We present results of the study aimed to analyse the morphology of evolving coronal X-ray flaring structures. Our analysis deals with the Yohkoh SXT observations obtained during flares. In order to remove the telescope PSF blurring and improve the spatial resolution on the images we incorporated the ANDRIL maximum likelihood deconvolution algorithm. Using deconvolved images, we determined the distribution of physical conditions (Temperature-T, Emission Measure-EM) as they change with time for several flares. The results obtained indicate that the ''magnetic connectivity'' pattern is vividly changing as flare progress.
We put forward new ''hierarchical'' model of the coronal magnetic field organisation in order to explain the observed evolution of flaring structures.
It is known, that appropriate numerical
techniques may be used in order to remove the instrumental
blurring from the measured images. This problem has been
considered in detail in relation with known troubles of Hubble
Space Telescope mirror (White et al., 1993, 1994). We present
results of application of the maximum likelihood method of image
enhancement to the interpretation of soft X-ray images obtained
by SXT telescope aboard the Yohkoh Japanese solar satellite.
Applied deconvolution procedure (ANDRIL) provides X-ray images
with angular resolution below 1 arcsec as taken in different
energy bands. These deconvolved images have been analysed in
order to derive maps of distribution of basic physical
characteristics of flaring plasma. In the present study we
analyse flares observed close to the limb since it is easier to
distinguish in this case the vertical stratification of the X-ray
emission. As described in more detail by Sylwester at al.,
(1996), we deconvolved the instrumental point spread function
(PSF) from the original images recorded by SXT. As the PSF shape
depends substantially on the position of the source within
telescope field of view, we have used approximate formulae
derived by Martens et al., (1995). In the deconvolution, we
divided each of the original 2.45 // × 2.45// CCD
pixel into 25 sub-pixels increasing in this way the spatial
resolution. However, this increase in resolution causes sometimes
problem with an artificial ''noise'' showing up on deconvolved
images taken with short exposures. The problem occurs in areas
where signal level is of the order of dark current (out of bright
flare kernels). Such ''noise'' may lead to unreasonable values of
derived parameters. We discuss later, how to avoid this problem,
at least partly. In Fig. 1, we show sequence of deconvolved
images obtained through Be119 filter for 11 August 1992 flare.
One may recognize there the presence of white patches
corresponding to these ''noisy'' regions. In order to lessen
Figure 1: Example initial sequence of seven consecutive deconvolved Be119 frames for 11 August 1992 flare. The frames have been coaligned to within one sub-pixel (0.5//) using the satellite pointing data. The frame cadence is ~ 10 s. In the upper left corner, GOES plot for the corresponding flare is shown with the arrow pointing to the time when selected frames have been obtained. Dotted lines (here and in the other figures) correspond to the heliocentric coordinate system with the leftmost one representing solar limb. The intensity scaling (individual for each image) is very flat (power ~ 1/5) in order to bring up weak structures.
the ''noise'' problem, we decided to add (weighting by exposure
time) several (usually five) consecutive deconvolved images
obtained using particular filter. We coaligned the images
according to the Yohkoh satellite pointing file to within one
sub-pixel. This summed picture represents the average deconvolved
image (over time interval of several tens of seconds). The frames
getting into the sum we selected in such way, that the mean times
corresponding to Be119 and Al12 images differ by no more than few
seconds and therefore the analysis of their ratio is meaningful.
As pointed out by Siarkowski et. al., (1996), the determination
of emission measure and temperature maps from image ratios
constitutes a ''delicate'' task, especially in case of the SXT
measurements where the satellite axis had been pointing in
slightly different directions between exposures. It has been
shown in that paper that the most important is precise alignment
of images to within fractions of arcsec. The satellite pointing
data are usually not that accurate and additional pointing
corrections are necessary. Appropriate alignment is even more
important when deconvolved images are to be compared
because of the smaller size of the sub-pixel. Therefore, before
deriving EM and T maps, we coaligned precisely
Be119 and Al12 images using multiplicative method (for details
see Siarkowski et. al., 1996). Usually the applied additional
shifts are of the order of fractions of arcsec. Ratios of
Be119/Al12 signals we have interpreted using standard isothermal
approximation and as a result two characteristic parameters (T,
EM) have been derived for each sub-pixel. In Fig. 2
and 3 we present sequences of high-resolution maps of emission
measure for two limb flares. The important result of described
analysis
Figure 2: The sequence of ten emission measure maps for 11 August 1992 flare. The frames have been precisely coaligned as discussed in the text. The scaling of images is again very flat. The frame next to GOES represents pre-flare situation as seen on the original Be119 image.
procedure is that the ''noise'' problem is mostly removed. It
comes out that the detailed morphology of X-ray sub-arcsec flare
structure is richer, smoother and more detailed on the emission
measure map as compared to that seen on summed image obtained
from a given filter. In the present paper we investigate
evolution of flare morphology basing on the emission measure
maps. Discussion of behaviour of the temperature for selected
kernels is given in a parallel research by B. Sylwester and
J. Sylwester (1998).
We have analysed the evolution of deconvolved high-resolution intensity maps and emission measure distributions using time-lap animation. There are number of conclusions coming out from such analysis as concerns several limb flares investigated:
Figure 3: The
same as in Fig. 2 for 30 November 1993 flare, except that
instead of GOES (unavailable), the integrated (over the frame
area) time variation of the Be119 flux is plotted. Arrows point
to the kernels (FK, CK) for which physical conditions have been
analysed by B. Sylwester and J. Sylwester, (1998).
Figure 4: Proposed hypothetical loop hierarchy structure of magnetic fields in the corona.
As a result of the analysis of these observations we came to the conclusion that classical ''magnetic connectivity'' pattern, with foot-points of the coronal loops ''permanently anchored'' to particular magnetic flux tube emerging from below the photosphere does not fit to the above findings. It seems necessary to reconsider this ''classical'' point of view.
Based on the present morphological study, we introduce a new ''cartoon'' of connectivity pattern for the field lines in the corona. This cartoon is shown in Fig. 4. We call it the ''hierarchical model''. In this model we postulate the existence of a hierarchy of loop systems in which larger loops are rooted somewhere at the ridges joining tops of low-lying arcades connecting the opposite magnetic polarities. This low-lying arcades, perhaps, yet again, follow this organisation on even smaller scales (fractal structures?). The ultimate smallest flux-tubes seem to be those which emerge from below the photosphere as the ephemeral regions within the plage domain of one dominant polarity. Such a picture may get support from the NIXT observations (Golub at al., 1990; Herant et al., 1991) where compact X-ray loops are observed within the particular Ha ribbon. The uppermost loop systems related to active region and/or the complex of active regions (possibly showing up as coronal helmet structures) are tighted up by solar wind. During the flare (or micro- or nanoflare), through the process of reconnection of magnetic field lines inside multiple turbulent reconnection kernels (cf. Jakimiec at al., 1998), the overall magnetic field pattern simplifies causing stepwise relaxation of the magnetic surplus energy.
We are grateful to Mrs. A. Kepa for preparing videos and figures related to this research. This work has been supported by a KBN Grant 2.P03D.006.10 of Polish Committee for Scientific Research.
Golub, L., Herrant, M., Kalata, K., Lovas, I., Nystrom, G., Pardo, F., Spiller, E. & Wilczynski, J. 1990, Nature, 344, 842 Herant, M., Pardo, F., Spiller, E. & Golub, L. 1991, , 376, 797 Jakimiec, J., Tomczak, M., Falewicz, R., Phillips, K.J.H. & Fludra, A. 1998, , submitted Martens, P. C., Lemen, J. R. & Acton, L. W. 1995, , 157, 141 Siarkowski, M., Sylwester, J., Jakimiec, J. & Tomczak, M. 1996, Acta Astronomica, 46, 15 Sylwester, J., Sylwester, B. & Siarkowski, M. 1996, , Magnetic Reconnection in the Solar Atmosphere, ed. R.D. Bentley & J.T. Mariska, 244 Sylwester, B. & Sylwester, J. 1998, JOSO Annual Report 1997, submitted White, R. L. 1993, STS Image Reconstruction Newsletter, 11 White, R. L. 1994, Restoration of HST Images and Spectra II, 104
BACK
Page created by Jaroslaw Bakala (
jb@cbk.pan.wroc.pl)
Web Curator:Janusz Sylwester (
js@cbk.pan.wroc.pl)
File translated from TEX by TTH, version 1.57.