structural analysis. They cover many solar cycles /3 / , so these eclipse observations are invaluable for analysing the coronal activity solar cycle, although coronal mass ejections (CME) have rarely been observed at eclipses. Obviously, orbital externally-occulted WL-coronographs are more suitable instruments to observe CME, so we will not consider here these events. Finally, the best spatial resolution observations of the corona were performed with the 3.6m aperture Canada-France-Hawaii Telescope (CFHT) on July 11, 1991, bringing a new insight in the study of the fine scale phenomena important for the physics of small-scale processes.

MODEL OF CORONAL STRUCTURES

Morvholo~ical Analysis

It is generally accepted that the analysis of quasi-stationary coronal structures of the Sun (including to some extent small-scale structures) gives an important insight into the knowledge of the solar magnetic fields, their evolution and their origin and, accordingly, of the solar activity. Indeed, it is quite obvious that a low beta plasma behaves like a tracer of the lines of force of the magnetic field, even when a dynamical corona is considered. The large value of the electrical (and thermal) conductivity of the coronal plasma is u additional factor when beta is not low. It is apparent on eclipse pictures that structures overlying the coronal hole regions are open, as are the magnetic field lines of force and, conversely, arch-like and loop-like structures axe observed around active regions and f i laments /4 /corresponding to closed magnetic field lines which will eventually be "disrupted" during a CME.

White-light eclipse pictures do not properly record the numerous coronal features unless special filters are used. The most useful method is to have a radial neutral filter to compensate the large radial gradient of light, especially over the inner corona. Then, coronal details appear more clearly. However, the overlavviug of the details superposed along the line of sight and, also, the large differences often observed between equatorial and polar regions, make the pictures still different from what would be "seen" by an experienced observer. This is illustrated on a f i g u r e / 2 / / 3 / / 4 / w h i c h shows, on the example of the well studied eclipse observation of June 30, 1973, a processed picture which leads to a result not far from the interpretation given when a drawing is produced from a naked eye observation. Such a W.L. picture shows best the distribution of densities in the corona, because the amount of observed light is linearly related to the amount of free electrons and, consequently, to the plasma densities, integrated along the line of sight. It is quite evident, from the inspection of this figure, that the plasma is largely confined to small-scale structures, that is to thread-like filaments and loops. This impression is largely confirmed by the inspection of an X-ray picture which shows emissions in the inner corona observed on the solar disc.

Plasma densities are directly deduced from W.L. observations (see further) ; temperature structures can be analysed with filtergrams made in line emissions. Both the richness and the complexity of structures are evident, suggesting permanent dynamical phenomena (flows) occuring essentially at very small scales and sudden instabilities occuring at laxge scales. Although all scales are largely revealed in these numerous observations, an even better spatial resolution is required to fully understand the physics of these structures. Attempts to improve the resolution were made recently by processing W.L. pictures, see Figure 1, but clearly better detectors or larger telescopes should be used. Model calculations can help in analysing the morphology of the corona. Let us consider the pressure equilibrium of the coronal plasma. Briefly, assuming an isothermal corona, the gas pressure P in the structures can balance the magnetic field pressure, at least in the middle corona where P ~ r -e , because the amplitude B of the large-scale magnetic fiehl (low-order harmonic components of the photospheric field) decreases like r -:~ there. Using a potential field approximation and some practical boundary layer conditions outside, numerical methods have been largely applied to "extrapolate" the magnetic field measured at photospheric levels. The deduced global morphology

(4)32 S. Koutchmy

Fig. 2. A naive mappin 9 of flow lines above chromospheric neu- tral lines computed for the June 30, 1973 solar total eclipse ; ace / 2 / o r / ~ / f o r a com~rison with the true coronal structure.

Density Structures Deduced from Eclivze Photometry

Since the intensity of the WL corona is directly related to the free electron densities due to the Thomson scattering, eclipse photometry provides direct measurements of coronal demdties. However, a good absolute intensity calibration is required and the contribution from the F-corona due to the scattering on dust particles should be subtracted. In first approximation, the coronal atmosphere is considered homogeneous and the measurement of the polarization ratio greatly help the separation of the K and of the F components, because the F-corona is not polarized for r _~ 5, s e e / 1 0 / ; the background A due to the parasitic light (instrument, sky) is not considered polarized and the polarization ratio PK of a model homogeneous K-corona can easily be predicted /11/, see Figure 3. Then, the observed polarization deduced from the measurement of both radial B~t and tangential BT brightnesses of the corona is:

Ptot - B T - B i ~ . _ K T - - K R ( 1 ) B,r + Ba KT + KR + F + A

and KT -- Kit

PK-- KT + KR

Further, the K-corona brightness can readily be deduced:

(2)

K = K T - K I t _ Ptot ( K + F + A ) (3) PK PK

from the measured K+F +A = B(r), Ptot and from the calculated PK. This approach has been extensively developed in the past to describe low-resolution observations and deduce average models of N~.

Although average density distributions are useful for comparing the corona at different latitudes and different phases of the sunspot cycle, they axe far from the real values. The most conservative estimate of the filling factor (f.f.) in the intermediate corona gives a value of less than 10%, so average values correspond to a vastly oversimplified picture of the corona, even without considering really small-scale structures with diameter less than 10 axcsec.

4-

3

2

1-

I'to t [ RQ] fiNe)

• / _ ~ / p(Ne 2)

,,.,,

i I I

1.5 2 2.5

Pk [°/o]

-60

-50

"40

-30

-20

-10

9(ro)

Fig. 3. Radial variation of the total effective line of sight integration length for WL observations (l(Ne)) and for emission line o b . . a t i o ~ . 0(Ne~)), when the radial vari- ation of the electron densflies corresponds to a hydro- static temperature of 1.5 MK. On the same scale in ab- scissae we show the corresponding polarization degree of the K-corona in the optical region (A = 0.55p) cal- culated using the work of Saito /II/ ; note that p = r cos(A - Ao).

Coronal Physics from Eclipse Observations (4)33

Loops

Threads /

/ / __--

----- Polar ----- Riys

T / - - 'I ' i i l l / . / l i , . -

T4 Regions ,11' void / ! " J - - - , l l l l

/ . L f f R:; l:nrs Thyd?t.. _ , , ,

I I Is MK _--- I ~'. MK

[I 29'9 s, 113 2;s 2 Ii3 +i!+i

Radial dlstancesin solar Radii

.1010

• 10 9

"I .ms

0 Z

II

Oi o

40 s

Fig. 4. Radial variations of the electron density in different parts of the coron- a from eclipse photometry. Inserted at the bottom note the graph with a set of lines to show the slopes corresponding to different hydrostatic temperatures.

Let us consider the surface local brightness (intensity) of the K-corona I = I(p) along a radial direction, a being the projected on the plane of the sky radial distances. It has been found convenient to represent I(p) by a sum of terms of the form p - m with m tak- ing different predominant values (from 17 to 3.5), depending on the value of the radial dis- tance. Between the surface and few solar radii (r < 3 or 4), the derived radial density dis- tribution, with mi = 17 and m2 = 7, follows nicely a curve described by a hydrostatic law with a constant temperature Th of 1 to 2 x 10'; K (106 K = 1 MK), see Figure 4. This seems to indicate that the departure from hydrostat- ic equilibrium due to the hydrodynamic flow (solar wind) is small in this region.

Assuming/~ the mean molecular weight of particles in the corona is known from consideration of abundances and ionization degrees (H and He are fully ionized in first approximation, so # _ 0.6), MH the mass of a proton and G the constant of solar gravity at the surface, we can write:

fire(r) = fire(l) e:~p[ - (# Mt! GR@/]c T]t) (1 - l / r ) ] (4)

where fire(l) is the density at the "surface" of the Sun. A more convenient form of (4) can be deduced:

fir,(r) = firo e = v - (13.9/T, , .h / r ) (S)

where Tu is now expressed in MK, r still in solar radii and h is the height above the solar surface in the same units.

This formula is obviously valid only in the inner corona, assuming 2]t is known. TH can be deduced from the knowledge of ge( r ) assuming a constant filling factor in the corona. Then Ne(r) is readily deduced from the measurement of I(p):

c o

I(p) __ ~® ~. f fire(r) W~(r) ¢(A - Ao) d Z (6) - - 0 0

where E is the isotropic part of the Thomson scattering coefficient, B~ is the average brightness of the solar disc, p = r cos(A - Ao), W~(r) is the dilution factor which takes into account the limb darkening of the solar disc, ¢ is a function of the solar apparent longitude A describing the anisotropy of the scattering ; Ao is the longitude corresponding to the solar limb and 1 is the line of sight (l.o.s.). Wi th a good approximation, we could assume that W~ "" Wo/r 2 and ¢ ~. 1.25, because the effective l.o.s, integration length left, see Figure 3, is of the order of 1 solar radii, lerr is indeed introduced to permit a straightforward interpretation of images seen on eclipse pictures. We define lelr as the length of the l.o.s, which corresponds to a decrease of the scattered intensity by a factor e "~ 2.7, compared to the value obtained at Ao : 0 ; assuming a radial variation of Ne(r) given by (4) and taking a hydrostatic temperature T = 1.5 MK, le~ can be deduced, see

JASR 14:4-D

(4)34 S. Koutchmy

Figure 3. Now the filling factor f.f. in its simplest form could also be introduced. Assume that along l,,Ir only a part a of the l.o.s, is occupied by a plasma of constant density, then the f.f. is equal to cz, assuming a homogeneous distribution normal to the l.o.s., which is not the ease because structures have a limited extend, so f.f. ~ oz. Another factor which is often mentioned in the literature to describe the density irregularity is: < Ne 2 > / < Ne >2. There is a relation between this last factor and the f.f. ; however, because a whole spectrum of irregularities exists in the corona and observations have a finite resolution, this relationship is not easy to deduce, nor to measure. Note that permitted line emissions (like radio emissions) are also a function of the coronal local temperature and not just of < N~ >.

As said before, the analysis of the polarization degree in the K-corona alone, PK, gives some insight in the knowledge of the distribution of Ne along each l.o.s. The largest polarization is produced by electrons at A = A, ; outside this plane (limb-plane or the "plane of the sky" defined as the plan normal to the 1.o.s. and passing through the center of the Sun) the polarisation decreases. Assuming a spherically symmetric distribution and Th ---- 1.5 MK, the polarization degree of the K-corona can be computed, see Figure 3. Experimental values are shown on Figure 5 for an eclipse corona observed two years before the minimum of solar cycle (1973) and an eclipse corona observed at sunspot maximum (1980). A large dispersion is clearly observed, although the radial behaviour of the average PK fits quite well the theoretically predicted values of Figure 3. Outside of r = 3 both theoretical values and the measured one are unreliable: the hydrostatic equilibrium is no more valid and the dominance of the F-corona makes the measurements of PK very noisy. At r < 3, it is indeed already the ease inside the polar coronal hole of 1973, see Figure 5, where a special analysis was performed /2 / to determine Ne assuming a homogeneous radial distribution with the result shown on Figure 4. However, even a coronal hole is very inhomogeneous, with plumes, threads and jets, as eleurly appearing on a good eclipse processed picture / 2 / / 4 / . The most popular model of a coronal hole based on the ATM-SkyLnb data, disagrees with this eclipse measurements / 2 / / 2 2 / a n d does not take into account inhomogeneities ; indeed when structures are considered, the problem with C.H. is completely open and a lot of efforts will have to be made in the future to resolve the discrepancies. A last special case is streamers which are very easy to study at eclipses.

5 0 / 0 -

0

500.

I00

50

10

5

1,

.5

v [krn.sec -I ] V : vs~ (~. k T/IJ.m H)1/2

_ ~ _ . = = ~ . - - _ - - 2 Mk

,o: ,;" fS-- ....

,¢- ,, /

10 6.

2 3 4 5 6 7

Fig. 6. In thick line, electron densities measured for a large s t reamer /1~ /a t the 1970 eclipse and in thin line and dot~, the velocities of outward motion deduced from the continuity equation ; in dashed lines are the veloc- ities given by the Parker's model for different tempera- tures and a homogeneous distribution of Ne.

Fig. 5. Polarization degree PK of the K-corona alone mea~sured near the time of sunspot mini- mum (1973) and at sunspot mazimum of activity (1980) by S. Koutchmy et al. / 1 ~ / and J. Df=rst /13/

Coronal Physics from Eclipse Observations (4)35

The contribution of a streamer which is close to the plane of the limb (A _ Ao) can easily be measured as s large modulation A I of the asimutal distribution of intensities ; measuring the corresponding polarization degree, it is also possible to evaluate I when the effective thickness AI along the 1.o.s. of the streamer is not too large. Therefore, (6) can be rewritten in a more simple form:

z~(p) = - ~ ~ w~(,) No(,) za (7)

A full analysis of such case was presented i n / 1 4 / . There, the distribution of Ne in the cross-section of the streamer is given by:

Ne : cte e z p - [(z 2 + ~/y2) 0.-2] e z p - (~z) (8)

where T/describes the ratio between the extension along the l.o.s, and the thickness observed in the plane of the picture and o" is the equivalent cross-section. Because a large streamer is made of several overlapping parts (sheets), it is impossible to consider in detail the distribution of Ne, so s gaussian shaped cross-section with ~ _~ I is s good first approximation. The deduced densities along the axis of the streamer are shown on Figure 6 ; a remarkable change of the behaviour of N~ is detected in the range r = 3 to 3.5 which corresponds to the smallest values of the effective cross-section 0.. Assuming this part corresponds to the crossing of the sound velocity by the outwardly expanding gas "inside" the streamer (by analogy with the Laval-nozzle) and using the equation of mass conservation:

No(,) 0.(r) ~ ( , ) = cte (9)

we readily deduce v(r), see Figure 6. An increasing velocity, up to the limit of visibility of the stalk of the streamer, is obtained ; therefore, no curvature is predicted nor observed with the present precision. This procedure was repeated more recently on the example of a narrower and longer streamer observed at the 1973 eclipse and the deduced velocities were given in /24/ . Again, no curvature is observed. However, the deduced velocities axe definitely decreasing after r : 5 and taking into account the position angle of the streamer, a curvature would have been observed if the gas was expelled freely and the magnetic field frozen in. That is clearly not the case, so a modification of the model should be considered, see /8 / , implying electric currents at each side of the streamer after r ~ 3.

FINE STRUCTURE

Although the heterogeneity of the quiet solar corona has been recognized for a long time, no concrete account of it have been made in the theory of the acceleration of the solar wind nor in the theories of extended coronal heating. Theoretical works are essentially based on time and space averages. A good coronal eclipse picture / 2 / / 4 / m a k e s it clear, however, that the coronal plasma is indeed confined in fine scale structures ; eventually, these structures clump together locally to form a far larger scale structure like a coronal streamer. In that sense we say the corona is striated with fine scale elongated structures. The physics of the small-scale structures is obviously different: taking into account the filling factor, densities increase dramatically and their tangential gradients reach high values suggesting discontinuities ; the structures can be detached or twisted. Finally, fine scale structures can interact, cross one each other, pointing to the reconnection - disconnection phenomena occuring at even smaller scales. The study of these small-scale structures is still rather frustrating ; there are several reasons for this, namely:

a. The spectrum of fine scale inhomogeneities is unknown. Moreover, no statistical account can be furnished due to obvious geometrical superposition effects, although the structures (often unresolved !) are preferentially stretching radially, at least for r ~ 1.5.

b. The absence of a general approach to this problem makes difficult any inference to the more general problem of stellar coronac and stellar winds. Up to now, only coronal loops and the rather extended streamers and coronal holes have received a t t e n t i o n / 8 / / 1 5 / .

c. Optical observations, which potentially could give the best spatial resolution, are mainly of "amateur- size" apertures, giving essentially unresolved observations. Moreover, the modern detectors like CCD cameras are not yet in use in coronal imagery. The last July 11, 1991 solar total eclipse has been an exception because of the unique possibility to use the best optical telescopes in the world and modern detector-systems at Mauna Kea (Hawaii). Results are not yet fully analysed but s few remarkable measurements are available, confirming the occurrence of very fine structures in the corona.

Before going further in studying a few cases of density heterogeneities, let us consider temperature variations in the corona. It is well established that essentially 2 main classes of emission lines exist (we do not consider the class I l l lines formed at T > 3 MK which exist only in very active region and micro-flaxes): class I corresponds to T : 1 to 1.5 MK (typical lines are those of Fe IX, X, XI) and class If, corresponding to T = 1.5 to 2.5 MK (Fe XIV, Fe XIII, Fe XV). On Figure 7, we display the emissions observed at the 1981 eclipse in

(4)36 S. Koutchmy

the green Fe XIV line and in the red Fe X line. Pictures were specially processed to increase the visibility of structures (octo-directional maximum of secondary derivatives): they clearly demonstrate the absence of obvious correlation between these temperature sensitive emissions, demonstrating the inhomogeneities in temperature. Examining the simultaneously observed picture in WL, which gives the distribution of N,., we however noticed a good correlation with the Fe XIV picture (T _~ 2 MK) which seems to be the most probable temperature in the corona outside active centers. This result was already appearing on Figure 4, when the hydrostatic temperature is computed over the main features of the inner corona, excluding polar rays. We conclude that the true temperature in the corona is around 1.5 to 2 MK and does not change by more than a factor 1.5 although producing large effects on the resulting emission of different ions ; however, the local pressure is not greatly affected by these changes of temperature, because density inhomogeneities are far more important.

Fig. 7. Structure of the inner corona observed at different temperatures: at left, in the 530.3 nm line of Fe X I V formed near ~ MK and at right, in the 637.4 nm line of Fe X formed near 1.1 MK. Pictures were simultaneously obtained by J. Sykora at the 1981 eclipse and subsequently processed at NSO-SP Lo remove the background.

Coronal Sheets and Tangential Discontinuties

When a good resolution coronal WL eclipse picture is examined, several slightly curved quasi-radial coronal sheets can be identified. Almost systematically they are parts of a large coronal streamer. Almost inevitably they show, from one side, the very characteristic signature of a coronal discontinuity: a large and definite jump of Ne /16/ ; Koutchmy identified these jumps of N~ with a tangential discontinuity, as defined in plasma physics. Let us summarize the characteristics of a coronal discontinuity (D), as obtained from the analysis of a characteristic case of a coronal sheet of known radius of cu rva tu re /16 / (no te that we do not consider a sheet exclusively "fiat") :

TABLE 1 Tangential Discontinuities: Typical Values using a 5 arcsec Resolution Picture

r[R@] 1.50 2.0 2.40

N[e- /cm a] 1.1 × 10 s 1.7 × 107 0.4 × 107

61[Mm] 4.0 4.5 5.0 V± n[e - /cm 4] 0.25 0.027 0.006 Vrn[e- /cm 4] 0.77 > 10 -3 10 -'~ 4 x 10 -5

ratio VX/Vr > 325 > 270 > 150 ABII [gauss] 0.8 0.25 0.12 ~/~wd [10 6 K] 1.25 1.0 0.8

w t ~ c RL -- [cm] 14 40 80

eB H

61 is the full observed "width" of the discontinuity ; it is close to the resolution of the picture, so the jump of densities at the boundary could be higher. However, the deduced numbers are already very significant: the gradient of densities V± across D is several orders of magnitude larger than the radial gradients ! Often an emptiness or "void" seems to exist along D, which makes the jump even larger. The strength of the magnetic field was calculated from the conditions needed to have a stable D ; the stability has been confirmed at least in the intermediate corona, by comparing eclipse pictures obtained at several hour intervals.

Coronal Physics from Eclipse Observations (4)37

We assume the temperature is constant across D; we note that the gyro-rsdius of e - : RL << 61 and that the evolution of D is 'very slow. If Nt and N2 are densities (N1 >> N2) at each side of D, v is the velocity and B the magnetic field, we have:

v± = 0 ; rill # ell2 ~ (10)

Ba. = 0 ; BIO :f: BII2 ; N1 # N2 J

The current j± across D is also null because we have no flux of particules across D (Nl v_Lj = N2 V.L2 = 0), but a bulk displacement of the whole D is not excluded. /°1 and P2 being the pressure at each side we finally have:

P1 + B21 = P2 + B~2 (11) 8a" 8a"

Taking B.LI ~ 0, ~ = 2Nej k Thyd, we get the value BII given in the above table which is probably the best estimation of the true magnetic field in the intermediate corona. A more sophisticated analysis would include the equation j = 1/4w V x B.

Rays, Threads and Plasmoids

These structures of the inner and of the intermediate corona are exclusively observed on very high resolution eclipse pictures. They seem to be an obvious coronaS counterpart of localized transition region phenomena occurring everywhere : above miniature ephemeral active regions that presumably overly enhanced chro- mospheric network regions (although inner network phenomena cannot be completely excluded), or above filaments, or in and especially around sunspots active regions or finally above polar regions (fine structure of polar plumes or rays). There are early reports of visual detection of the inner coronas fine jets ; the first quantitative works were done in 1968-73 / 1 7 / a n d since that time, several new measurements were performed /18/conf i rming the early theoretical predic t ions/19/ . From the analysis of several excellent plates obtained at past eclipses, we got measurements /18 / shown on Figure 8. The most convenient parameter to report is the transverse photometric "size" (or FWHM) which should be comparable to the extension along the l.o.s. ; then using formula (7), Ne can be obtained. From the estimation of the cross-section ~ro in Figure 8, with Al ~ FWHM and using equation (9), we get the variation of velocities Vo shown at the bot tom ; absolute values are deduced assuming they are close to the observed proper motion. This is rather arbitrary, because proper motions are very difficult to measure and emission lines line-profile analysis gives results which are

70•90

difficult to interpret.

;log (~D[arc sec 21

J J

Ne J

t i R o l 4 3 2 1.5 I I I I

IogV[km se¢ -1]

I 400 300 200 100

50

v=

log Ne [cm -3]

(Ne CFHT/

(Ne) Ne /

I I ~73 I~I L. l u - I

i ° 1 L _ J

.i011

.io Io

.!0 9

.I08

1.25 1.10 1.0 I I I

Fig. 8. Diagram displaying diffe- rent parameters deduced from the analysis of fine scale coronal struc- tures coming from different eclip- ses (indicated by the year of the ~Oth century). Bores correspond to a measure of the cross-section in aresec 2 (left ordinate) and the dashed line gives the correspond- ing estimated densities ; at ~he bot- tom are shown radial velocities de- duced from the continuity equation and velocities (with error bars) d- educed from proper motion mea- surements.

~ 81 pole

VI CFHT vo

(4)38 S. Koutchmy

The 1991 eclipse experiments at CFHT brought completely new results which are not yet fully evaluated. Surprisingly, small-scale structures at sub-arcsec scales showed up as a highly variable phenomenon. Elon- gated coronal threads of < 1 Mm diameter are well detected above equatorial active regions at more than 100 Mm from the surface, with a life-time of order of 100 sec ; rays with a larger cross-section have larger life-time. Proper motions are obvious on threads but they seem to fade out and the background is changing. A larger event than threads was also observed at CFHT, namely an isolated cloud of coronal plasma fading out in the ambiant corona with a lot of dynamical associated phenomena/20/ . Precise measurements of the position of the center of gravity can easily be performed over the time sequence of more than 6000 video- CCD frames. However, only the component of the proper motion projected on the plane of the limb is seen. Additionally, the cloud which is of 3 arcsec size at the beginning of the video time sequence is splitting-up several times and ephemeral thread-like structures are produced. Ultimately the cloud disappeared (at the end of the sequence), after 200 see of time. Values of the measured proper motions vto~ in the plane of the sky and also radially Vr were reported /24/ . This observation of a coronal plasmoid and dynamical associ- ated phenomena needs a more thorough analysis but already indicates that an important part of the corona (if not all the corona !) is in a dynamical state at sub-arcsec and resolution. Ejected plasma clouds are initially d i a m a g n e t i c / 2 1 / a n d because we see them fading out in the background corona, MHD small-scaie phenomena should be operating and are probably responsible for the heating.

CONCLUSIONS

Future eclipse observations could still be a good complement to the planned program of space coronal observations on Koronas satellites and especially on SOHO.

Because of the good to excellent photometric precision performed at total eclipses, an absolute calibration for a few space observations can be provided during a short interval. Good spatial resolution observations need the use of a radial neutral filter. However, future eclipse observations should now be focused on experiments which are difficult to perform on a spaceborne platform: line profile analysis of line emissions in the intermediate corona, IR analysis including polarimetry, etc. It is not clear if eclipse observations can yield more on the analysis of very small-scale phenomena like the plasmoid phenomena observed in 1991, because not only a large aperture telescope is needed, but also a sub-arcsec seeing should prevail ! Accordingly, spaceborne observations free of seeing effects could be more effective when considering that problem. A large aperture groundbased mirror coronagraph /23 / would however permit these observations ; IR adaptive optics system should be a part of that instrument ; such instrument in space could be the ult imate for analysing sub-arcsec structures.

ACKNOWLEDGEMENTS

Results presented here are mainly coming from collaborations that I benefited from many scientists whose names partly appear in the references and to whom I am indebted. The typing was diligently performed by Valdrie Demallly and Colette Douillet.

REFERENCES

1. C. Lebecq, S. Koutchmy and G. Stellmacher, The 1981 Solar Total Eclipse: II. Global Absolute Photo- metrical Analysis, Astron. Astrophys. 152, 157, (1985)

2. S. Koutchmy, Study the June 30, 1973 Trans-polar Coronal Hole, Solar Physics 51,399 (1975)

3. M. Loucif and S. Koutchmy, Solar Cycle Variations of Coronal Structures, Astron. Astrophys. Suppl. 77, 45 (1989)

4. S. Koutchmy, Solar Corona, in: Illustrated Glossary for Solar and Solar- Terrestrial Physics, Eds Bruzek and Durrant, Reidel, 39 (1977)

5. P. Ambroz, The Coronal Magnetic Field - Numerical Expansion Methods and Observations, in: Solar Magnetic Fields and Corona, vol. 1, Proc. XIII Consultation Meeting on Solar Physics, Nauka, Sib. Div. 138 (1989)

6. T.J. Hoeksema, Extending the Sun's Magnetic Field through the Three-dimensional Heliosphere, Adv. Space Res., voi. 9, 4, 141 (1989)

7. B.V. Somov and S.I. Syrovatsky, Appearance of a Current (neutral) Sheet in a Plasma Moving in the Field of a 2-D Magnetic Dipole, J.E.T.F. 61, 5, 1864 (1971)

8. S. Koutchmy and M. Livshits, Coronal Streamers, Space Science Rev., in press (1992)

Coronal Physics from Eclipse Observations (4)39

9. S. Koutchmy, M. Loucif and O. Koutchmy, Relations between Large-scale Coronal Structure, Chromo- spheric Filaments and the Solar Cycle, in: Proceedings STP Workshop, Meudon June 1984, P. Simon Ed. (1085)

I0. S. Koutchmy and P. Lamy, The F-corona and the Circum-solar Dust: Evidences and Properties, in: Proceedings IAU Coll. N ° 85, Giese and Lamy Eds, Reidel, 83 (1985)

11. K. Saito, A non-Spherical Axisymmetric Model of the Solar K-corona of the Minimum Type, Annals of the Tokyo Astron. Oks. XII, 2, 53 (1972)

12. S. Koutchmy, J.P. Picat and M. Dantel, Polarimetric Analysis of the Solar Corona, Astron. AstropAys. 59, 349 (1977)

13. J. Dfirst, Two Colour Photometry and Polarimetry of the Solar Corona of 16 Feb. 1980, Astron. Astrophys. 112, 241 (1982)

14. S. Koutchmy, Etude Hydrodynamique du Grand Jet Coronal observd/t l'Eclipse du 7 mars 1970, Solar Physics 24, 374 (1972)

15. J.B. Zirker, Coronal Holes and High Speed Wind Streams, Colorado Assoc. University Press (1977)

16. S. Koutchmy, Un ModUle de Grand Jet Coronal avec Renforcement de REgion Active, Astron. Astrophys. la , 79 (1971)

17. S. Koutchmy and G. Stellmacher, Photometric Study of Chromospheric and Coronal Spikes Observed during the Total Eclipse of 30 June, 1973, Solar Physics 49, 253 (1976)

18. S. Koutchmy, Small-scale Coronal Structures, in: Solar and Stellar Coronal Structure and Dynamics, Proceedings of the NSO/SP 9th Workshop, R.C. Altrock Ed., 208 (1988)

19. H. Alfven, On the Filamentary Structure of the Solar Corona, in: The Solar Corona, IAU Symp. 16, J. Evans Ed., Acad. Press, 35 (1963)

20. J.C. Vial, S. Koutchmy and the CFHT Team, Evidence of Plasmoid Ejection in the Corona from 1991 Eclipse Observations with the CFHT, in: Proceedings of an ESA Workshop on Solar Physics and Astrophysics at lnterferometry Resolution, 87 (1992)

21. G.W. Pneuman, Ejection of Magnetic Fields from the Sun: Acceleration of a Solar Wind containing Diamagnetic Plasmoids, Astrophys. J. 265, 468 (1983)

22. F. Crifo-Magnant and J.P. Picat, A Density Model for the N-Polar Coronal Hole at the 1973 Eclipse, Astron. Astrophys. 88, 97 (1980)

23. S. Koutchmy and R. Smurtt, High Resolution Observations of the Solar Corona: Why and How?, in: Proceedings I0 th NSO-SP Workshop, O.v.d. Luhe Ed., 560 (1989)

24. S. Koutchmy, Streamer Eclipse Observations, in "Ijt SOHO Workshop", Annapolis, Aug. 25-28, 1992, ESA SP (1992)