SUMER spectral observations of postflare supra-arcade inflows
D.E. Innes ([email protected])
Max-Planck-Institut für Aeronomie, Max-Planck-Str. 2, 37191 Katlenburg-Lindau, Germany
D.E. McKenzie
Department of Physics, Montana State University, P.O. Box 173840, Bozeman, Montana
59717-3840
Tongjiang Wang
Max-Planck-Institut für Aeronomie, Max-Planck-Str. 2, 37191 Katlenburg-Lindau, Germany
Abstract. On 21 April 2002 a large eruptive flare on the west limb of the Sun developed
a bright, very dynamic, postflare arcade. In TRACE 195 Å images, a series of dark, sunward
moving flows were seen againt the bright extreme ultraviolet (EUV) arcade. SUMER obtained
a series of spectra of the dark EUV flows in the lines C , Fe  and Fe  at a fixed position
above the limb. These spectra give spatially resolved line-of-sight velocities and emission
measures for the arcade plasma over a temperature range 2 × 104 to 107 K. The flows are dark
in all SUMER lines. The UV continuum longward (∼ 1350 Å) and shortward (∼ 675 Å) of the
Lyman limit is used to determine the amount of absorbing material with temperature below
2 × 104 K. There is some evidence of cold absorbing material just before the arcade emission
reaches the height of the SUMER observations. Along most of the dark channels there is no
change in continuum ratio and we therefore conclude, as originally suggested by McKenzie
and Hudson (1999), that they are plasma voids.
Keywords: Sun: flares – UV radiation
1. Introduction
Large postflare soft X-ray arcades (supra-arcades) rising over 200 arcsec into
the solar corona have been seen and described for a number of long duration
solar flares. The arcades look like fans of coronal rays directed outwards into
the corona from the top of what appears to be a typical post-flare loop system. In several of the supra-arcade events sunward moving structures have
been seen in SXT images (McKenzie and Hudson 1999; McKenzie 2000).
All events with sunward flow are associated with large coronal mass ejection
(CME) events. The majority of these flows look like dark trails falling from
the corona through the flare arcade toward the Sun. They seem to slow and
stop as they reach the height of the main arcade flare loops. From their appearance and their association with CMEs one may think that the sunward flows
are ejected chromospheric material falling back to the Sun; however so far
not a single one of them has been seen in Hα or cold line images. McKenzie
and Hudson (1999) and McKenzie (2000) suggested, after analysing soft Xray images, that they are low density flux tubes contracting down to their
equilibrium position.
c 2003 Kluwer Academic Publishers. Printed in the Netherlands.
downflow.tex; 13/03/2003; 13:37; p.1
2
On 21 April 2002, a spectacular postflare coronal arcade of an X1.5 flare
was observed on the south west limb of the Sun. It was also associated with
a large CME. The flare and CME were observed by many instruments and a
number of papers on the event have already been written (Wang et al. 2002;
Caspi et al. 2002; Gallagher et al. 2002). RHESSI images detected soft Xray emission from the disk, AR9906, at the start of the event. The CME
was very fast. It was seen in LASCO C2 and by UVCS and had an outward
speed of about 2500 km s−1 . TRACE obtained high cadence images in 195 Å
of the flare arcade dynamics and these have been coaligned to the SUMER
spectra by Wang et al. (2002) to reveal a 3-D view of some of the flows in
the flare corona. The evolution is very complex. Several large, hot loops are
seen expanding into the corona. The arcade is seen to move and restructure
in response to both disk and coronal activity. Dark sunward moving trails are
seen against the bright EUV arcade emission about 30 min after the first flare
X-rays.
The spectrometer SUMER was pointing at a fixed position in the corona
above the active region when the flare erupted and obtained a time series of
spectra in the lines C , Fe  and Fe . These three lines are formed at
approximately 4 × 104 , 106 and 107 K respectively and thus cover a very wide
temperature range. If the dark EUV flows are cooling loop material one might
expect to see them in the C  line, or if they are associated with reconnection
jets then they could show up as Doppler shifts in Fe  or Fe . Therefore
a careful analyis of the spectra has the potential to unveil the origin of the
dark flows. The cadence was also short enough to see different sections of
the dark inflows as they cross the position of the SUMER slit. This paper
presents a first analysis of the SUMER spectra at the time of the inflows
and is directed at resolving the cause of the darkness: dense, cold absorbing
plasma or plasma voids.
In an accompanying paper (Innes et al. 2003) a second aspect of the spectra, 1000 km s−1 flows in the Fe  line associated with the dark inflows,
is reported. A third paper is planned to discuss the structure and oscillations
seen in the TRACE images of the dark inflows. Various models of the inflows
have been suggested by McKenzie and Hudson (1999) andMcKenzie (2000).
These and other alternatives will be investigated in more detail in a fourth
paper.
2. Observations and data analysis
The SUMER spectrograph was pointing at a 4 × 300 arcsec2 long narrow
strip of the solar corona about 100 arcsec above the limb centered at Sun coordinates (981,-184). Stigmatic images were taken of the C  1335 Å (T ∼
4 × 104 K), Fe  1349 Å (T ∼ 106 K) and Fe  1354 Å (T ∼ 107 K) lines
downflow.tex; 13/03/2003; 13:37; p.2
3
2
(ph/s/cm /sr/em)
Intensity per emission measure
Spectra of postflare inflows
Fe XII
Fe XXI
C II
Log
Temperature
Figure 1. Line intensity contribution functions for the lines observed by SUMER. Ionization
equilibrium and Sun coronal abundances have been used.
with a cadence of 50 s. The contribution functions computed using CHIANTI
(Dere et al. 1997) of the three lines are shown in Fig 1. For most of the time,
only 2 Å centered on the lines were transmitted. Each hour a full spectral
window from 1333-1373 Å was transmitted. Direct measurement of the line
intensities can give constraints on the plasma emission measure at various
temperatures using a model of the ionization and the element abundances.
The line shifts give constraints on the plasma flow velocities.
The continuum emission around each line is a combination of first and second order radiation and may provide valuable clues to the origin of the dark
EUV flows. In the wavelength region observed, the first order (∼ 1350 Å)
emission is above the Lyman limit and the second order (∼ 675 Å) is below.
So second order photons are absorbed by H  along the line-of-sight and the
first order not. Thus, if there is cold material along the line-of-sight, this may
show up as decrease in the second to first order continuum ratio.
The two orders can be disentangled by using the known detector response.
As described by Wilhelm et al. (1997), the central part of the SUMER detector is coated with KBr. This essentially increases the detector sensitivity to
photons by about an order of magnitude at wavelengths greater than 1100 Å.
For example, at 1350 Å the KBr is almost 20 times more sensitive than the
downflow.tex; 13/03/2003; 13:37; p.3
4
Fe XXI
C II
Sun-Y (arcsec)
Fe XII
bare
bare
KBr
Wavelength
Figure 2. The SUMER spectrum taken at 1:26:33 UT, showing the change in detector sensitivity due to the KBr coating on the middle section. The profile is the spatially integrated
emission between the dotted lines. The dashed lines show the three spectral regions transmitted
thoughout most the observing sequence.
uncoated part of the detector. At wavelengths less than about 800 Å both the
bare and the KBr are comparable. One needs observations on both the coated
(KBr) and uncoated (bare) part of the detector. The C  was on the bare part
of the detector and the Fe  was on the KBr. As can be seen in Fig. 2, the
continuum around the C  is appreciably below the continuum around the
Fe  and Fe . For each window, there is an equation relating the observed
count rate to the first and second order fluxes and the detector sensitivity. The
two equations can be solved to obtain the first and second order continuum
levels.
The C  is a doublet and except for a few transient events, the emission
is stray light from the disk. It results in two faint emission lines along the
spectrometer slit (Fig. 3a). The positions of the lines do not change although
their strength can vary depending on the active region disk brightness. We
therefore average the intensity from specific pixels to determine the continuum. The pixels chosen to obtain continuum levels around the C  on the
bare part of the detector are shown in Fig. 3a. The pixels used to obtain the
continuum count rate on the KBr coated part, around the Fe , are shown
in Fig. 3b. Here we make a distinction between continuum on the blue and
downflow.tex; 13/03/2003; 13:37; p.4
5
Spectra of postflare inflows
a)
b)
Fe XII 1349
C II 1335
red
blue
pixel no.
pixel no.
Figure 3. The average line profiles in the a) C  and b) Fe  windows. The pixels used to
estimate the continuum are marked with solid horizontal lines. As discussed in the text, the
continuum on the left (marked blue) and right (marked red) behave differently.
red side of Fe . As will be seen, there are phases in the evolution when the
red-to-blue intensity ratio changes due to either Fe  or Fe  line emission
shifting into the continuum windows.
The SUMER spectra have been divided by the most recent flat-field, taken
in Nov 2001. There is some evidence that this no longer represents the present
small scale structure of the detector so that caution must be exercised when
interpreting stucture with a regular 5-6 pixel pattern. The other important
correction applied to the SUMER data was the geometric correction which
converts the raw ‘inverse cushion’ images to ‘rectangular grid’ images. The
standard corrections for deadtime and local gain were also applied.
TRACE 195 Å images were obtained at full resolution (pixel size 0.5 arcsec) and with a cadence of approximatly 20 s. There were short interuptions
in TRACE data around 0:50, 1:10, 1:40 and 2:00 UT. The images have been
processed with the solarsoft routine trace prep to remove the dark pedestal
and ccd characteristics.
Images at 304 Å, 171 Å, 195 Å and 284 Å were obtained by EIT between
1:00 and 1:20 UT. The images at 171 Å, 195 Å and 284 Å show the same flare
arcade structures as the TRACE image. The 304 Å image shows colder loops
extending into the corona at 1:19 UT at the position of bright C  emission in
downflow.tex; 13/03/2003; 13:37; p.5
6
SUMER spectra. The 304 Å image was calibrated using the solarsoft routine
eit prep.
The TRACE, EIT and SUMER images were first coaligned to within 5
arcsec using the given coordinates and pixel sizes in the image headers. Then,
within that range, we adjusted the overlay for the best correspondence of the
emissions. Using TRACE and SUMER time series between 0:30 and 4:00
UT, alignment to within 1 arcsec was achieved.
3. Overview of the Event
The three phases of the flare evolution have been outlined by Wang et al.
(2002). The initial phase starts with the ejection of disk material. First two
jets can be seen in the TRACE images and a few min later hot loops erupt
and expand through the corona above the flare site. In the corona, SUMER
first detects the jet in C  with a blue shift of 170 km s−1 . The subsequent
loop ejection, can be seen in TRACE as an expanding front with plane-ofsky speed about 120 km s−1 . The loop reaches the height of the SUMER
observations just before 1:00 UT. The next phase, starting at about 1:10 UT,
is the inflow phase (Fig. 4c,d).
4. SUMER observations
SUMER intensities in the three windows are shown in Fig. 5 along with the
corresponding time-space map of TRACE intensities. The TRACE image is
constructed by extracting the time series of 195 Å intensities at the position of
the SUMER slit (Sun-X=981 and −346 ≤ Sun-Y≤ −50). Each of the SUMER
frames in this figure is the sum of continuum and line intensity. The initial
Fe  hot, hook shaped structure is, as explained in Wang et al. (2002), the
rising loop in Fig. 4b. ‘jet-1’ can only be seen in the C  image.
The Fe  data is almost featureless. The intensity increases after 1:25 UT
when the flare arcade rises into the SUMER field-of-view. This general pattern is reflected in the TRACE. The TRACE passband, however, includes
an Fe  line as well as Fe  and Bremstrahlung continuum so the intensity distribution also shows features seen in Fe  such as the dark channels
around (1:40,-300) cutting into the bright arcade emission.
4.1. T Fe  
After 1:15 UT, much of the emission in the Fe  and C  windows is continuum. The continuum is Bremstrahlung emission from the hot flare arcade.
The separate Fe  and C  line and continuum intensities are shown in Fig. 6.
downflow.tex; 13/03/2003; 13:37; p.6
7
Spectra of postflare inflows
a)
b)
jet-1
jet-1
hot loop
jet-2
c)
d)
Figure 4. TRACE 195 Å filter images showing the development of the flaring region. (a) The
jets at the time of flare onset (b) Difference image showing the jets and the ejected loop at flare
onset (c) The initial flare arcade (d) the supra-arcade and inflows. The position of the SUMER
observations is marked with a white vertical line.
Here, the line emission is that remaining after continuum subtraction. The
continuum levels are computed from the pixels indicated in Fig. 3. The first
frame shows the Fe  line intensity. There is clearly less Fe  after 1:15 UT.
The disappearence of the Fe  is seen more clearly in Fig. 6b, the running
difference image. It can be seen that Fe  disappeared from the whole slit
area a little earlier at 1:10 UT. This unfortunately is exactly the time of a
TRACE data gap. The front is also picked up in the Fe  continuum on the
blue side of the line. It is marked ‘Fe  shift’ on Fig. 6c because it is due to
Fe  line emission shifted into the ‘blue’ continuum window. The Doppler
shift associated with the front is at least 250 km s−1 and it moves up the slit
with a speed 600 − 1000 km s−1 . The continuum on the other side of the line
is shown in Fig. 6d. This has basically the same structure but there is no fast
front at 1:10 UT. Three areas have been marked ‘Fe  shift’ because analysis
of the SUMER spectra at these times suggests that Fe  line emission with
downflow.tex; 13/03/2003; 13:37; p.7
8
a) Fe XXI
b) Fe XII
hot
loop
c) C II
d) TRACE
jet-1
Figure 5. The time evolution of SUMER intensities observed at the position marked in Fig 4,
compared to the TRACE 195 Å intensity at the same position. The white contours outline the
main Fe  emission. All intensities are integrated over the 2 Å wide windows and displayed
with a logarithmic scale. The white rectangle shows the position of the dark inflows analysed
in Sect. 5.
Doppler velocity ∼ 1000 km s−1 is shifted into the Fe  window at these
times (Paper II).
4.2. T C  
For the interpretation of the dark flows, the C  line emission is probably
most interesting. There are 4 main emission structures in the first hour of the
flare. The first is related to ‘jet-1’. The second and third, as shown in the EIT
He  304 Å image taken at 1:19:25 UT, are also related to disk loop or jet-like
structures. Fig. 7 shows there were cold structures reaching as far out as the
SUMER slit at this time. SUMER spectra show they formed at this height
around 1:10 UT and lasted until 1:30 UT.
The amount of material in the cold structures along the line-of-sight can
be estimated from the C  and EIT 304 intensities. As seen in Fig. 7, the
spatial overlap between the two is very good and it is quite likely that the
two lines are essentially coming from the same plasma. A lower limit to the
downflow.tex; 13/03/2003; 13:37; p.8
9
Spectra of postflare inflows
a) Fe XII line
c) Fe XII
blue
b) Fe XII line difference
Fe XII shift
d) Fe XII
red
Fe XXI shift
e) C II line
f) C II continuum
hard X-ray
Figure 6. The time evolution of the SUMER line intensities and continuum around the line,
scaled logarithically. Contours outline the position of main Fe  emission as in Fig 5. a)
Fe  line intensity b) Fe  running difference with 5 min time lag c) Fe  continuum on
the blue side of the line. Features of high Fe  line shift are marked. d) Fe  continuum on
the red side of the line. Features of high Fe  line shift are marked. e) C  line intensity f)
Continuum around C .
downflow.tex; 13/03/2003; 13:37; p.9
10
a)
Fe XII
Fe XXI
Sun-Y (arcsec)
C II
EIT
SUMER
Sun-X (arcsec)
Emission Measure
b)
C II (photospheric)
C II (coronal)
EIT
Sun-Y (arcsec)
Figure 7. SUMER spectra at the time of the EIT 304 Å image showing the C  structures in
the active region. a) EIT 304 image aligned with the SUMER C , Fe  and Fe  spectra b)
Emission measure from the EIT 304 intensities at the position of the SUMER slit compared
to the emission measure computed from C  intensities assuming photospheric and coronal
carbon abundance.
downflow.tex; 13/03/2003; 13:37; p.10
Spectra of postflare inflows
11
emission measure can be obtained by assuming the line intensity of any line is
coming from plasma at the peak of the line’s contribution function. The EIT
304 is dominanted by He  at temperatures below 105 K, and the peak in the
He  contribution function in ionization equilibrium is at 8 × 104 K. The C 
1335 has its peak contribution at 4 × 104 K. The emission measure in Fig. 7
for the 304 band is computed with the solarsoft routine eit flux, assuming a
temperature 8 × 104 K. The emission measure of the C  is determined from
the contribution function (Fig. 1) at a temperature 4×104 K. It is amazing that
the two emission measure estimates (Fig 7b) are within a factor two of one
another. The uncertainly associated with emission measure modeling at these
low temperatures must be at least this big when one takes into consideration
the uncertainty in the abundances, the ionization, the thermal mixing etc. If
this emission measure is correct, then with a structure width, 15 arcsec, the
density would be 109 cm−3 . The other C  structures in the SUMER spectrum
are found at similar locations. They are most likely also related to cold loops
in the corona.
Another interesting feature of the C  intensity is the brightening along the
length of the slit at 1:15 UT. This is exactly the time of the first hard X-ray
burst recorded by RHESSI (Caspi et al. 2002; Gallagher et al. 2002). It is seen
in no other line or in the continuum. It is either caused by an increase in the
C  stray light or resonance scattering in the corona.
5. Column density of dark inflows
Inflows are seen in TRACE images across the whole region from about 1:10
UT onwards. The most spectacular tracks in the TRACE images are seen in
the south after 1:30 UT. Fig. 8 shows coaligned TRACE and SUMER spectra
at the time of the southern inflows. The dark TRACE tracks are also seen in
the UV continuum and in Fe  observed by SUMER but they are not seen
in either the C  or the Fe  line emission. Therefore we can essentially rule
out dense plasma structures with temperature greater than 104 K.
The UV continuum can be used to investigate the possibility that they are
caused by even colder plasma. Because the count rates in the continua are
low the spectra have been averaged over four space and two time bins. Some
of the structure is therefore lost. The wavelength windows used to obtain the
continua are shown in Fig. 3. The Fe  continuum has been divided into red
and blue because line shifts from Fe  and Fe  cause differences in the
two continuum levels. In the following analysis, the blue side has been used
to obtain the continuum intensities.
We concentrate on the time of very clear dark inflow tracks in the south
at the height of the SUMER slit. During this phase of the arcade evolution,
the arcade is rising up to this height. SUMER and the corresponding TRACE
downflow.tex; 13/03/2003; 13:37; p.11
12
C II
Fe XXI
Sun-Y
Fe XII
TRACE
Sun-X
SUMER
Wavelength
Figure 8. SUMER spectra coaligned to one of the TRACE images at the time of several
inflows in the South. The vertical white line on the TRACE image is the SUMER slit position.
Dotted lines connect the dark TRACE inflows to the corresponding positions in the SUMER
spectra. The black dot in the middle of the Fe  window near Sun-Y=-270 is a detector
artifact.
space-time images of the inflows (white box in Fig. 5) are shown in Fig. 9.
The TRACE image, (Fig. 9a) shows several very clear dark channels in the
bright arcade emission. The northern inflow, ‘F1’, is interesting because it
appears at this height inside the arcade emission. It is an inflow falling through
existing hot arcade plasma. The other inflows appear as dark channels holding
off the arcade emission at their positions. There seem to be three separate
inflows, ‘F2, F3, F4’. The contours on Fig. 9 are the Fe  flux distribution,
displayed in Fig. 9b. The dark Fe  channels correlate almost exactly with
those in the TRACE image.
Fig. 9c, shows the first order continuum, computed using the continuum
level on the blue of the Fe  line and the continuum in the wing of C . Here
the contrast across the four inflows does not appear to be so well pronounced
as in Fe  and TRACE. The next figure (Fig. 9d), is the ratio of the Fe 
continuum intensity on the red and blue side. It is mostly featureless (grey).
The white spot,‘C3’, at (1:30, -310) has been pointed out in Fig. 6 as a ‘Fe 
shift’ region.
downflow.tex; 13/03/2003; 13:37; p.12
13
Spectra of postflare inflows
a)
b)
F1
F2
F3
F4
c)
d)
C3
f)
e)
C1
C2
C3
C1
Figure 9. Time-space maps near inflows (a) TRACE (b) Fe  (c) First order continuum (d)
Ratio of the continuum red and blue of Fe , scaled logarithmically between 0.25 and 4
(white) (e) Second order continuum (f) The ratio 4 times the first order continuum to the second order continuum, scaled logarithmically between 0.25 and 4. The ratio for Bremstrahlung
emission is 1. The contours outline the Fe  emission. The white vertical lines in b) show
the time intervals used to create the plots in Figs. 11 and 12. The flows F1-F4 are shown in
Fig. 11 and the channel, C1, is shown in Fig. 12.
The last two figures are critical for interpretation of the inflows. The first,
Fig. 9e, is the second order continuum. The second is the ratio of the first
order continuum multiplied by a factor four to the second order continuum,
scaled logarithmically between 0.25 and 4. Bremstrahlung radiation would
have the ratio 1 and so if it was simply Bremstrahlung emission the image
would be grey. The image becomes greyer and more uniform across the region
of bright continuum along the top. The bottom half of this image is distinctly
whiter. The ratio is typically 1.5. There are, however, three regions with a
ratio greater than 4, marked ‘C1, C2, C3’. The region,‘C3’, is associated with
a large red-to-blue Fe  continuum ratio which implies that the first order
continuum may be increased due to Doppler shifted Fe  (Paper II). The
large first-to-second order ratio at ‘C1’ and ‘C2’ cannot be so easily explained
as an increase in the first order continuum. The more likely explanation is
downflow.tex; 13/03/2003; 13:37; p.13
14
C II
Fe XXI
Sun-Y
Fe XII
TRACE
Sun-X
SUMER
Wavelength
Figure 10. SUMER spectra coaligned to the TRACE image at the time of unusually low
second to first order continuum ratio. The vertical white line on the TRACE image is the
SUMER slit position. A dotted line connects the dark TRACE inflow to the region of dark C 
continuum (the second order).
absorption of short wavelength, second order, photons by cold intervening
material. There is a dark channel at ‘C1’ in image Fig. 9e. In the SUMER
spectrum, Fig. 10, it can be seen as a dark region in the C  continuum at
Sun-Y=-300 and in the Fe , but not in the Fe  continuum.
Two time intervals have been selected for more detailed investigation. The
intervals are drawn in Fig. 9b. The earliest, between 1:33 and 1:38 UT, covers the dark track, ‘C1’. The later one between, 1:48 and 1:53 UT, covers
a time when the four inflows are crossing the SUMER slit position. The
time-averaged fluxes at these two times are shown in Figs. 11 and 12.
Fig. 11 shows the fluxes at the later time, when there are four inflows.
All four, ‘F1-F4’, show up clearly in the Fe , the first and second order
continuum and TRACE (Fig. 11a,b,c). Fig. 11d compares the continua on the
red and blue side of Fe . Again the important figures for the absorption
are the last two. Fig. 11e compares the structure of first and second order
emission across the region. Both the first and second order decrease by about
the same amount across the flows. This implies that at this time and height,
the dark channels are caused by less emission measure not absorption along
the line-of-sight. Fig. 11f, compares the observed fluxes to Bremstrahlung
downflow.tex; 13/03/2003; 13:37; p.14
Spectra of postflare inflows
a)
F4 F3
15
b)
F2
F1
c)
d)
f)
e)
Figure 11. Flux across the dark inflows in Fig. 9, averaged between 1:48 and 1:53 UT. (a)
TRACE and Fe  (dashed) (b) TRACE and first order continuum (dashed) (c) TRACE and
second order continuum (dashed) (d) Continuum blue of Fe  and red of Fe  (dashed) (e)
First order continuum and second order continuum (solid) (f) Second order continuum (solid)
and four times the first order continuum. The fluxes in (a-e) have been normalized to the
smoothed value. (f) shows log intensity in W sr−1 m−2 Å−1 .
emission. The dashed line is four times the first order flux and the solid line
is the second order flux.
At the earlier time, 1:33 to 1:38 UT, there is no clear darkening in the
TRACE image, but there is the dark channel in the second order continuum
(Fig. 9) and the Fe . The plots in Fig. 12 show the same. In Fig. 12f, there
is a deep dip around the position of ‘C1’, The second order is a factor 5 less
than expected from the first order. This is really a lower limit to the ratio
because the C  continuum is essentially non-existent.
Finally in this section, we compute the implied H  column density from
the continuum ratios. The results are sumarized in Table I. The logarithm of
downflow.tex; 13/03/2003; 13:37; p.15
16
a)
b)
C1
c)
C1
d)
C1
e)
f)
C1
C1
Figure 12. Fluxes across the dark channel in Fig. 9, averaged between times 1:33 and 1:38
UT. The fluxes are the same as in Fig. 11
the ratio (4 times the first order continuum flux to the second order continuum) gives the optical depth τ. The column density, nH dl, is computed using
the H  absorption cross section, 10−18 cm2 , at 670 Å (Rumph et al. 1994). The
column densities across most of the structures are comparable to the column
density in the cold loop structures seen in He  and C . There the column
density (emission measure/density) was also between 1017 − 1018 cm−2 . The
absorption cross section at 195 Å is a factor 10 less than the cross-section at
670 Å (Rumph et al. 1994), so this small amount of material cannot possibly
be causing the EUV darkening. The only region with clear structure was the
channel ‘C1’. If the width, dl, is 10 arcsec as observed, the H  density, nH , is
2 × 109 cm−3 . The counts in the C  continuum in the dark channel are very
few and this is only a lower limit.
downflow.tex; 13/03/2003; 13:37; p.16
Spectra of postflare inflows
17
Table I. H  column density
Time
Position
ratio
τ
nH dl (cm− 2)
1:36
1:36
1:50
1:50
-300
-250
-300
-250
5.0
1.2
1.8
1.2
1.6
0.2
0.6
0.2
1.6 × 1018
2 × 1017
6 × 1017
2 × 1017
6. Discussion
The dark inflows seen at 195 Å in TRACE were also seen in SUMER Fe 
and UV continuum. These were not seen in C  and Fe . The ratio of the
UV continuum longward and shortward of the Lyman limit indicate a small
amount of plasma with temperature below 104 K. Only at one position is
absorption of second order photons associated with a dark Fe  channel. It
is seen above the TRACE arcade and not detected as a darkening in TRACE.
The derived column density of H  is not enough to cause any observable
dimming in the 195 Å. It is therefore our conclusion that the dark tracks are
plasma voids and not caused by intervening cold plasma clumps. The cause
of the voids will be investigated later.
Acknowledgements
These observations could not have been obtained without the help of the
SUMER planners Cristian Vocks and Werner Curdt. We would also like to
thank Hugh Hudson for his valuable comments. The SUMER project is financially supported by DARA, CNES, NASA and the ESA PRODEX programme (Swiss contribution). SUMER and MDI are part of SOHO, the Solar
and Heliospheric Observatory of ESA and NASA.
References
Caspi, A., S. Krucker, and R. P. Lin: 2002, ‘Correlation of RHESSI and
TRACE Observations of the Rise Phase of the 21 April 2002 X1.5
Flare’. In: American Geophysical Union, Fall Meeting 2002, abstract
#SH52A-0465. pp. A465–+.
Dere, K. P., E. Landi, H. E. Mason, B. C. Monsignori Fossi, and P. R.
Young: 1997, ‘CHIANTI - an atomic database for emission lines’. Astron.
Astrophys. Suppl. 125, 149–173.
downflow.tex; 13/03/2003; 13:37; p.17
18
Gallagher, P. T., B. R. Dennis, S. Krucker, R. A. Schwartz, and A. K. Tolbert:
2002, ‘RHESSI and TRACE observations of the 21 April 2002 X1.5 flare
’. Solar Phys. 210, 341–372.
Innes, D. E., D. E. McKenzie, and T. J. Wang: 2003, ‘Observations of
1000 km s−1 Doppler shifts in 107 K solar flare supra-arcade’. Solar Phys.
pp. submitted, Paper II.
McKenzie, D. E.: 2000, ‘Supra-arcade Downflows in Long-Duration Solar
Flare Events’. Solar Phys. 195, 381–399.
McKenzie, D. E. and H. S. Hudson: 1999, ‘X-Ray Observations of Motions
and Structure above a Solar Flare Arcade’. Astrophys. J. Lett. 519, L93–
L96.
Rumph, T., S. Bowyer, and S. Vennes: 1994, ‘Interstellar medium continuum,
autoionization, and line absorption in the extreme ultraviolet’. Astron. J.
107, 2108.
Wang, T. J., S. Solanki, D. E. Innes, and W. Curdt: 2002, ‘Initial features of an
X-Class flare observed with SUMER and TRACE’. In: Proc. IAU Colloq.
188: Magnetic Coupling of the Solar Atmosphere, ed. H. Sawaya-Lacoste
(ESA SP-505, Noordwijk: ESA). p. 607.
Wilhelm, K., P. Lemaire, W. Curdt, U. Schuhle, E. Marsch, A. I. Poland, S. D.
Jordan, R. J. Thomas, D. M. Hassler, M. C. E. Huber, J. . Vial, M. Kuhne,
O. H. W. Siegmund, A. Gabriel, J. G. Timothy, M. Grewing, U. Feldman,
J. Hollandt, and P. Brekke: 1997, ‘First Results of the SUMER Telescope
and Spectrometer on SOHO - I. Spectra and Spectroradiometry’. Solar
Phys. 170, 75–104.
downflow.tex; 13/03/2003; 13:37; p.18