Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://star.arm.ac.uk/~ambn/2001/350teriaca_final.ps Äàòà èçìåíåíèÿ: Wed Jan 24 13:34:36 2001 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 05:20:36 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: prominence

Electron density variations during ultraviolet transient
events
L. Teriaca y (lte@arcetri.astro.it), M. S. Madjarska
(madj@star.arm.ac.uk) and J. G. Doyle (jgd@star.arm.ac.uk)
Armagh Observatory, College Hill, Armagh BT61 9DG, N. Ireland
Abstract. High­resolution temporal observations performed with the SUMER
spectrometer on SoHO provide an opportunity to investigate the electron density
variations in the `quiet Sun' solar transition region due to UV transient events. Two
datasets obtained in the density sensitive lines belonging to the O iv 1400 š Amultiplet
were searched for such events, leading to the identification of two explosive events,
on 1996 July 10 and 1997 May 31. In both cases, the O iv 1401:16/1404:81 density
sensitive line intensity ratio shows a clear variation, corresponding to enhancements
in the electron density by factors of ¸ 3. This is fully consistent with recent 2.5D
MHD simulations. The 1996 July 10 dataset also provided us with the opportunity
to monitor the behaviour of the electron density through an UV blinker. Despite an
increase of a factor of two in the line intensities, no variation of the electron density
was found. This suggests that the intensity enhancement is due to an increase in the
filling factor.
Keywords: Sun: Transition region--Ultraviolet: SoHO--Sun: explosive events: elec­
tron density:
1. Introduction
Brueckner and Bartoe (1983) using observations made with the high
resolution telescope and spectrograph (HRTS), described several high
velocity events seen in lines from ions formed at transition region tem­
peratures (particularly C iv and Si iv). The authors classified them as
turbulent events and jets. The first ones showed velocity components up
to 250 km s \Gamma1 with a spatial dimension of ¸ 2000 km. Following these
first observations, several other authors (Dere, Bartoe, and Brueckner,
1989; Porter and Dere, 1991; Innes et al., 1997a; Chae et al., 1998a;
P'erez et al., 1999a; Landi et al., 2000) have studied the phenomena,
extending and completing the description of the general characteristics
and naming them as `explosive events'.
Primarily observed in the network lanes at the boundaries of the
super­granulation cells (Porter and Dere, 1991), the explosive events
are preferentially found in regions with weak and mixed polarity fluxes
that display magnetic neutral lines (Chae et al., 1998a). They are often
y now at: Osservatorio Astrofisico di Arcetri, 50125 Firenze, Italy
c
fl 2001 Kluwer Academic Publishers. Printed in the Netherlands.
teriaca—final.tex; 23/01/2001; 14:40; p.1

2 Teriaca et al.
observed in bursts of up to 30 minutes (Innes et al., 1997a; Chae et al.,
1998a, P'erez et al., 1999a), while the average lifetime of a single event
is ¸ 60 s (Dere, 1994). The presence of bi­directional jets (Innes et
al., 1997b) with velocities comparable to the local Alfv'en speed (Dere,
1994), together with the often observed association with episodes of
photospheric magnetic flux cancellation lasting more than 1 hour (Dere
et al., 1991; Dere, 1994; Chae et al., 1998a), suggest a sequence of mag­
netic reconnection events (Parker, 1988; Porter and Dere, 1991; Dere,
1994; Innes et al., 1997b; Wilhelm et al., 1998). Recent MHD modelling
of shear­induced magnetic reconnection carried out by Karpen, Antio­
chos and DeVore (1995) and Karpen et al. (1998) shows the appearance
of features in which the density is 4­10 times higher than the pre­event
local electron density.
A new recently discovered class of transient phenomena charac­
terizing emission lines formed at transition region temperatures was
introduced by Harrison (1997) with the name of `blinkers'. Observations
performed with the CDS spectrometer (Harrison et al., 1995; 1997) on­
board SoHO show enhancements in the flux of transition region lines
at network junctions (Harrison, 1997; Harrison et al., 1999). Blinkers
are mainly observed in lines of O iii, O iv and O v with modest or
no detectable increase at higher or lower temperatures. They show a
typical lifetime of ¸ 17 minutes over an area of ¸ 5 10 7 km 2 , with an
average intensity increase in O v of ¸ 1:5 which, in extreme cases, can
reach values as high as five times the pre­event level (Harrison et al.,
1999).
Despite the large number of observational works on UV explosive
events and blinkers, large uncertainties about their basic physical pa­
rameters such as electron density and temperature still exist. Hayes and
Shine (1987), using the ratio of O iv 1401 and Si iv 1402, show some
evidence of density enhancements associated with short­lived bursts in
an active region. Dere (1994), using the ratio of O iv lines, reports a
density of 7 10 10 cm \Gamma3 in an explosive event. More recent indications
of density enhancement during explosive events have been presented
by Wilhelm et al. (1998). In order to derive the electron density of
transition region plasma, the most reliable method involves the use
of density­sensitive line ratios. However, only lines from the same ion
and with similar wavelength should be used (Mason and Monsignori
Fossi, 1994; Jordan, 1996; WikstÜl et al., 1998). The fact of belonging
to the same ionization stage eliminates the effects of possible differen­
tial departure from ionization equilibrium, specially during dynamical
events. These constraints are matched by the O iv 1400 š A multiplet
(Cook et al., 1995; Brage et al., 1996) which falls inside the wavelength
range covered by the SUMER spectrometer on­board SoHO. WikstÜl
teriaca—final.tex; 23/01/2001; 14:40; p.2

Electron density variations during UV transient events 3
Table I. SUMER Quiet Sun observations
Starting Date/Time Det. Slit Exp. Time (s) Solar X Solar Y
1996 July 10 17:10 A 1 00 x120 00 20 + 3 + 0
1997 May 31 15:01 B 0.3 00 x120 00 12 +600 +576
et al. (1997) presented a critical discussion on many of the possible
density sensitive line ratios observable with SUMER, suggesting that
the O iv 1400 š A multiplet maybe the only one yielding reliable density
diagnostics for the transition region. As shown by Doschek et al. (1998)
and O'Shea et al. (2000) lines from O V are also excellent electron
density diagnostics. However, for transient events, the O IV lines are
more suited due to their count rate. In the present contribution we
studied the behaviour of the O iv 1401.16/1404.81 line ratio during
the appearance of explosive events and an UV blinker. In the next
section we discuss the observations and the data reduction. In x 3, data
analysis will be presented, paying particular attention to the problems
related to the blends affecting the O iv lines. Results are described in
x 4, while their implications and relation to current theoretical models
are discussed in x 5.
2. Observations and Data Reduction
SUMER is a normal incidence spectrograph operating over the wave­
length range ¸500 to 1610 š A (details can be obtained from Wilhelm et
al., 1995; 1997; Lemaire et al., 1997). Both of the datasets presented
in this work were obtained on the `quiet Sun' in a sit­and­stare mode
with no compensation of solar rotation applied.
The 10th July 1996 dataset (see Table I and Figure 1) consists of 42
spectra, exposing for 20 seconds the central part of detector A through
a 1\Theta120 arcsec 2 slit. For each spectrum, four spectral windows (50
pixels wide) were transmitted to the ground, respectively centered on
O iv 1399:77 š A, O iv 1401:16 š A, O iv 1404:81 š A and O iv 1407:38 š A.
The selected series of spectra cover a time period of 16 minutes during
which, at solar Y= 0, the Sun rotates ¸ 2:5 arcsec.
Data for the second dataset were obtained on 31 May 1997 (see
Table I and Figure 2) with a 0.3\Theta120 arcsec 2 slit on the bottom part of
detector B using exposure times of 12 seconds. They consist of a tem­
poral series comprising of 560 spectra. For each spectrum, four spectral
teriaca—final.tex; 23/01/2001; 14:40; p.3

4 Teriaca et al.
Figure 1. O iv 1401.16 intensity map of the 1996 July 10 dataset (DS1). Dashed
lines define the region (EE 1) along which the temporal variation of the electron
density was followed (see Table II). Areas enclosed by solid lines are the ones over
which an integration was performed in order to obtain averaged values for cell center
(CC 1) and network (N 1). The evolution of an UV blinker was also studied (BL)
along the region defined between the dash­dotted lines. Black columns at 5 and 15
mins. represent missing data.
windows (25 pixels wide) were transmitted to the ground, respectively
centered on Si iv 1393:76 š A, O iv 1401:16 š A, Si iv 1402:77 š A and O iv
1404:81 š A. The observations cover a period of 112 minutes, during
which the Sun rotates ¸ 8 arcsec (¸ 27 times the slit width).
The reduction of SUMER raw images followed several stages, i.e.
dead time correction, local gain correction, flat­field subtraction (the
July 1996 data were flat­field corrected on­board), radiometric calibra­
tion (in order to pass from count px \Gamma1 s \Gamma1 to Watt m \Gamma2 Sr \Gamma1 š A \Gamma1
)
and a correction for geometrical distortion.
teriaca—final.tex; 23/01/2001; 14:40; p.4

Electron density variations during UV transient events 5
Figure 2. O iv 1401.16 logarithmic intensity maps of the 1997 May 31 dataset (DS2).
Left panel: Time series of the intensity along the slit. Dashed lines define the region
(EE 2) along which the variation of the electron density was followed (see Table III).
Areas enclosed by solid lines are the ones over which an integration was performed
in order to obtain averaged values for cell center (CC 2), network (N 2) and bright
network (BN) regions. Right panel: The image on the left panel has been binned over
groups of 21 slit positions, showing the area of the Sun covered during the entire
sequence (see text for details).
3. Data analysis
The two datasets studied here present different characteristics which
complement each other. The July 1996 dataset (hereafter DS1) contains
four O iv lines and offers the possibility to make an evaluation about
line blending importance. The large spectral windows allow a precise
determination of the line parameters and background contribution. Due
to the wider slit it is also possible to achieve a reasonable temporal
resolution, while the small drift on the solar surface (2.5 times the
slit width) allows us to consider DS1 as an example of the temporal
evolution of a `quiet Sun' area at 2 arcsec resolution. The May 1997
dataset (hereafter DS2) presents only two of the oxygen lines belonging
to the O iv 1400 š A multiplet. In the DS2 case, the slit covers an area
which is 27 times the slit width, resulting more in a raster­type scan
than a temporal sequence.
3.1. Background subtraction
In the DS1 case, all line parameters were measured applying a multi­
Gaussian fit with a constant background. In the DS2 case, due to
the small spectral windows, a different procedure was used. First, a
multi­Gaussian fit with constant background was applied in order to
measure the background itself and the contribution of S i 1401.51 to
the O iv 1401.16 line (see x 3.2). Enhancements in the line wings,
teriaca—final.tex; 23/01/2001; 14:40; p.5

6 Teriaca et al.
Figure 3. S i 1401.51 to O iv 1401.16 ratio as a function of time/position along
EE 2. Crosses represent values obtained from the Gaussian fit, while the solid line
represent a spline fit to the data.
particularly during explosive events, can lead to an over­estimation of
the background level, particularly for the stronger lines when the small
25 pixel wide spectral window is used. Thus, we decided to use as
background values, in all three lines, the ones obtained for the weak
O iv 1404.81 line. After this the intensity in the above lines was cal­
culated, integrating over the full spectral window and then subtracting
the previously obtained background values. The effect of line blends
are, hence, evaluated and removed (see below).
3.2. Line blending
The major problem in measuring transition region electron densities
using the O iv 1400 š A multiplet comes from the fact that some of
these lines are affected by blending. The situation is even more com­
plex in SUMER data due to the superimposition of second order lines
belonging to O iii. We only discuss below the O iv lines at 1401.16 š A
and 1404.81 š A; problems associated with 1399.77 š A and 1407 š A can be
found in Teriaca (2001).
The O iv 1401.16 line ( 2 P 0
3=2 -- 4 P 5=2 ) is blended with S i 1401.51
š A. This blend can be easily resolved using a two Gaussian fit. The
contribution of S i varies from ¸ 4% of the O iv 1401.16 line in the
bright network (BL and BN, see Figures 1 and 2 respectively) up to
37% in the very dark cell center (CC 2, see Figure 2) with average
values over the entire datasets of 13% and 12% for DS1 and DS2,
respectively. Some difficulties arise during explosive events, when the
S i line is blended with the enhanced wings of the O iv line. In the DS1
case, this has been solved using a constrained Gaussian fit where the
width and position of S i were obtained from the averaged values over
the entire DS1 dataset.
teriaca—final.tex; 23/01/2001; 14:40; p.6

Electron density variations during UV transient events 7
Figure 4. O iv 1401 line intensity variation as a function of time along the central
pixel of the BL region showed in Figure 1.
In the DS2 case, the situation was complicated by the small spectral
window. Even if any of the explosive events seen in the Si iv lines
were visible in the O iv 1401.16 line, it is difficult to estimate the S i
contribution as well as the background level. In Figure 3 we can observe
the S i/O iv ratio along EE 2. Even if we observe some scatter in the
points characterized by the appearance of explosive events, a clear trend
from the darker to the brighter regions is clearly visible. From these
considerations the final S i 1401.51 contribution to the O iv 1401.16 was
obtained applying a spline fit to the data in Figure 3. This assumption
is also justified by the observational evidence that explosive events are
not observed in lines formed at temperatures below 4 10 4 K (Dere,
1994).
The O iv line at 1404.81 š A ( 2 P 0
3=2 -- 4 P 3=2 ) is heavily blended in
SUMER data with the O iii 702.34 second order line and S iv 1404.771
(Feldman and Doschek, 1979; Cook et al., 1995). These blends cannot
be resolved using a line fitting technique and an evaluation using an
atomic database is required. The S iv 1404.771 contribution to the total
1404 feature is only 3.6% (Teriaca, 2001).
After correcting for the small contribution of S iv, the CHIANTI
(Landi et al., 1999) database was used in order to estimate the theo­
retical ratio O iii (703.845+703.85)/702.34., obtaining a value of 0.2.
This compares very well with the measured mean value of 0.212 from
SUMER reference spectra for the quiet sun as given by Doschek et
al. (1999). This line intensity ratio is density independent in the range
between 10 9 and 10 11 cm \Gamma3 . The O iii 703.845 š A and 703.85 š A lines
are observed as an unique spectral line, hereafter also indicated as O iii
703. Using the above value of 0.212 for the ratio O iii 703/702.34, the
intensity of the O iii 702.34 line can be estimated and subtracted from
the total intensity of the spectral feature at 1404 š A. In DS1, this was
done for each chosen area finding that the contribution of O iii 702.34 to
teriaca—final.tex; 23/01/2001; 14:40; p.7

8 Teriaca et al.
the total spectral feature varied between 16% and 37%, with an average
value over the entire dataset of 31%. In order to stress the importance of
evaluating the O iii blend point by point, we also present (see Table II
and Figure 7 and 9) the value obtained assuming everywhere a fixed
65% contribution of O iv 1404.81 to the total spectral feature.
In the DS2 case, the 1407 š A spectral feature is not present and the
average value of 65% derived above was used for all the selected areas.
We conclude that, in average, the blend at 1404 š A comprises ¸4% of
S iv 1404.771, ¸31% of O iii 702.34 and, finally, ¸65% of O iv 1404.81.
Judge et al. (1998) for the quiet Sun, suggested an O iii contribution of
20% and assumed a zero contribution for S iv. Madjarska et al. (1999),
studying a prominence, find contributions of 18% and 5% for the two
lines respectively.
3.3. Explosive events and blinkers identification
Explosive events are usually observed in strong resonance lines such
as C iv 1548 and 1550 or Si iv 1393 and 1402. In DS1, we have only
the quite weak O iv intersystem lines around 1400 š A. This makes it
difficult to visually identify explosive events. A solution of this problem
comes from the closeness between O iv 1401.16 and Si iv 1402.77. The
latter is a much stronger line and its left wing is present (weakly) in the
right part of the 50 pixel wide O iv window. During an explosive event
in which the blue component is present, the amount of Si iv 1402.77
inside the O iv 1401.16 window becomes very large and it appears as
a brightening in the O iv 1401.16 intensity map (see Figure 1, along
EE 1 at t¸ 11 minutes). In these conditions, only red wing dominated
events remain undetected. However, there is evidence for a prevalence
of blue wing dominated events over the red wing ones by a factor of
3:2 or more (Dere et al., 1989, Innes et al., 1997a) and in events where
both wings are enhanced, the left one is often the strongest (Dere et
al., 1989; Landi et al., 2000). Using this technique, an event in DS1
(see Figure 1 and 7) was found at solar Y ¸ 53. The event is clearly
present in two successive spectra, leading to a minimum duration of
¸ 40 seconds. A region 7 pixels wide (EE 1) was hence selected along
the slit (region between dashed lines in Figure 1) and, binning over
different time intervals (see Figure 7 and Table II), the line parameters
were determined. Further examination shows the explosive event being
visible also in the O iv 1401.16 line itself, as an event with a dominant
blue­shifted component with a velocity up to 100 km s \Gamma1 . From the
position of the left wing of the Si iv 1402.77 line during the event, it is
possible to estimate velocities up to 150 km s \Gamma1 .
teriaca—final.tex; 23/01/2001; 14:40; p.8

Electron density variations during UV transient events 9
Another possibility of detecting regions with enhanced dynamical
activity is through the analysis of the line width, looking for particularly
broad profiles. In fact, transition region and coronal emission lines
show a full width half maximum (FWHM) that exceed the Doppler
thermal broadening (Mariska, 1992). When opacity effects can be ex­
cluded (Doyle et al., 2000), this excess is usually explained through
unresolved motion associated with small and large scale motion and/or
wave propagation (non­thermal velocity). When we consider all the
selected regions in DS1, we note that the FWHM of O iv 1401 is always
close to the value obtained averaging over the entire dataset with the
exclusion of the region with the explosive event already identified and,
partially, with the exclusion of the first region selected along EE 1 (see
Table II), where we found a FWHM of 0.327 š A, 30 % larger than the
value obtained averaging the entire dataset (instrumental broadening
included).
During the first 300 seconds of DS1 it is possible to observe a
strong enhancement in the flux of all the spectral features present
in the dataset. The brightening is located around solar Y = 5 (see
Figure 1) and has an approximate size in the Y direction of L ¸ 7
pixels (' 5000 km). In Figure 4, the total intensity in the spectral
range around the O iv 1401.16 line (range chosen avoiding the wing
of Si iv 1402) is shown as a function of time along the pixel positions
passing through the point of maximum brightening (the central pixel
of region BL in Figure 1). From Figure 4 it is possible to estimate an
approximative duration of ¸ 200 seconds for the event under examina­
tion. We are not able to determine the dimension in the X direction,
but, assuming the same value along the Y direction, we have an area
úL 2 ¸ 8 \Theta 10 7 km 2 , in agreement with the typical blinker area reported
by Harrison et al. (1999). A region 7 pixels wide was, hence, selected
for DS1 and the temporal evolution of this brightening was followed
binning in time over 5 successive spectra (100 seconds) (see Figure 1
and 9).
DS2 was also searched for explosive events performing a visual ex­
amination of the Si iv 1393.76 and 1402.77 slit images (see Figure 8),
finding a sequence of several explosive events at solar Y ¸ 540, starting
approximatively 25 minutes after the beginning of DS2 and lasting
¸ 30 minutes. Also in this case a location along the slit was determined
(EE 2) and its temporal and spatial evolution studied. In order to have
some idea about the presence of enhanced wings due to explosive event
activity, we define, for each selected interval along EE 2, the quantity
teriaca—final.tex; 23/01/2001; 14:40; p.9

10 Teriaca et al.
Figure 5. Theoretical dependence of the O iv 1401.16/1404.81 intensity line ratio as
calculated with the CHIANTI database for three different temperatures: log Te = 5:1
(dotted line), log Te = 5:2 (solid line) and log Te = 5:3 (dashed line). The crosses rep­
resent values of the theoretical line ratio computed by Brage et al. (1996) including
proton impact excitation at a temperature of 1:4 10 5 K (log Te = 5:15).
W e (wings enhancement) as:
W e = I b + I r
I peak
(1)
where I peak is the peak intensity of the Si iv 1402.77 š A line, while
I b and I r are the line intensities averaged over 0.17 š A wide intervals
centered 0.22 š A away from the central position on the blue and on the
red sides of the spectral line. The values of W e along EE 2 are reported
in Figure 8(b).
3.4. Line ratios
The inference of the electron density in an ionized plasma using the
line ratio of forbidden lines of transition region ions like O iv has
been used by several authors, (eg. Feldman et al., 1976; Dere et al.,
1982; Dwivedi and Gupta, 1991; Brage et al., 1996; Wilhelm et al.,
1998; Judge et al., 1998; Del Zanna and Bromage, 1999; P'erez et al.,
1999b). We use here only the O iv 1401.16/1404.81 intensity line ratio
(hereafter RD ). Figure 5 shows the variation of RD as a function of
electron density obtained from the CHIANTI atomic database for three
different temperatures: log T e = 5:1, 5.2 (O iv formation temperature)
and 5.3. The CHIANTI database allows a complete solution of the
teriaca—final.tex; 23/01/2001; 14:40; p.10

Electron density variations during UV transient events 11
Figure 6. ffl(– lu )/Ne as a function of the electron density as calculated from the
CHIANTI database for O iii 599, O iv 554, O v 629 (dashed lines) and O iii 703.845,
O iii 703.85, O iv 1401 and O iv 1404 (solid lines). The vertical dashed line represents
the maximum density value measured in DS1. For each line, calculations have been
performed adopting the formation temperature of the ion to which the line belongs.
statistical balance equation system for the ion in question, including
electron impact excitation and de­excitations, but it does not include
proton excitation rates. However, proton collisions can be important
at redistributing electrons amongst the fine structure levels of a minor
species ion formed in the transition region, for example see Doyle et
al. (1980).
Brage et al. (1996) presented theoretical calculations of the RD line
ratio that include proton excitation. Their results are also shown in
Figure 5 and are fully consistent with the ones obtained with CHIANTI.
Errors on line intensities determination have been calculated through
the photon noise and, hence, the errors on the intensity line ratio es­
tablished. RD shows also a weak temperature dependence resulting in
a change of 0.06 for a 0.1 dex variation of T e (at N e = 10 10 cm \Gamma3 ). This
last error has been quadratically summed to the ones on the intensity
line ratio, obtaining the total errors reported in Table II and III. The
final errors on the electron density have been evaluated through the
CHIANTI database.
One way of verifying whether the observed solar plasma undergoes
temperature variation is through the ratio of allowed lines from different
ionization stages of the same element (this eliminates uncertainties over
abundances). Intensities of lines from allowed transitions respond to
teriaca—final.tex; 23/01/2001; 14:40; p.11

12 Teriaca et al.
Table II. Results of the 1996 July 10 dataset (DS1). Values in boldface refer to the
explosive event.
Time Ratio a Ne a Ratio b Ne b
(s) (cm \Gamma3 ) (cm \Gamma3 )
EE 1
0 -- 79 (c) 3.18 \Sigma 0.14 (1.5 +0:2
\Gamma0:2 ) \Theta 10 10 2.94 \Sigma 0.18 (1.2 +0:2
\Gamma0:2 ) \Theta 10 10
80 -- 159 2.42 \Sigma 0.11 (6.0 +1:1
\Gamma1:1 ) \Theta 10 9 2.23 \Sigma 0.14 (4.2 +1:3
\Gamma1:2 ) \Theta 10 9
160 -- 239 2.79 \Sigma 0.14 (9.9 +1:7
\Gamma1:6 ) \Theta 10 9 2.72 \Sigma 0.19 (9.1 +2:3
\Gamma2:0 ) \Theta 10 9
240 -- 319 2.30 \Sigma 0.12 (4.8 +1:1
\Gamma1:1 ) \Theta 10 9 2.11 \Sigma 0.15 (3.2 +1:3
\Gamma1:2 ) \Theta 10 9
420 -- 499 2.00 \Sigma 0.12 (2.3 +1:0
\Gamma1:0 ) \Theta 10 9 1.90 \Sigma 0.16 (1.5 +1:2
\Gamma1:1 ) \Theta 10 9
500 -- 579 1.77 \Sigma 0.11 (0.6 +0:8
\Gamma\Gamma\Gamma ) \Theta 10 9 1.76 \Sigma 0.14 (0.5 +0:2
\Gamma\Gamma\Gamma ) \Theta 10 9
580 -- 659 2.27 \Sigma 0.13 (4.6 +1:2
\Gamma1:2 ) \Theta 10 9 1.98 \Sigma 0.15 (2.2 +1:2
\Gamma1:2 ) \Theta 10 9
660 -- 699 3.33 \Sigma 0.23 (1.7 +0:4
\Gamma0:3 ) \Theta 10 10 2.71 \Sigma 0.23 (0.9 +0:3
\Gamma0:2 ) \Theta 10 10
700 -- 799 2.36 \Sigma 0.14 (5.4 +1:3
\Gamma1:2 ) \Theta 10 9 2.07 \Sigma 0.16 (2.9 +1:3
\Gamma1:2 ) \Theta 10 9
800 -- 899 2.01 \Sigma 0.12 (2.4 +1:0
\Gamma0:9 ) \Theta 10 9 1.76 \Sigma 0.13 (0.5 +0:5
\Gamma\Gamma\Gamma ) \Theta 10 9
0 -- 899 (d) 2.39 \Sigma 0.07 (5.6 +0:7
\Gamma0:7 ) \Theta 10 9 2.16 \Sigma 0.08 (3.6 +0:7
\Gamma0:6 ) \Theta 10 9
BL
20 -- 119 2.43 \Sigma 0.10 (6.1 +1:0
\Gamma0:9 ) \Theta 10 9 2.52 \Sigma 0.14 (6.9 +1:5
\Gamma1:4 ) \Theta 10 9
120 -- 219 2.33 \Sigma 0.08 (5.1 +0:8
\Gamma0:8 ) \Theta 10 9 2.45 \Sigma 0.12 (6.3 +1:2
\Gamma1:1 ) \Theta 10 9
220 -- 319 2.41 \Sigma 0.10 (5.9 +1:0
\Gamma0:9 ) \Theta 10 9 2.46 \Sigma 0.14 (6.4 +1:4
\Gamma1:3 ) \Theta 10 9
380 -- 479 2.35 \Sigma 0.12 (5.3 +1:1
\Gamma1:1 ) \Theta 10 9 2.17 \Sigma 0.14 (3.7 +1:3
\Gamma1:2 ) \Theta 10 9
480 -- 579 2.42 \Sigma 0.11 (6.0 +1:1
\Gamma1:1 ) \Theta 10 9 2.40 \Sigma 0.15 (5.8 +1:5
\Gamma1:4 ) \Theta 10 9
580 -- 679 2.27 \Sigma 0.11 (4.6 +1:0
\Gamma0:9 ) \Theta 10 9 2.28 \Sigma 0.15 (4.6 +1:4
\Gamma1:3 ) \Theta 10 9
680 -- 779 2.33 \Sigma 0.11 (5.1 +1:0
\Gamma0:9 ) \Theta 10 9 2.43 \Sigma 0.16 (6.1 +1:6
\Gamma1:5 ) \Theta 10 9
780 -- 879 2.33 \Sigma 0.12 (5.1 +1:1
\Gamma1:0 ) \Theta 10 9 2.44 \Sigma 0.18 (6.1 +1:8
\Gamma1:6 ) \Theta 10 9
ALL 1 (e) 2.23 \Sigma 0.06 (4.2 +0:5
\Gamma0:5 ) \Theta 10 9 2.23 \Sigma 0.06 (4.2 +0:6
\Gamma0:5 ) \Theta 10 9
N 1 2.41 \Sigma 0.07 (5.8 +0:7
\Gamma0:7 ) \Theta 10 9 2.26 \Sigma 0.08 (4.5 +0:7
\Gamma0:7 ) \Theta 10 9
CC 1 2.28 \Sigma 0.08 (4.7 +0:7
\Gamma0:7 ) \Theta 10 9 2.30 \Sigma 0.10 (4.8 +0:9
\Gamma0:9 ) \Theta 10 9
(a) : Calculated assuming 31% and 3.6% contributions of O iii 702.34 and S iv
1404.771, respectively to the total – 1404 š A spectral feature
(b) : Calculated using the local R O IV
1404 value
(c) : Characterized by broad line profile (see text x 3.3)
(d) : Average over the entire EE 1 region
(e) : Represents the average over the whole dataset
electron density changes in the same way, and any significative change
in the line ratio can be read as a signature of a temperature variation.
Following this technique, Harrison et al. (1999), using lines such
as O iii 599, O iv 554 and O v 629 š A, concluded that no significant
teriaca—final.tex; 23/01/2001; 14:40; p.12

Electron density variations during UV transient events 13
temperature variation occurs during blinkers. In our case we have al­
lowed transitions (O iii 703.845 and O iii 703.85) and an intersystem
transition (O iv 1401.16), and we are interested in checking whether
the same technique can be used here. In an optically thin plasma the
line intensity for a transition from the upper level u to the lower level
l is given by:
I lu = 1
4ú
Z
h
G(N e ; T e ) lu N 2
e dh; (2)
where G(N e ; T e ) lu is the Contribution function (containing all the density­
and temperature­dependent terms). This can be written as:
G(N e ; T e ) lu = ffl(– lu )
N e
N ion
N el
N el
NH
NH
N e
; (3)
where N ion /N el is the relative abundance of the ionic specie obtained
from the ionization balance calculations (containing a strong temper­
ature dependence), N el /NH is the element abundance with respect to
hydrogen and NH /N e the hydrogen abundance relative to the electron
density (¸ 0:8). ffl(– lu ) is the emissivity of a spectral line (erg cm \Gamma3
s \Gamma1 ) normalized to the number density of the emitting ion N ion given
by:
ffl(– lu ) = hc
– lu
N u
N ion
A ul ; (4)
where N u is the population of the upper level u and A ul is the spon­
taneous transition probability from u to l. In the case of an isothermal
plasma in ionization equilibrium, the ionic fraction can be removed
from the integral obtaining:
I lu = 1
4ú
N ion
N el
N el
NH
NH
N e
Z
h
ffl(– lu )
N e
N 2
e dh: (5)
In Figure 6 we show ffl(– lu )/N e as a function of the electron density
for O iii 703.845, O iii 703.85, O iv 1401 and O iv 1404 (solid lines).
The trend for the allowed lines used by Harrison et al.(1999) is also
shown for comparison (dashed lines). The vertical dotted line indicates
the maximum density found in DS1 (1:7 10 10 cm \Gamma3 ). It is possible
to observe how, for densities below this value, the O iii 703 and the
O iv 1401 lines show the same density dependence (within 3% for
N e Ÿ 10 10 cm \Gamma3 and within 8 % for N e Ÿ 1:7 10 10 cm \Gamma3 ). However,
an increase of the electron density (for densities above 10 10 cm \Gamma3 ) will
lead to a reduction of the ratio O iv 1401.16/(O iii 703.845 + O iii
703.85) (hereafter RT ), while an increase in temperature will increase
it. Thus, RT can be used as an indicator of temperature increases.
teriaca—final.tex; 23/01/2001; 14:40; p.13

14 Teriaca et al.
Figure 7. Electron density obtained along EE 1. The horizontal dashed line indicates
the average density over the whole DS1 (see Table II). (a): Values obtained assuming
a fixed 65 % contribution of O iv 1404.81 on the whole 1404 š A feature. (b): Values
obtained estimating locally the amount of O iv 1404.81 in the whole 1404 š A feature
(see text for details). (c): O iii 703 (703.845 + 703.85) (open circles) and O iv 1401.16
(filled circles) line intensities along EE 1 as function of time. (d): O iv 1401.16 to
O iii 703 line ratio (RT within the text) as function of time. Side panels: Slit images
before (g), during (h) and after (i) the explosive event.
4. Results
Results obtained for DS1 and DS2 are shown in Table II and Figure 7
and 9, and in Table III and Figure 8, respectively. The two datasets
considered here have an important difference. As already discussed in
x 3, DS1 represents the temporal evolution of a strip of the solar surface
(1 \Theta 120 arcsec 2 ) over ¸ 6 minutes interval (after which the slit has
moved to a new position), while DS2 permits us to build­up a raster
teriaca—final.tex; 23/01/2001; 14:40; p.14

Electron density variations during UV transient events 15
Figure 8. (a): Electron density values obtained along EE 2. The horizontal dashed
line indicates the average density over the whole dataset (see Table III). These values
were obtained assuming a fixed 65 % contribution of O iv 1404.81 on the whole
1404 š A feature. (b): Enhancement factor of the Si iv line wings (see Equation 1).
Lower panels: Slit images at three specified locations.
of a region 8 \Theta 120 arcsec 2 with a resolution of 0.3 and 1 arcsec in
the X and Y directions, respectively (see Figure 2, right panel). It is
evident that DS2 allows us to identify network and internetwork areas
in a more reliable way, while DS1 gives us the possibility of evaluating
the importance of the line blends (see x 3.2). The average contribution
of O iv 1404.81 to the total 1404 spectral feature (hereafter R O IV
1404 )
has been evaluated to be 0.65. This value has been adopted for all the
measurements in DS2. For DS1 we will discuss the results obtained
evaluating locally the R O IV
1404 , unless specified differently.
4.1. Electron density in network and internetwork
Areas of DS1 and DS2 have been selected with the purpose of measuring
the electron density in different regions of the solar atmosphere. An
average over the whole datasets (ALL 1 and ALL 2) was performed in
teriaca—final.tex; 23/01/2001; 14:40; p.15

16 Teriaca et al.
order to obtain average density values that can be considered as typical
for the quiet Sun. N e values of (4.2 +0:6
\Gamma0:5 )\Theta10 9 and (9.0 +0:7
\Gamma0:7 )\Theta10 9 cm \Gamma3
for DS1 and DS2, respectively, were obtained. Electron density values
of (4.8 +0:9
\Gamma0:9 )\Theta10 9 and (8.5 +1:0
\Gamma1:1 )\Theta10 9 cm \Gamma3 have been obtained in cell
center areas CC 1 and CC 2, respectively, while values of (4.5 +0:7
\Gamma0:7 )\Theta10 9
and (3.9 +0:6
\Gamma0:6 )\Theta10 9 cm \Gamma3 are derived in network areas N 1 and N 2 (see
Table II and III). The above results lead to ratios between the electron
density in cell centers and network of 1.1 \Sigma 0.3 and 2.18 \Sigma 0.3 for
DS1 and DS2, respectively. These values are in agreement with recent
results from Del Zanna and Bromage (1999), who report a value of 1.9
\Sigma 0.7.
It is interesting to note that, in DS1, we obtain a higher value of
N e in the cell center than in the network, only when the local R O IV
1404 is
calculated. If the R O IV
1404 values obtained for CC 1 (0.64) and N 1 (0.69)
were assumed representative of cell center and network conditions and
used to calculate the density in CC 2 and N 2, a still higher ratio
between N e in cell center and in the network in DS2 (3.18 \Sigma 0.4) could
be found. The fact that the R O IV
1404 is higher in the network than in the
cell centers agrees with the fact that the network contrast is higher in
O iv than in O iii (Gallagher et al., 1998). However, in the region BN
of DS2 an electron density of (8.7 +1:0
\Gamma1:1 )\Theta10 9 cm \Gamma3 was found, twice the
N e value found in region N 2 of the same dataset.
4.2. Explosive events: EE 1 and EE 2
Figure 7 presents the temporal variation of the electron density along
EE 1. In Figure 7(a) the electron density, calculated assuming a fixed
R O IV
1404 of 0.65, is shown as a function of time, while in Figure 7(b)
values obtained calculating the local values of R O IV
1404 are reported (see
x 3.2). Figure 7(c) shows the intensities of the O iv 1401 (filled circles)
and O iii 703 (open circles) lines, while the ratio between the intensities
of the former and the latter line is showed in Figure 7(d). For reasons of
clarity all data points along EE 1 shown in Figure 7 have been identified
with a letter from a to j.
A signature of density variation is evident during the first 300 sec­
onds (a -- d), with a variation of a factor of (¸ 2:5) between a and b.
The variation is still large (¸ 3) when the local R O IV
1404 value is used
(see Figure 7(b)). During the same time interval the intensity of both
lines decreases maintaining a constant ratio RT (Figure 7(d)) in all the
points except a, in which a slight increase is observed. In a, an excess in
the O iv 1401 line width was detected (see x 3.3). After t = 350 s (see
Figure 7) the slit has moved to a different region of the Sun. During
the last 400 seconds (e -- j), the intensity of both lines is more or less
teriaca—final.tex; 23/01/2001; 14:40; p.16

Electron density variations during UV transient events 17
constant with the remarkable exception of h. This is the point where a
strong explosive event was observed (see x 3.3) in the blue wing of the
Si iv 1402 line and in O iv 1401.16 š A, with an observed electron density
enhancement of a factor ¸ 3 with respect to the pre (g) and post­event
(i) values. However, the correction at the time of the explosive event
(see Table II, row in bold) is quite problematic due to the difficulties
in evaluating the two components of the 1407 blend, in the case that
one or both of them could present non­Gaussian components. This
would imply a lower limit to the density variation associated with the
explosive event. In fact, there is clear evidence that still lower values of
density are observed in f and j (with values ¸ 7 and 16 times lower than
at the peak of the explosive event, respectively, depending whether a
constant R O IV
1404 value is used or not, see Table II).
In f and j the intensity of the observed lines is particularly low, lead­
ing to large errors in the determination of RD . This coupled with the
fact that the values falls in the steepest part of the theoretical curve of
electron density versus intensity line ratio (see Figure 5), leads to large
errors in the final electron density. The observed factor of ¸ 3 increase
in the electron density can, hence, be regarded as a lower limit for the
electron density increase. Another reason to consider the above value as
a lower limit comes from the dimensions of the integrated area in space
and time. In order to have a sufficient S:N ratio, we binned over 40
seconds in time (in point h) and over 7 spatial pixels in the Y direction
(¸ 5000 km). These values must be compared with an explosive event
average life of 60 seconds (Dere, 1994) and typical sizes of 2000 km.
The larger integration area, in particular, has probably `diluted' the
signature of the explosive event in the quiet Sun background. A sharp
increase of RT , in correspondence with h, is visible in Figure 7(c).
In x 3.4, we have seen how the above line ratio could be used as a
probe of the temperature variation in the solar atmosphere, provided
the electron density is Ÿ 2:5 10 10 cm \Gamma3 . At the density values observed
in h, the quantity ffl(– lu )/N e for O iv 1401 starts to deviate from a
constant value (see Figure 6). In these conditions, an increase in the
electron density (at constant temperature) will lead to a reduction in
the O iv/O iii ratio, so the observed increase in the line ratio can­
not be due to the different response of the two lines to an increasing
electron density. On the other hand, the value of RT in h is probably
overestimated.
The problem arises in the determination of the O iii 703 line in­
tensity, which includes both O iii 703.845 and 703.85. In h, signatures
of the explosive event are observed in Si iv 1402 and in O iv 1401.
These two lines have formation temperatures of 6:4 10 4 and 1:6 10 5 K,
respectively. Since the formation temperature of O iii (10 5 K) is just
teriaca—final.tex; 23/01/2001; 14:40; p.17

18 Teriaca et al.
Table III. Results of the 1997 May 31 dataset
(DS2).
Time Ratio Ne
(min.) (cm \Gamma3 )
1 -- 8 2.05 \Sigma 0.11 (2.7 +0:9
\Gamma0:9 ) \Theta 10 9
9 -- 16 2.57 \Sigma 0.13 (7.4 +1:4
\Gamma1:3 ) \Theta 10 9
17 -- 24 2.70 \Sigma 0.12 (8.9 +1:4
\Gamma1:3 ) \Theta 10 9
25 -- 28 3.33 \Sigma 0.16 (1.7 +0:3
\Gamma0:3 ) \Theta 10 10
29 -- 30 3.80 \Sigma 0.21 (2.6 +0:5
\Gamma0:4 ) \Theta 10 10
31 -- 32 3.40 \Sigma 0.15 (1.9 +0:3
\Gamma0:3 ) \Theta 10 10
33 -- 34 3.80 \Sigma 0.19 (2.6 +0:5
\Gamma0:4 ) \Theta 10 10
35 -- 36 3.40 \Sigma 0.20 (1.9 +0:4
\Gamma0:3 ) \Theta 10 10
37 -- 38 3.35 \Sigma 0.23 (1.8 +0:4
\Gamma0:4 ) \Theta 10 10
39 -- 41 3.20 \Sigma 0.19 (1.5 +0:3
\Gamma0:3 ) \Theta 10 10
42 -- 44 3.06 \Sigma 0.18 (1.3 +0:3
\Gamma0:2 ) \Theta 10 10
45 -- 48 3.00 \Sigma 0.16 (1.3 +0:2
\Gamma0:2 ) \Theta 10 10
49 -- 52 3.40 \Sigma 0.18 (1.9 +0:3
\Gamma0:3 ) \Theta 10 10
53 -- 56 3.80 \Sigma 0.22 (2.6 +0:6
\Gamma0:5 ) \Theta 10 10
57 -- 60 3.57 \Sigma 0.17 (2.2 +0:4
\Gamma0:3 ) \Theta 10 10
61 -- 64 3.26 \Sigma 0.19 (1.6 +0:3
\Gamma0:3 ) \Theta 10 10
65 -- 72 3.18 \Sigma 0.15 (1.5 +0:2
\Gamma0:2 ) \Theta 10 10
73 -- 80 2.64 \Sigma 0.11 (8.2 +1:3
\Gamma1:2 ) \Theta 10 9
81 -- 88 3.01 \Sigma 0.18 (1.3 +0:3
\Gamma0:2 ) \Theta 10 10
89 -- 96 2.56 \Sigma 0.16 (7.3 +1:7
\Gamma1:6 ) \Theta 10 9
97 -- 104 2.25 \Sigma 0.13 (4.4 +1:2
\Gamma1:1 ) \Theta 10 9
105 -- 112 2.20 \Sigma 0.13 (4.0 +1:2
\Gamma1:1 ) \Theta 10 9
1 -- 112 (a) 3.04 \Sigma 0.07 (1.3 +0:1
\Gamma0:1 ) \Theta 10 10
ALL 2 (b) 2.71 \Sigma 0.06 (9.0 +0:7
\Gamma0:7 ) \Theta 10 9
N 2 2.19 \Sigma 0.07 (3.9 +0:6
\Gamma0:6 ) \Theta 10 9
CC 2 2.67 \Sigma 0.10 (8.5 +1:0
\Gamma1:1 ) \Theta 10 9
BN 2.69 \Sigma 0.07 (8.7 +1:0
\Gamma1:1 ) \Theta 10 9
(a) : Averaged over the entire EE 2 region
(b) : Represents the average over the whole dataset
in the middle of the previous two temperatures, there is no reason to
think that signatures of the explosive event should not be present also
in the O iii 703 š A line. As discussed in x 3.3, this line (including O iii
703.845 and 703.85 observed as one feature) is blended on its blue side
teriaca—final.tex; 23/01/2001; 14:40; p.18

Electron density variations during UV transient events 19
with the 1407.38 O iv line. When a multi­Gaussian fit is performed,
the O iii line parameters can be easily determined yielding values very
close to the ones observed in g and i. This can be interpreted as the
emission coming from background and foreground plasma. In fact, in
many explosive events it is possible to identify a rest component with
characteristics very close to the ones observed in the surrounding `quiet'
plasma (Chae et al., 1998b; Landi et al., 2000). The difficulties come
when we look at the O iv line parameters. In this case, we find a
very broad profile interpreted as the sum of the O iv background and
foreground components plus the enhanced components of O iii and
O iv. Anyway, even if all the excess (with respect to the value in g) of
the intensity observed in the broad component is assumed as coming
from O iii, the value of RT in h is still higher (¸ 30 %) than in g and
i. This suggests that some increase in temperature is occurring during
the explosive event. It is worth noting that an increase of RT is also
observed in a, where O iv 1401 shows enhanced broadening that could
be interpreted as a small dynamic event. It is important to point out
that the same reasoning about the under­estimation of the O iii line
intensity during the explosive event leads to an over­estimation of the
amount of O iv 1404.81 and, hence, again to an under­estimation of
the electron density during the explosive event.
In Table III and Figure 8, the results obtained along the region
EE 2 are presented. In Figure 8(a), the N e values along EE 2 (obtained
with a constant R O IV
1404 of 0.65) are shown, while in Figure 8(b) the
wing enhancement quantity W e (see Equation 1) is reported. A strong
correlation between the electron density and the quantity W e appears
evident in Figure 8. It is important to remember that, in this dataset,
the slit moves to a different area of the Sun after ¸ 4 minutes. This
means that the values shown in Figure 8(a) and 8(b) may result from
both spatial and temporal evolution of the transition region plasma.
However, it appears clear that density values ¸ 3 times higher than
the average over the entire dataset are associated with explosive event­
like phenomena. Also in this case the considerations about the size in
space and time of the integration area, with respect to the typical values
for an explosive event, are valid. Along EE 2 we have an integration
in time that, in the best case, is equal to 2 min. and a spatial average
over ¸ 4300 km. This suggests that the above electron density increase
can, once again, be considered as a lower limit.
4.3. The UV blinker
In Figure 9 it is possible to observe the temporal variation of N e along
the region BL in DS1. In Figure 9(a) the electron density, calculated
teriaca—final.tex; 23/01/2001; 14:40; p.19

20 Teriaca et al.
Figure 9. Electron density obtained along BL. The horizontal dashed lines indicate
the average density over the whole DS1 (see Table II). (a): Values obtained assuming
a fixed 65 % contribution of O iv 1404.81 on the whole 1404 š A feature. (b): Values
obtained estimating the contribution of O iv 1404.81 on the whole 1404 š A feature
through the O iii 703 line (see text for details). (c): O iii 703 (open circles) and
O iv 1401.16 (filled circles) line intensities along BL as a function of time. (d): O iv
1401.16 to O iii 703 line ratio (RT within the text) as function of time.
assuming a constant R O IV
1404 of 0.65, is shown as a function of time,
while in Figure 9(b) we report the N e values obtained through the
local R O IV
1404 values (see x 3.2). Figure 9(c) shows the intensities of
the O iv 1401 (filled circles) and O iii 703 (open circles) lines, while
the ratio between the intensities of the former and the latter line is
showed in Figure 9(d). It is evident that, when a constant R O IV
1404 is
used, no appreciable N e variation is detected (see Figure 9(a)). The
general picture partially changes when the local R O IV
1404 value is applied.
However, the absence of changes in the electron density during the first
teriaca—final.tex; 23/01/2001; 14:40; p.20

Electron density variations during UV transient events 21
300 seconds remains confirmed, but the situation changes in the second
part of the dataset. Starting at t = 400 s, N e has reduced by a factor
of ¸ 2 with respect to previous values. After this point, we observe an
increase in N e close to the values shown during the first 300 seconds.
The RT value remains constant for the first three data points (see
Figure 9(d)) despite the occurrence of a variation of a factor 1.5 in the
line intensities (see Figure 9(c)). The selection of the areas of DS1 over
which the binning along BL is performed was done in order to get the
region with enhanced intensity inside a single data point (see x 3.3).
Nevertheless, the brightening is not uniform across the area defining
the second data point. In order to verify whether the electron density
is related with the line intensity, a sub­region (two successive spectra
times three spatial pixels along solar Y) of the above region was selected
around the brightest pixel. Integrating over this sub­region we obtain
a higher line intensity (in all lines) but the same electron density as in
the second data point.
In Figure 10(a) the spectral profiles around the O iv 1401 line for
the first three data points are shown. It is interesting to observe that,
for the second data point (dotted line), the O iv 1401 line is broader
than for the previous and the following points but without an evident
signature of high velocity components like the ones observed during
explosive events. The line appears also blue­shifted (¸ \Gamma5 km s \Gamma1 )
compared with the other two data points. When the smaller sub­region
centered around the peak of the brightening is considered, the blue­
shift increases up to ¸ \Gamma11 km s \Gamma1 . During the last 400 seconds a
steady decrease of RT is observed, while the line intensities do not show
any appreciable variation. During the same time period the density
appears to increase by a factor of ¸ 2. At electron densities around
10 10 cm \Gamma3 , RT should be scarcely affected by the variation in N e (see
x 3.4). However, the O iii 703 line has been used to determine the R O IV
1404
(influencing the final determination of N e ) as well as the RT values (see
Figure 9(d)). This can lead, in the case of an under­estimation in the
O iii line intensity, to an under­estimation of N e (through the R O IV
1404
value) as well as to an over­estimation of RT . We are confident of our
measurements of the O iii 703 line intensity, but the fact that part of
the observed behaviour in the last 400 seconds of Figure 9(b) and 9(d)
could be due to the above effect can not be excluded. In any case, RT is
constant within a factor of 1.2. This is also the level found by Harrison
et al. (1999) within which the O iv/O iii ratio is constant (see Figure 6
in that paper). In Figure 10(b) the spectra around the O iv 1401 line
relative to the fourth, fifth, sixth and seventh data points (shown in
Figure 9) are reported.
teriaca—final.tex; 23/01/2001; 14:40; p.21

22 Teriaca et al.
Figure 10. Spectral profiles around the O iv 1401.16 line from BL. Note the blue
wing of the Si iv 1402 line partially visible on the right edge of the picture and part
of the second order Ar viii 700.245 on the left edge. The chromospheric S i line at
1401.5136 š A is also clearly visible on the red wing of the stronger O iv 1401 line.
In (a) the first three data points of Figure 9 are shown, while in (b) the successive
four data point are reported.
5. Discussion
In the last decades, the idea of magnetic reconnection as a key process
in releasing the magnetic energy, considered the source of numerous
solar activity phenomena (from coronal heating to solar flares), has
gained importance (Giovanelli, 1946; Gold and Hoyle, 1960; Priest,
1981; Parker, 1988; Forbes, 1991). In the solar atmosphere all the
magnetic structures are rooted in the photosphere where the plasma
dominates over the magnetic field (plasma fi ? 1). In these conditions
photospheric motions displace the magnetic structures and opposite
polarity fields may, eventually, meet. When this happens the magnetic
field may reconnect through the formation of a current sheet, from
which plasma is ejected in both directions along the magnetic field
lines with velocities of the order of the Alfv'en speed. In this context,
observations of bi­directional jets with velocities up to 150 km s \Gamma1
(Innes et al., 1997b) together with the detection of associated episodes
of photospheric magnetic flux cancellation (Dere et al., 1991; Dere,
1994; Chae et al., 1998a), strongly suggest magnetic reconnection as a
possible mechanism for UV explosive events.
teriaca—final.tex; 23/01/2001; 14:40; p.22

Electron density variations during UV transient events 23
Recently, Karpen et al. (1995) have performed 2.5 MHD modelling
of the interaction between two magnetic dipoles subjected to uniform
shearing. Together with the formation of high velocity flows along the
current sheet, the authors also report the formation of features with
electron densities from 4 to 10 times higher than the pre­event values.
Such results have been confirmed in a more recent work by the same
authors (Karpen et al., 1998) in which two dipoles of markedly unequal
field strength were considered. Also Roussev et al. (2000) have recently
presented results on the modelling of reconnection jets, again finding
electron density increases of a factor of 2--4 in the transition region. In
x 4.2 we have shown evidence of enhancements of at least a factor of 3 in
the electron density associated with UV explosive events, together with
an indication of a possible temperature increase. These observations
give further support to the idea that explosive events are signatures of
magnetic reconnection occurring in the solar atmosphere.
Another result concerns the evolution of the electron density during
an UV blinker (see x 4.3). The blinker studied in the present work
shows a duration time of ¸ 200 seconds and, hence, can be considered
belonging to the short­duration tail of the blinker distribution found
by Harrison et al. (1999), similarly to the events observed by Gallagher
et al. (1999). Both of the previous studies were carried out using the
CDS spectrometer on­board SoHO. The higher spectral resolution of
SUMER permits us to exclude the presence of high velocity motions
such as the ones observed during explosive events, at least for the
present case (see Figure 10). However, the spectral profile relative to
the peak of the blinker (profile 2 in Figure 10a), shows an increased
broadening (from an average observed FWHM of 0.255 š A to a value of
0.285 š A) and a blue­ward shift of ¸ \Gamma5 km s \Gamma1 . Chae et al. (1998b)
suggest that CDS blinkers could consist of several small­scale, short
lived SUMER `unit brightening events' with a size of a few arcsec and
a lifetime of a few minutes, and characterized by a spectral profile that
are not as broad as those of explosive events but still with significatively
enhanced wings. They have also suggested that both explosive events
and blinkers are due to magnetic reconnection with the differences in
the spectral profiles arising from different magnetic geometries.
Harrison et al. (1999) proposed a schematic model for UV blinkers,
in which newly emerging magnetic flux in cell centers is carried by
the convective motions towards the network, where the magnetic field
is concentrated (see Figure 13 in that paper). Here the fields merge
and, eventually, reconnect through the formation of current sheets. The
formation of a current sheet and the resulting formation of shocks and
jets should lead to an increase in the electron density (as observed in
the case of explosive events), while the thermalization of the accelerated
teriaca—final.tex; 23/01/2001; 14:40; p.23

24 Teriaca et al.
particles should lead to an increase in the plasma temperature. In
the case considered here, no variation in density and temperature is
observed, despite an intensity increase of 1.5 -- 2 times. This, together
with the absence of high velocity motions, strongly suggests that the
observed brightening is due to an emission measure increase result­
ing from the merging of fields carrying plasma at transition region
temperatures in the network, as proposed by Harrison et al. (1999).
In this context, differences between blinkers and other type of UV
transient events could depend on whether magnetic reconnection takes
place or not (perhaps also depending on the geometry of the magnetic
field). In this view blinkers, Chae's `unit brightening events' and explo­
sive events could be different examples of progressively higher presence
of magnetic reconnection in regions of merging magnetic field.
Acknowledgements
We would like to thank the anonymous referee for very helpful com­
ments and suggestions. Research at Armagh Observatory is grant­aided
by the N. Ireland Dept. of Culture, Arts and Leisure, while partial
support for software and hardware is provided by the STARLINK
Project which is funded by the UK PPARC. This work was supported
by PPARC grant PPA/G/S/1999/00055. The SUMER project is finan­
cially supported by DLR, CNES, NASA, and PRODEX.
References
Brage, T., Judge, P.G., and Brekke, P.: 1996, Astrophys. J. 464, 1030.
Brueckner, G.E. and Bartoe, J.­D.F.: 1983, Astrophys. J. 272, 329.
Chae, J., Wang, H., Lee, C.Y., Goode, P.R., and Sch¨uhle, U.: 1998a, Astrophys. J.
497, L109.
Chae, J., Wang, H., Lee, C.Y., Goode, P.R., and Sch¨uhle, U.: 1998b, Astrophys. J.
504, L123.
Cook, J.W., Keenan, F.P., Dufton, P.L., Kingston, A.E., Pradhan, A.K., Zhang,
H.L., Doyle, J.G., and Hayes, M.A.: 1995, Astrophys. J. 444, 936.
Curdt, W., Feldman, U., Laming, J.M., Wilhelm, K., Sch¨uhle, U., and Lemaire, P.:
1997, Astron. Astrophys. Suppl. S. 126, 281.
Del Zanna, G. and Bromage, B.J.I.: 1999, J. Geophys. Res. 104, 9753.
Dere, K.P., Bartoe, J.­D.F., and Brueckner, G.E.: 1982, Astrophys. J. 259, 366.
Dere, K.P., Bartoe, J.­D.F., and Brueckner, G.E.: 1989, Solar Phys. 123, 41.
Dere, K.P., Bartoe, J.­D., Brueckner, G.E., Ewing, J., and Lund, P.: 1991, J. Geophys
Res. 96, 9399.
Dere, K.P.: 1994, Adv. Space Res. 14(4), 13.
teriaca—final.tex; 23/01/2001; 14:40; p.24

Electron density variations during UV transient events 25
Doschek, G.A., Feldman, U., Laming, J.M., Warren, H.P., Sch¨uhle, and Wilhelm,
K.: 1998, Astrophys. J. 507, 991.
Doschek, E.E., Laming, J.M., Doschek, G.A., Feldman, U., and Wilhelm, K.: 1999,
Astrophys. J. 518, 909.
Doyle, J.G., Kingston, A.E., and Reid, R.H.: 1980, Astron. Astrophys. 90, 97.
Doyle, J.G., Teriaca, L., and Banerjee, D.: 2000, Astron. Astrophys. 356, 335.
Dwivedi, B.N. and Gupta, A.K.: 1991, Solar Phys. 138, 283.
Feldman, U., Doschek, G.A., and Patterson, N.P.: 1976, Astrophys. J. 209, 270.
Feldman, U. and Doschek, G.A.: 1979, Astron. Astrophys. 79, 357.
Forbes, T.G.: 1991, Geophys. Astrophys. Fluid Dyn. 62, 15.
Gallagher, P.T., Phillips, K.J.H., Harra­Murnion, L.K., and Keenan, F.P.: 1998,
Astron. Astrophys. 335, 733.
Gallagher, P.T., Phillips, K.J.H., Harra­Murnion L.K., Baudin F., and Keenan F.P.:
1999, Astron. Astrophys. 348, 251.
Giovanelli, R.G.: 1946, Nature 158, 81.
Gold, T. and Hoyle, F.: 1960, MNRAS 120, 89.
Hayes, M. and Shine, R.A.: 1987, Astrophys. J. 312, 943.
Harrison, R.A., Sawyer, E.C., Carter, A.M. et al.: 1995, Solar Phys. 162, 233.
Harrison, R.A.: 1997, Solar Phys. 175, 467.
Harrison, R.A., Fludra, A., and Pike C.D., et al.: 1997, Solar Phys. 170, 123.
Harrison, R.A., Lang, J., Brooks, D.H., and Innes, D. E.: 1999, Astron. Astrophys.
351, 1115.
Innes, D.E., Brekke, P., Germerott, D., and Wilhelm, K.: 1997a, Solar Phys. 175,
341.
Innes, D.E., Inhester, B., Axford, W.I., and Wilhelm, K.: 1997b, Nature 386, 811.
Jordan, C.: 1996, Astrophys. J. Suppl. S. 237, 13.
Judge, P.G., Hansteen, V., WikstÜl, ü., Wilhelm, K., Sch¨uhle, U., and Moran, T.:
1998, Astrophys. J. 502, 981.
Karpen, J.T., Antiochos, S.K., and DeVore, C.R.: 1995, Astrophys. J. 450, 422.
Karpen, J.T., Antiochos, S.K., DeVore, C.R., and Golub, L.: 1998, Astrophys. J.
495, 491.
Landi, E., Landini, M., Dere, K.P., Young, P.R., and Mason, H.E.: 1999, Astron.
Astrophys. Suppl. S. 135, 339.
Landi, E., Mason, H.E., Lemaire, P., and Landini, M.: 2000, Astron. Astrophys. 357,
743.
Lemaire, P., Wilhelm, K., and Curdt, W. et al.: 1997, Solar Phys. 170, 105.
Madjarska, M.S., Vial, J.­C., Bocchialini, K., and Dermendjiev, V.N., 1999, in
Proceedings 8th SoHO workshop, Paris, ESA SP­446, 467.
Mariska, J.T.: 1992, The Solar Transition Region, Cambridge University Press,
Cambridge.
Mason, H.E., and Monsignori Fossi, B.C.: 1994, Astron. Astrophys. Rev. 6, 123.
O'Shea, E., O'Neill, T., Keenan, F.P. and Doyle, J.G.: 2000, Solar Phys. 196, 321.
Parker, E.N.: 1988, Astrophys. J. 330, 474.
P'erez, M.E., Doyle, J.G., Erd'elyi, R., and Sarro, L.M., 1999a, Astron. Astrophys.
342, 279.
P'erez, M.E., Doyle, J.G., O'Shea, E., and Keenan, F.P.: 1999b, Astron. Astrophys.
351, 1139.
Porter, J.G. and Dere, K. P.: 1991, Astrophys. J. 370, 775.
Priest, E.R.: 1981, Solar Flare Magnetohydrodynamics, Gordon & Breach, London.
Roussev, I., Galsgaard, K., Erd'elyi, R., and Doyle, J.G.: 2000, Astron. Astrophys.
(submitted)
teriaca—final.tex; 23/01/2001; 14:40; p.25

26 Teriaca et al.
Teriaca, L.: 2001, PhD thesis (Queens University/Armagh Observatory).
WikstÜl, ü., Judge, P.G., and Hansteen, V.: 1997, Astrophys. J. 483, 972.
WikstÜl, ü., Judge, P.G., and Hansteen, V.: 1998, Astrophys. J. 501, 895.
Wilhelm, K., Curdt, W., Marsch, E., Sch¨uhle, U., Lemaire, P., Gabriel, A., Vial,
J.­C., Grewing, M., Huber, M.C.E., Jordan, S.D., Poland, A.I., Thomas, R.J.,
K¨uhne, M., Timothy, J.G., Hassler, D.M., and Siegmund, O.H.W.: 1995, Solar
Phys. 162, 189.
Wilhelm, K., Lemaire, P., Curdt, W. et al.: 1997, Solar Phys. 170, 75.
Wilhelm, K., Innes, D.E., Curdt, W., Kliem, B., and Brekke, P.: 1998, in Solar Jets
and Coronal Plumes, Guadalupe, France, ESA­SP 421, 103.
teriaca—final.tex; 23/01/2001; 14:40; p.26