Off-limb EUV observations of the solar corona and transients with the CORONAS-F/SPIRIT telescope-coronagraph

The SPIRIT telescope aboard the CORONAS-F satellite (in orbit from 26 July 2001 to 5 December 2005), observed the off-limb solar corona in the 175 Å (Fe IX, X and XI lines) and 304̊A (He II and Si XI lines) bands. In the coronagraphic mode the mirror was tilted to image the corona at the distance of 1.1...5 Rsun from the solar center, the outer occulter blocked the disk radiation and the detector sensitivity was enhanced. This intermediate region between the fields of view of ordinary extreme-ultraviolet (EUV) telescopes and most of the white-light (WL) coronagraphs is responsible for forming the streamer belt, acceleration of ejected matter and emergence of slow and fast solar wind. We present here the results of continuous coronagraphic EUV observations of the solar corona carried out during two weeks in June and December 2002. The images showed a “diffuse” (unresolved) component of the corona seen in both bands, and non-radial, ray-like structures seen only in the 175 Å band, which can be associated with a streamer base. The correlations between latitudinal distributions of the EUV brightness in the corona and at the limb were found to be high in 304 Å at all distances and in 175̊ A only below 1.5Rsun. The temporal correlation of the coronal brightness along the west radial line, with the brightness at the underlying limb region was significant in both bands, independent of the distance. On 2 February 2003 SPIRIT observed an expansion of a transient associated with a prominence eruption seen only in the 304 Å band. The SPIRIT data have been compared with the corresponding data of the SOHO LASCO, EIT and UVCS instruments.


Observations of the solar corona by wide field instruments
The solar corona is routinely observed now with two types of instruments: the temperature sensitive, space-based Xray and EUV telescopes (e.g. SOHO EIT, Delaboudinière et al., 1995), to study the inner corona up to R∼1.3 (hereafter R is the distance from the solar center in the units of the solar radius) and the electron density sensitive space and ground-based WL coronagraphs to study the outer corona above R=2. Excellent examples of ground-based observations of the corona, starting from the limb level during the solar eclipses can be found elsewhere (Koutchmy, 1994;Wang et al., 2007) but the solar eclipses are infrequent and last only a few minutes. Most of the data have been obtained up to now with the SOHO LASCO C2 and C3 coronagraphs (Brueckner et al., 1995). The LASCO C1 coronagraph observed the corona in the green line in the intermediate range R=1.1. . . 3.2 (Schwenn et al., 1997) only at the solar minimum in 1996-1998. Recently, the upper limit of the regular coronal observations with the EUV telescopes was increased to R=1.7 (STEREO SECCHI EUVI, Wülser et al., 2004) and the lower limit for the space-based WL coronagraphs reduced to R=1.3 (STEREO SECCHI COR-1, Thompson and Davila, 2007). The ground-based MK4 coronagraph at Mauna Loa Solar Observatory is able now to observe the corona from R=1.1 (Burkepile et al., 2005).
In numerous observations it was determined that the inner corona consists of a "diffuse" part, possibly formed by many unresolved loops and well described by a hydrostatic approximation, and bright, raylike or threadlike axisymmetric and non-axisymmetric structures of different spatial scales, forming a base of streamers (e.g. Koutchmy, 2001;Koutchmy and Molodensky, 2005). The structure and local parameters of the coronal plasma in the EUV range at distances below R=1.6 were studied by many researchers (e.g. Wilhelm et al., 1998Wilhelm et al., , 2002Warren, 1999;Parenti et al., 2000;Li et al., 2000;Cirtain et al., 2006;DeForest, 2007, and others). The parameters of the outer corona above R=2, mainly the electron density and its 3-dimensional distribution, were retrieved from the observations with the WL coronagraphs (e.g. Lamy et al., 1997;Quemerais and Lamy, 2002;Saez et al., 2005). The intermediate region between R=1.6 and 2 has not been studied in detail, in particular, in the EUV, although it is essential for theoretical modelling of streamers and solar wind generation. Besides the problem of physical matching of the EUV and WL data, it seems to be important to extend the X-ray and the EUV measurements from the disk to at least R=2. . . 3, with similar spatial and temporal resolution to watch spatial and temporal variations of the plasma structure and the development of Coronal Mass Ejections (CMEs).
The ability of an ordinary EUV telescope to register a coronal brightness at the distances larger than R=1.3 is limited by the dynamic range of the detector based on Charge Coupled Devices (CCD), which is typically about 4.5 orders of magnitude, and the straylight level (∼1% of the mean intensity at the disk). Delaboudinière (1999) was the first to overcome the latter limitation and observed the corona with the EIT telescope in the equatorial plane to the distances R∼2.6. These observations were made while the spacecraft was in the offset position and the solar disk radiation was blocked by a mask in front of the detector.
The solar corona at R=1.3...12 is studied in the Far UV spectral range with the Ultraviolet Coronagraph Spectrometer aboard the SOHO observatory (Kohl et al., 1995;Raymond et al., 1997). This instrument has two UV spectral channels: the OVI channel operating in the range 945-1123Å (473-561Å in the second order), and the Ly α channel operating in the range 1160-1350Å (580-635Å). The images are obtained in the discrete spectral lines Ly α 1216Å, FeXII 1242Å, O VI 1032Å, Si XII 499/521Å, Mg X 610/625Å and some others in the slit-like fields of view (FOV) of 14 ×40 (Ly α channel) and of 28 ×40 (O VI channel). In the standard synoptic mode a panoramic view of the corona is constructed by a series of discrete exposures at 8 angular and up to 12 radial positions with the total observational time of ∼20 h. UVCS is a very effective diagnostic device for the analysis of local ionization equilibrium or line-of-sight velocity distributions, but cannot study spatial and temporal dynamics of the coronal structures in a wide field of view.
The SPIRIT EUV telescope-coronagraph is a new instrument for wide-field observations of the solar disk and the corona in the 175 and 304Å EUV spectral bands (Zhitnik et al., 2002). The instrument is able to observe wide segments (more than 90 • in latitude) of the solar corona from the limb to R=5 with high spatial (5 ) and temporal (5 min) resolution, which has no analogues in the practice of the solar astronomy. The main task for the CORONAS flight was to prove the concept and the design of such instrument and to define its potential for studying the solar corona and tran-sients. In the period of high solar activity (2002)(2003) the SPIRIT EUV telescope operated in the coronagraphic mode during 15 observation sessions lasting from several hours to one week. The current report presents the main parameters of the instrument and some significant results of the SPIRIT observations in comparison with the data of SOHO LASCO, EIT and UVCS.

SPIRIT EUV telescope-coronagraph
2.1 Design of the instrument and the observation procedure One of the two SPIRIT EUV telescopes had two channels operating in 175 and 304Å. It had been designed according to the Herschel optical configuration which provides a good angular resolution and the largest possible efficiency due to only one reflection. The distance between the offaxis parabolic mirrors and the detectors in SPIRIT was three times larger than in the Ritchey Chretien telescopes with similar resolution (e.g. SOHO/EIT, SECCHI/EUVI), so the straylight in our case was at least one order less. In order to operate as a coronagraph, the telescope was provided with drivers to tilt the mirror up to +/−1.5 • from its nominal position (Zhitnik et al., 2003). The optical design of the SPIRIT telescope-coronagraph is shown in Fig. 1a. A driven lid in front of the input window was used as an outer occulter to block the direct illumination of the mirror by the intense solar disk radiation. The corona was observed through additional lateral optical windows. Both the central and the lateral windows have been covered with similar thin film filters. The detector had an image intensifier with a variable amplification, a lens and a CCD-array for the registration of visible images. The exposure time in the disk mode was 9 s, in the coronagraphic mode 300 s. Due to longer exposures and higher amplification of the detector, a sensitivity of the instrument in the coronagraphic mode was 143 times (in the 175Å band) and 316 times (in the 304Å band) more than in the disk mode. It should be noted that in both modes the exposures in the 175 and 304Å bands were strictly simultaneous.
Due to the telescope design restrictions, the occulters were mounted eccentrically to the telescope optical axis, so the viewing range in the coronagraphic mode depended on the current position of the FOV towards the instrumental axes. The FOV had the geometrical size of 46 ×52 . In total, the instrument was able to observe the corona from the limb to the radial distance of R=5. The orientation of the FOV with regard to the solar axes varied in time, due to a slow roll of the satellite (ordinarily the roll velocity was less than 3-4 • 1/day). The roll angle was not stabilised, but its value was recovered on ground using the attendant pictures of the sky taken by two star cameras in the SPIRIT assembly. The vignetting function of the coronagraph (Fig. 1b) had a steep rise from the limb to R=4, so it acted as a radial filter   analogous to that used in the WL coronagraphs, to equalize the coronal signal at different heights.
The observational cycle consisted of three steps. First, the mirror was tilted in order to place the FOV into the preprogrammed position relative to the solar disk (taking into account the current roll angle). Then the shifted image of the disk was registered in the disk mode with the open occulter, to control the real position of the FOV towards the Sun. Then the occulter was closed, and several images of the corona were taken in the enhanced coronagraphic mode. As a rule, images of the disk were taken every 2-3 h, between those from 4 to 12 coronal images were taken with a cadence of 15-30 min. Ordinarily, the images in the coronagraphic mode were transmitted to the Earth in the binned mode with the pixel size of 5.4 , with the total number of frames and the image format being limited by the available telemetry volume. Blurring of the coronal images produced by the drift of the satellite pointing system and variation of the roll angle during the exposure time did not exceed 10-15 .
2.2 Plasma temperature response functions for the 175 and 304Å bands Figure 2a and b presents the plasma temperature response functions (PTR) of the SPIRIT 175 and 304Å bands in comparison with the similar channels of the SOHO/EIT telescope (Slemzin et al., 2005). The SPIRIT 175Å channel had the PTR similar to that of the 171Å EIT channel. Calculations with the use of the CHIANTI code (Dere et al., 1997) showed that more than 80% of its input was provided by three lines of the Fe IX, X and XI ions (T max ∼0.6...1.5 MK). The PTR of the SPIRIT 304Å channel had two peaks: the major peak corresponded to the He II line (T max ∼80.10 3 K) and the minor peak at T ∼1...2 MK, associated with the coronal Si XI 303.3Å, Fe XV 284.2Å and Fe IX 171.1Å lines (the latter registered in the second reflection order of the mirror multilayer coating). Estimations with the help of the CHIANTI code showed that in quiet solar regions, more than 50% of the total signal in the SPIRIT 304Å band corresponded to the He II 304Å line. The response functions of the SPIRIT and EIT 304Å bands are similar except for the larger contribution of the hot ion lines, in our case, caused by a wider spectral function. The contribution of the hot component can be easily distinguished by a comparison of the 304 and 175Å images, just as Delaboudinière (1999) did for the EIT images.

Processing and calibration of the images
The images of the disk and corona were first pre-processed using a standard procedure which included a background subtraction, a correction of nonlinearity of the brightness scale (appeared at the highest amplification), and a flat-field correction. Using the data of the star cameras the images were rotated to compensate for the roll of the satellite, in order to fit the Y-axis of the image with the solar rotation axis.
In some cases the images of the corona were summed over the period of several hours for a better signal-to-noise ratio.
Then the images of the corona were superimposed on the nearest disk image to build a composite image. To perform a photometric analysis, the coronal images were calibrated by dividing the signal in each pixel to the value of the vignetting function according to its radial position and then the disk and coronal parts were matched at the distance of R=1.3. The angular distribution of the straylight was studied during several partial solar eclipses observed with SPIRIT from the CORONAS-F orbit. Figure 3a shows the images taken during the solar eclipse of 11 June 2002; Fig. 3b -the straylight radial functions for the 175 and 304Å bands. The straylight functions were sufficiently narrower and steeper than the measured radial brightness distributions of the corona. In the analysis presented here a contribution of the straylight was neglected.

Movies of the solar corona in the 175Å and 304Å bands
The In order to enhance the coronal structures, the disk and the coronal part of the images were presented in a logarithmic scale and separately scaled in intensity relative to the maximum brightness. Figure 4 shows the three selected frames from the movies demonstrating the common features observed in the corona.
In the 175Å band (Fe IX-XI lines) the corona contained a "diffuse" (unresolved), quasi-symmetric part and a structure of bright rays starting from the solar surface, some of them deviated from the initial direction of tens degrees at the distance R=1.3-1.5 (examples in Fig. 4: the radial rays are marked by the number 1, the non-radial ones by the number 2). Some very bright rays which evidently originated from the active regions appeared near the limb (e.g. a group of rays in the northeast sector marked by the number 4 in Fig. 4 In 304Å the corona was inhomogeneous and more "diffuse" than in 175Å. Only traces of the coronal rays seen in 175Å were faintly visible in this band. The brightness in the corona was higher when some bright regions appeared at the disk near the limb (or behind it). The rays seen in the SPIRIT 175Å picture (Fig. 5a and c) between the limb and the inner boundary of the LASCO FOV (R=2.3) marked by the numbers from 1 to 6 evidently have counterparts in the streamer structure seen in the LASCO images. It is worth keeping in mind that the coronal brightness in the LASCO images is proportional to the total electron density n e of the plasma, whereas in the 175Å band it is proportional to n 2 e of the plasma component with the temperature near 1 MK.

Comparison of SPIRIT images with LASCO and UVCS
The correlation between the latitudinal distributions of the coronal brightness in the SPIRIT 304Å and LASCO images ( Fig. 5b and d) is not high or even negative: some of the brightest features in 304Å at R=1.1. . . 1.5 correspond to the regions of weaker brightness in the LASCO picture at R>2.3 and vice versa. Figure 6 demonstrates a comparison of the SPIRIT and UVCS images for 16 June 2002: (a) the SPIRIT image in 175Å from the limb to R=1.6 superimposed on the UVCS image (R=1.6...3.5) in the OVI line (1032Å); (b) the SPIRIT image in 304Å superimposed on the UVCS image in the Ly α line (1216Å). In the first case only some of the rays seen in 175Å (e.g. features numbered as 1 and 2 in Fig. 6a) Fig. 4. Specific coronal features observed by the SPIRIT EUV coronagraph in the 175 and 304Å bands: 1 -radial rays, 2 -non-radial rays, 3 -a rising loop, 4 -a fan of rays linked with the active region at the disk.
excitation. Whereas the Fe ion lines of the 175Å band are excited by collisions, in the O VI line the collisional and radative components are of one order of value (Raymond et al., 1997). Besides, the collisional components in the Fe IX-X-XI and O VI images correspond to different excitation temperatures: T e ∼1 MK and 0.3 MK, correspondingly. The corona in the pair of the SPIRIT 304Å and the UVCS Ly α 1216Å images (Fig. 6b) do not show the structural elements seen in the SPIRIT 175Å image. The angular distributions of brightness in both cases are roughly similar: the brightness in the sectors numbered by 1 is higher than those in sector 2 in both bands. It corresponds to the suggestion that the coronal radiation in both cases is generated by the same mechanism of resonant scattering. The Ly α line is less sensitive to the line-of-sight velocity than the He II line (Kohl et al., 1995). in the 175Å band is the largest in the relatively dense closed loops localized above the active regions. These structures cannot be resolved with our spatial resolution (1 • bin corresponds to 1.2×10 4 km at the limb), but their averaged brightness is proportional to the brightness of the active regions at the limb. At larger distances the major brightness is concentrated in non-radial rays, so the correlation vanishes.
In the 304Å band the main mechanisms of excitation of Helium are: collisional excitation in the closed loops near the limb (probably enhanced by the non-thermal electrons or ionic diffusion) with the minor contribution of the radiative excitation by the photoionization-recombination (P-R) process (Zirin, 1975;Macpherson et al, 1999;Andretta et al., 2003;Judge et al., 2004). Above the limb this band also contains a contribution of the collisionally excited Si XI and Fe XV lines. In total, these components result in a positive correlation between the coronal brightness in the 304Å band with the brightness at the limb. At the distances R>1.5 the electron density is low, and the radiative excitation, namely, the resonance scattering becomes dominant (Delaboudinière, 1999), which also provides a positive correlation with the brightness at the limb. Figures 7b and 8b show temporal variations of the coronal brightness along the west equatorial line during the week 16 June 2002-22 June 2002, normalized to unity at each distance and their correlations with the limb distribution. The scan line does not cross any bright coronal rays and corresponds to the "diffuse" corona above the quiet solar region.
In contrary with latitudinal distributions, the temporal variations are well correlated with those at the limb in both bands, independent of the distance. This result confirms that the corona at the distances R=1. . . 2 is closely linked with the brightness distribution in the underlying disk regions, varying with the solar rotation.

Radial distributions at the equatorial plane
We analyzed the radial distributions of the EUV coronal brightness in the equatorial plane using the SPIRIT data of 16 June 2002 described in the previous section. Figure 9 shows the normalized radial distributions in the SPIRIT 175 and 304Å bands (relative to the brightness at the limb) compared with the EIT data taken from the paper of Delaboudinière (1999). These data were obtained on 4 April 1996 when the SOHO spacecraft was in the off-set position, and presented as normalized radial brightness distributions in the Fe ion lines (an averaged distribution of practically coincident data in the 171, 195 and 284Å bands) and in the 304Å band along the east-west line in the eastern direction. It should be noted that the SPIRIT data correspond to the maximum of the solar activity, whereas the EIT data were taken at the minimum of the activity.
There are three important conclusions which result from Fig. 9. 1. The normalized radial distribution in 304Å from the SPIRIT data coincides with the corresponding one in 175Å in the region R=1. . . 1.8. The EIT distributions in 304Å and in the coronal Fe lines also coincide in the region R=1. . . 1.2. It suggests that in these regions the emission in the 304Å band (which consists of the major He II line and accompanying Si XI and Fe XV lines) has the same dependence on the coronal electron density as the corresponding emissions in the Fe bands which are excited by electron-ion collisions. The region of predominantly collisional excitation of the He II line is larger at the solar maximum (the SPIRIT case) in comparison with that at the solar minimum (the EIT case) because of higher electron density in the middle corona at the solar maximum.
2. At the distances larger than R=1.8 (in the SPIRIT case) and R=1.2 (in the EIT case) the radial distributions in the SPIRIT and EIT 304Å bands are shallower than the corresponding Fe line distributions. At the distances of R>2 the SPIRIT and EIT 304Å distributions practically coincide. According to the assumption of Delaboudinière (1999), it means that in these regions the resonant scattering of the disk radiation dominates in the He II line emission of the corona.
3. A comparison between the SPIRIT and EIT radial distributions shows the variation of the coronal EUV emission in the equatorial plane with the solar activity: in the range R=1.1...2.5 the normalized distribution at the solar maximum sufficiently exceeds that at the solar minimum and approaches it at larger distances.
A detailed quantitative analysis of radial distributions of the coronal EUV emission from the SPIRIT data will be done in further works.

Observations of transients in the middle corona
During the SPIRIT coronagraphic observations we detected several transients associated with eruption processes and coronal mass ejections (CMEs). Simultaneous observations in two spectral channels associated with different plasma temperatures give unique information to test and improve the theoretical models of the eruption processes. As an example, Fig. 10 demonstrates the selected pictures of the CME after a failed prominence eruption observed on 2 February 2003 between 13:00 and 18:00 UT. To enhance the contrast of transients, the coronal parts are shown as the base difference images obtained by the subtraction from each image of the  reference image taken at 10:24:14 UT. Earlier, from 1 February 16:00 UT to 2 February 10:00 UT, a slowly rising prominence appeared in 304Å as a bright feature (marked by the number 1 in Fig. 10a). Between 06:00 and 10:00 the prominence returned back and produced a brightening at the lower boundary of the FOV (the feature 2 in Fig. 10b and c). After 13:29 UT a bright spot appeared at the distance R∼1.5 (the feature 3) which could indicate the beginning of the eruption process. At 16:56:07 this feature was transformed into a big, expanding loop (the feature 4 in Fig. 10c). After 18:30 the loop left the field of view at R∼2.3. The whole eruption process was seen only in the 304Å band and may be definitely related to the He II line, because in 175Å we can only see the traces of the rising loop. So we can conclude that the temperature of the erupting matter was sufficiently less than 1 MK. The LASCO CME catalogue described this event as a weak CME moving with the projected radial speed V =165 Km/s (in the linear fit). It should be noted that in 304Å, due to the Doppler dimming effect, we can detect only rather cold prominences and CMEs moving with the radial speed V ≤200 km/s (Labrosse et al., 2006). Instead, in 175Å a detection of transients is limited only by the temperature conditions: the plasma temperature must be of the order of 1 MK.

Conclusions
The SPIRIT EUV telescope-coronagraph is proved to be an efficient instrument to study the solar corona and coronal transients from the solar limb to several solar radii. These data fill the gap between the images of the corona obtained by the ordinary EUV telescopes and WL coronagraphs. The instrument is efficient to detect CMEs, eruptive prominences and other slow moving transients. Depending to its temperature, they can be observed in Fe ion or He ion line bands.