�������� ���� �� ���� ��������� ������. ����� ������������� ��������� : http://star.arm.ac.uk/preprints/421.ps ���� ���������: Wed Jun 16 19:44:07 2004 ���� ��������������: Mon Oct 1 21:38:33 2012 ���������: ��������� �����: coronal hole

Astronomy & Astrophysics manuscript no. 1069 June 16, 2004
(DOI: will be inserted by hand later)
CDS wide slit time�series of EUV coronal bright points
I. Ugarte�Urra 1 , J.G. Doyle 1 , V.M. Nakariakov 2 , and C.R. Foley 3
1 Armagh Observatory, College Hill, Armagh BT61 9DG, N. Ireland
2 Physics Department, University of Warwick, Coventry, CV4 7AL, UK
3 Mullard Space Science Laboratory, University College London, Holmbury St. Mary, Dorking, Surrey RH5 6NT, UK
Received / Accepted
Abstract. Wide slit (90 ## � 240 ## ) movies of four Extreme Ultraviolet coronal bright points (BPs) obtained with the Coronal
Diagnostic Spectrometer (CDS) on board the Solar and Heliospheric Observatory (SoHO) have been inspected. The wavelet
analysis of the He # 584.34 �, O # 629.73 � and Mg ###/## 368 � time�series confirms the oscillating nature of the BPs, with
periods ranging between 600 and 1100 seconds. In one case we detect periods as short as 236 seconds. We suggest that these
oscillations are the same as those seen in the chromospheric network and that a fraction of the network bright points are most
likely the cool footpoints of the loops comprising coronal bright points. These oscillations are interpreted in terms of global
acoustic modes of the closed magnetic structures associated with BPs.
Key words. Sun: oscillations -- Sun: corona -- Sun: transition region -- Sun: chromosphere -- Sun: UV radiation -- Sun: magnetic
fields
1. Introduction
Coronal oscillations have been studied for more than thirty
years in several wavelength ranges (Aschwanden 2003, and
references therein). In recent years, observations made with
the Transition Region And Coronal Explorer (TRACE) and
instruments onboard the Solar and Heliospheric Observatory
(SoHO) have provided evidence of these oscillations in dif�
ferent scenarios: sunspots, TRACE loops, polar plumes, etc.
(e.g. O'Shea et al. 2001; Banerjee et al. 2002; Marsh et al.
2002; Deforest & Gurman 1998). Another good candidate to
host coronal oscillations are Extreme Ultraviolet (EUV) coro�
nal bright points (BPs). They are identified as small (10 -- 30
Mm) regions of enhanced emission over the quiet Sun and
coronal holes. They comprise tiny loops (Sheeley & Golub
1979) and lie over the network boundaries of the supergranu�
lar cells (Egamberdiev 1983), which are characterized for their
oscillating nature (Dame et al. 1984). BPs are the result of the
interaction of opposite magnetic polarities with magnetic re�
connection likely to play a major role in the BP appearance
(Priest et al. 1994). In this respect, BPs have been seen to dis�
appear at coronal temperatures after the full cancellation of one
of the magnetic polarities (Madjarska et al. 2003). These au�
thors also commented on small�scale brightenings within the
BP which showed velocity variations in the range 3 -- 6 km s -1 .
The analysis of the transition region line S ## 933 � flux fluc�
tuations showed clear evidence for a period just under 500 s
(Ugarte�Urra et al. 2004). These authors commented on oscil�
Send o#print requests to: I. Ugarte�Urra (iuu@star.arm.ac.uk),
http://star.am.ac.uk/preprints/
lations seen in another BP in O # 629 � which looked like a
damped wave.
Over the last years of successful space observations it has
become clear that it is necessary to combine observations of the
same phenomenon from di#erent instruments with di#erent ca�
pabilities. In the case of BPs, the role of the magnetic field is
crucial in the evolution of the features and whenever possible
its study should complement the imaging and spectral analysis.
Here, we extend the BP oscillation study by looking at time�
series data taken with the Coronal Diagnostic Spectrometer in
its wide�slit mode, coupled with data from other instruments
on board SoHO. Observations are discussed in Sect. 2, Sect.
3 describes the wavelet analysis results of the study of four
BPs, while Sect. 4 establishes a comparison with the magnetic
field evolution in one of the cases. In Sect. 5, we discuss the
results in the context of the chromospheric oscillations in the
network, as well as, the possible wave mechanisms and recon�
nection models. The conclusions are given in Sect. 6.
2. Observations
2.1. CDS
The Coronal Diagnostic Spectrometer (CDS) onboard SoHO
was designed to determine the characteristics of the solar at�
mosphere plasma through the study of the emission lines char�
acteristics in the 150 -- 800 � EUV spectral range (Harrison
et al. 1995). Two spectrometers, a normal (NIS) and a graz�
ing incidence (GIS), can be used for that purpose. In the case
of NIS, the selection of a wide slit (90 ## � 240 ## ) allows one

2 I. Ugarte�Urra et al.: CDS wide slit time�series of EUV coronal bright points
Table 1. Details of the CDS wide slit (90 ## � 240 ## ) observational programs, including their identification number, acronym, date and time of
observation, center coordinates of the slit for that time in arcseconds, total duration in minutes, exposure time in seconds for each of the images
and average time cadence in seconds with its standard deviation.
Program Acronym Date � Starting time Coordinates [ ## ] Duration [min] Exp. time [s] Avg. cadence [s]
s6744 CHJET v1 1997/01/20 � 13:23 UT 0,--992 128 60 70.4 � 1.6
s6829 CHJET v1 1997/01/30 � 09:30 UT --31,--982 128 60 70.4 � 1.6
s6840 CHJET v1 1997/01/31 � 08:02 UT --31,--984 93 60 70.0 � 1.7
s6697 MOVIE V1/v3 1997/01/16 � 08:02 UT 0,0 218 25 31.2 � 0.4
to obtain high temporal cadence images of a target at the ex�
pense of spectral resolution. Therefore, we can create a time�
series of a specific solar region at a cadence of 1 image ev�
ery 30 seconds, as seen by di#erent spectral lines with dif�
ferent formation temperatures. As images produced by close
lines would overlap, we are constrained to the study of iso�
lated and unblended lines. This mode has proven to be very
useful in the study of di#erent solar features: blinkers, coronal
jets, prominences, quiet Sun network and sunspots. The present
BP observations were done in 1997 (see Table 1), using the
wide slit in a sit�and�stare mode, i.e. the pointing is kept fixed
and the plasma moves under the slit. Three bright lines of the
NIS spectral range were used: He # 584.34 �, O # 629.73 �
and Mg ## 368.07 �. Their formation temperature in ionization
equilibrium is respectively, in logarithmic scale, 4.5, 5.4 and
6.0 (Mazzotta et al. 1998; Young et al. 2003). The Mg ## line
is blended with a Mg ### 367.67 � line which has a formation
temperature of 5.9, therefore the image of the slit centered at
368 � is the product of the BP emission in these two tempera�
tures. Table 1 shows the details of these observations. Each NIS
pixel corresponds to 1.68 ## square, although the spatial resolu�
tion of CDS is closer to 6 ## (Pauluhn et al. 1999). A standard
reduction was applied to the images in order to correct for bias,
flat�field, cosmic rays and instrumental e#ects.
The BPs were visually identified in the Mg ###/## window
as compact regions with an enhanced emission over the sur�
rounding corona. However, we note that network elements and
blinkers can both have a coronal response (Gallagher et al.
1998). Leaving aside, for the moment, the possible link that
could exist between both phenomena, we confirmed the BP na�
ture by inspecting images from the Extreme ultraviolet Imaging
Telescope (EIT). In these images, BPs are seen as point�like
or loop�like structures that last for several hours (Zhang et al.
2001). All of them last for the whole CDS sequence, which
varied between 93 and 218 mins.
2.2. EIT
Images from the EIT were used to identify and follow the life
evolution of the BPs. The BP corresponding to the CDS dataset
s6744 can be identified at #13:00 UT on 20 January 1997 in
the 171, 195 and 284 � bandpasses, and remains still visible at
15:09 in the 195 � band. For the s6829 dataset, the BP was a
fuzzy enhanced region at 7:00 UT in the 195 � band that slowly
became brighter and more recognizable by 9:00 UT. Its inten�
sity was still increasing at 13:04 UT. For the s6840 dataset, the
BP appeared some time between 5:00 UT and 7:00 UT (visi�
ble in 171, 195 and 284 � bands). By 9:00 UT it had reached
its maximum brightness in 195 � and was still visible at 13:00
UT, although fainter. The BP in dataset s6697 was visible for
several hours (at least from 8 UT to 14 UT).
2.3. MDI
MDI (Michelson Doppler Imager) high resolution magne�
tograms (0.6 ## /pixel) were also analyzed for the BP observed
on January 16 1997. They cover the whole CDS observational
period with a cadence of one exposure per minute. The magne�
tograms were corrected for di#erential solar rotation and for the
geometrical projection of the line�of�sight magnetic flux (Chae
et al. 2001; Hagenaar 2001). In order to increase the signal to
noise, every five consecutive exposures were averaged result�
ing in a final cadence of #5 minutes.
3. Wavelet analysis
BPs are characterized by a high flux variability in EUV emis�
sion (Habbal & Withbroe 1981; Habbal et al. 1990). Our ob�
jective in this paper is to extend to more cases the periodicity
analysis done by Ugarte�Urra et al. (2004) for two BPs and try
to determine whether oscillations are a common feature in their
brightness evolution and whether they show a coronal response.
We are aware that oscillations in the quiet Sun network have
been found to be intermittent (Banerjee et al. 2001), therefore,
we are interested in the use of a technique that allows the study
of non�stationary signals along the time�series. The wavelet
analysis is a technique that permits the extraction of the rel�
evant signals present in a time�series by convolving the series
with a wavelet function. The result is a two dimensional (time
and period) power spectrum, like that shown in the center panel
of Fig. 1, where the regions with higher power values, and
therefore higher response to the wavelet function, are plotted
in darker gray. The analysis su#ers from edge e#ects due to the
fact that we are dealing with time�series of a finite length. The
region in the power spectrum under this influence is called the
cone of influence (###), which is indicated by the cross�hatched
regions. Periodicities found inside the ### are unreliable and
will not be considered. We have used the software and defini�

I. Ugarte�Urra et al.: CDS wide slit time�series of EUV coronal bright points 3
tions provided by Torrence & Compo (1998) and the wavelet
function chosen for the convolution is the Morlet, which pro�
vides the best compromise between time and period resolution
(De Moortel & Hood 2000). An important part of the analysis
is to determine the relevance of the peaks in the wavelet power,
associated with specific periods and specific moments of time.
A randomization method is used to estimate the significance
levels (O'Shea et al. 2001). The basic assumption of the method
is that if there is no periodicity in the time�series, any order of
the intensity values would be as likely as any other. By com�
paring the power spectrum obtained from the observed series to
the one obtained from each of the permutations in time of the
intensity values, we can estimate the probability p that a certain
peak in the power spectrum would be produced by chance. We
will consider as relevant signals those with a probability (1- p)
greater than 95%. We calculated this probability for the global
wavelet spectrum, an average in time of the power spectrum,
and for the two strongest peaks (called first and second max�
imum) of each time slice. More details about the analysis can
be found in Ugarte�Urra et al. (2004) or O'Shea et al. (2001).
We produced a light curve for each BP and each spectral
window by integrating the emission coming from an area of
3 � 3 or 4 � 4 pixels, depending on the exposure time, which
increases the signal to noise without loss of spatial informa�
tion due to the broad point spread function of the instrument
(Pauluhn et al. 1999). While for the case of the BP located
at equatorial latitudes the region of integration was corrected
for solar rotation, this correction was not necessary for the
BPs located at the poles. The general trend of intensity in�
crease/decrease of the light curves were subtracted from each
BP with an appropriate running average in order to analyze the
variations that it experiences with respect to the trend.
We describe now the results obtained for each of the BPs.
The sequences with a 60 seconds cadence designed for coronal
hole studies showed a saturation of the O # 629 � BP counter�
part, so the study could only be carried out on He # 584 � and
Mg ###/## 368 �. Special attention is given to the sequence of 30
seconds exposure where O # 629 � can be safely analyzed and
where a comparison with high resolution MDI magnetograms
is available.
3.1. January 20 dataset
This is one of the sequences run at polar coordinates. The BP
appears in the He # 584 � window as a bright network ele�
ment not very di#erent from the surrounding area. Its emission
ranges between values comparable to those of other network el�
ements and values 35% higher. In the Mg ###/## window, the BP
is clearly identified as a 20 ## round structure with an intensity
between 1.5 -- 2 times the emission of the surrounding corona.
The flux lightcurve was obtained from integrating the emission
of the BP central part (3�3 pixels). The wavelet analysis of the
time�series shows an oscillatory pattern with a period of #822
seconds at the beginning of the time�series in He # 584 �. It ex�
tends over 3 -- 4 complete cycles as can be seen in the sinusoidal
dashed line plotted over the detrended time�series (Fig. 1, top
panel). It is preceded and followed by two brightenings of a
Fig. 1. Wavelet analysis results for the He # 584 � time series of the
BP observed on January 20, 1997. Top panel: time variation of the
number of counts after detrending; a sine function with a period of
822 seconds, corresponding to the peak in the wavelet analysis, is
over�plotted. Middle panels: wavelet power spectrum with the cone
of influence over�plotted (cross�hatched region) and global wavelet.
The dashed line represents the maximum period outside the cone of
influence. The periods corresponding to the first two maximums in
the global wavelet are indicated below the graph, together with the
associated level of probability in brackets. Bottom panel: variation in
time of the probability associated with the two strongest peaks in each
time slice. The solid line is associated to the first maximum location
(white dots on the power spectrum) and the dotted line to the second
maximum (black dots). The 95% confidence level is given by the dot�
dashed line.
longer temporal response, of around 1000 seconds. The ampli�
tude of the intensity variations is of the order of 5 -- 20%. The
wavelet analysis shows that no oscillations are present above
the probability threshold in the Mg ###/## time�series. The am�
plitude of the intensity variations (2 -- 6%) are much smaller
than those in He #
3.2. January 30 dataset
The BP, seen also in Mg ###/## as a non�structured fuzzy bright
compact region over the background coronal hole, lies over a
He # network element very close and sometimes indistinguish�
able from the neighbouring network elements. These neigh�
bouring elements also play an active role in the evolution of

4 I. Ugarte�Urra et al.: CDS wide slit time�series of EUV coronal bright points
Fig. 2. He # (solid line) and Mg ###/## (dashed line) light curves of the BP observed on January 30. Two di#erent locations are considered: BP
center and an o#set of 10 ## north of the BP center. Left panels: comparison of the temporal behaviour of the two spectral lines at the indicated
location; the flux has been normalized to the maximum value of each curve to make the comparison easier. Right panels: comparison of the
two locations for the same spectral line. The vertical dotted lines indicate the time when the events peaked in the o#�center box.
the BP. Again, as in the previous dataset, sole inspection of the
He # 584 � images does not tell us which of the network el�
ements are associated with a BP. The flux in the appropriate
network element is comparable to the flux of other elements in
many of the snapshots, although it can double their brightness
at di#erent times. To produce the light curve we selected a 3�3
pixel box in the central region of the Mg ###/## image and its
corresponding counterpart in the cooler line. The He # emission
shows a high variability again in this case. We find intensity
changes with amplitudes of the order of 4�10%, slightly lower
than the previous dataset, and between 3�6% for Mg ###/##.
The wavelet analysis reveals the presence of a band in the
power spectrum corresponding to a period of 634 -- 754 seconds
for as long as 3 cycles. The probability ranges between 80 and
96%, perhaps indicative of a real oscillatory behavior, although
it is below our initial probability requirements. Mg ###/## does
not show an oscillatory pattern in the power spectrum. It expe�
riences, however, changes in the flux similar to the ones shown
by He #. The top left panel of Fig. 2 shows the detrended light
curve for He # (solid line) and Mg ###/## (dashed line) normal�
ized to their maximum values in order to make the comparison
easier. Apart from the general trend, removed here, the figure
shows common brightenings suggesting a connection between
events at both temperatures.
There are, as well, events which are restricted to only one
of the observed spectral lines, like the one which takes place
in Mg ###/## 40 minutes after the start. This is an interesting
event as seen in the wide slit movie. It is originated in a nearby
network element that brightens and establishes an emission
link, probably loops, with the element below the coronal BP.
Unfortunately, the available resolution does not allow us to es�
tablish this more firmly. The bottom left panel of Fig. 2 shows
the light curve obtained from a box of integration located over
the bridge between the network elements, north of the BP cen�
ter. In He #, this box is located 5 ## eastwards, as the peaks of
emission in He # and Mg ###/## are slightly o#set. The right
panels of Fig. 2 show the similar information, but this time we
show in each graph the light curves of the two locations for the
same spectral line. The vertical dotted lines indicate the time
when the events peaked in the o#�center box. The large rise in
the He # flux (34% from trough to crest) starting at around 27
minutes after the beginning of the observations in the neigh�
bouring network becomes visible in Mg ###/## over the di#use
outskirts of the BP, 10�13 ## o#�center, for just the duration of
the event. The right panels show how the brightening, origi�
nated o#�center in He #, produces a general rising of the coronal
emission in the BP (center and o#set), followed by a steeper in�
crease, reaching 50%, at the top of that bridge and peaking 281
seconds after the initial He # brightening. There is no such re�
sponse in the He # emission after the first event. It soon regains
its original flux, except for a little spike 143 seconds after the
coronal peak. The final spike in the decay of the He # flux could

I. Ugarte�Urra et al.: CDS wide slit time�series of EUV coronal bright points 5
be a response to the coronal emission. The formation of He # is
still under discussion and one of the possible mechanisms is a
reprocessing of EUV coronal radiation (Andretta et al. 2003).
3.3. January 31 dataset
A third BP was observed 60 ## north of the January 30 BP's
location, inside a coronal hole. It appears in Mg ###/## 368 � as
a 20 ## compact bright region with a peak emission #2 times the
surrounding corona. In He # 584 � we find network elements
which are 1.8 -- 2.7 times brighter than the elements with no
coronal counterpart.
The length of the time�series is only 93 minutes, shorter
than previous datasets. The box of integration (3 � 3 pixels)
was located in the center of the emission. The light curves ob�
tained for both spectral lines are shown in the top two panels
of Fig. 3, with the corresponding error bars obtained from the
quadratic sum of the errors of individual pixels. The errors in
individual pixels were obtained from the standard formulae for
NIS (Thompson 2000). The bottom panel shows a comparison
of the variations experienced by the BP in both lines after de�
trending and normalizing. These variations are very similar in
both lines suggesting a connection between the events. Some
of these events peak at the same time in both lines, but in other
cases, like the one which peaks in Mg ###/## 25 minutes after
the start or the one peaking at minute 48, are seen in He # with
some time delay. The delay is three datapoints (204 seconds) in
the first case and two (143 seconds) in the second. Brightenings
with no clear counterpart in the other line are present as well.
The variety of events suggests that the evolution of the flux in
the BP (at this resolution) has a complex picture with di#erent
processes producing di#erent observational patterns. The am�
plitudes of the variations are of the order of 5�18% for He # and
3�10% for Mg ###/##.
The wavelet analysis results of the two time�series are
shown in Fig. 4. The power spectrum (central panel) of both
time�series has a high power band at around 700 seconds. In
He #, the peak is at a period of 688 seconds, while in Mg ###/## it
is at 750 seconds, with a 99�100% confidence if we attend to the
result of the wavelet analysis of the global wavelet spectrum.
The time dependent plot of the probability shows, however,
that certain parts of that band are below the 95% confidence
level. In He # the probability is over 95% from minute 41 in the
time�series. In the case of the coronal line, some points lie be�
low the threshold, although most of the time around 95% con�
fidence. The time�series can be reconstructed using its wavelet
power spectrum, following the indications given by Torrence
& Compo (1998). In this reconstruction, which basically sums
the contributions of the di#erent periods at the di#erent times,
we can disregard the higher periods thus revealing signals with
lower periods. If we filter the periods greater than 1000 s, the
confidence level rises above the threshold. Sinusoidal functions
of periods 688 seconds and 719 seconds have been plotted over
the detrended light curves (top panels of Fig. 4), where 719 sec�
onds is the middle value between the 688 and 750 values that
dominate the Mg ###/## power spectrum band. It stands out in
the comparison that there is a regularity in the appearance of
Fig. 3. He # 584 � and Mg ###/## 368 � flux changes observed in a
BP on January 31, 1997. Top two panels: flux variations for each of
the lines. Bottom panel: both light curves have been detrended and
normalized to its maximum value (He #, solid line; Mg ###/##, dashed).
the larger brightenings suggesting that a periodic phenomenon
is taking place.
3.4. January 16 dataset
This bright point, located at Sun center, is visible for several
hours in EIT (at least from 8 UT to 14 UT) and it appears as
a point structure of 3 � 3 pixels in the 195 � passband. Fig. 5
shows the appearance of the BP in the CDS wide slit images.
In Mg ###/## 368 �, we see it as a di#use elliptical cloud with
enhanced emission 2.1 -- 2.5 times over the background. In
O # 629 � we can identify between 2 and 4 network elements
(depending on the instance) at the junction of three network
cells. The network elements present high variable emission and
their increase/decrease is not always simultaneous. He # 584 �
images are very similar to the O # images with the network el�
ements clearly delimiting the edges of the cells. A MDI high
resolution magnetogram is also displayed in the figure and the
contours of the Mg ###/## and He # emission are over�plotted for
comparison. A bipolar region associated with the coronal emis�
sion and dominated by the negative fragments is present. As we
see in the O # and He � images, there are other bright network
elements in the field of view, but they do not have a coronal
counterpart in Mg ###/##. It is easy to notice that the coronal
emission is associated with the presence of strong opposite po�
larities. The positive fragments associated with the BP and the

6 I. Ugarte�Urra et al.: CDS wide slit time�series of EUV coronal bright points
Fig. 4. Wavelet analysis results of the BP observed on January 31, 1997 at 08:02 UT. The left panel corresponds to the He # 584.34 � time�series
and the right panel to Mg ###/## 368 �. The dashed lines plotted over the two light curves at the top panels account for sinusoidal functions with
periods of 688 and 719 seconds respectively, corresponding to the peaks of the power spectrum.
Fig. 5. CDS wide slit images in Mg ###/## 368 �, O # 629.73 � and He # 584.34 �, of the coronal bright point observed on January 16, 1997,
plus a MDI high resolution magnetogram. In the images, black regions are the brightest sources and white the faintest. In the magnetogram,
positive polarities are in white and negative in black. All images were taken at 10:01 UT. Solid (Mg ###/##) and dotted (He #) contours are used
as a reference for comparison.
surrounding bright area have peaks in the magnetic flux den�
sity between 97 and 187 Mx cm -2 , while the positive fragment
seen at an o#set of (15 ## , 65 ## ) (associated with the bright O #
network element) does not reach 40 Mx cm -2 .
Due to a higher cadence than previous sequences (31 sec�
onds), we decided to use a larger box of integration, 4�4 pixels,
to increase the signal to noise. This box was located at the cen�
ter of the peak emission in Mg ###/##. That same position in the
He # image corresponds to the left network element below the

I. Ugarte�Urra et al.: CDS wide slit time�series of EUV coronal bright points 7
BP in Fig. 5 and the same occurs for the O # image, where
in this instance the network element is smaller than the neigh�
bouring elements at the network junction. A first glance to the
time�series of each of the spectral lines indicates that the gen�
eral BP emission increases monotonically all along the obser�
vation period and that the variability is higher as the emission
increases. This general trend has been removed, using an ap�
propriate running average for each of the light curves, in order
to study the variations with respect to the trend. The amplitude
of these variations are of the order of 3 -- 11% in the case of the
He � 584 � emission, 10 -- 20% in the case of O # 629 � with
events reaching up to 45%, and 3 -- 8% for Mg ###/## 368 �.
The wavelet analysis of the He # light curve (top left panel
of Fig. 6) reveals a diagonal band in the power spectrum, 97
minutes long over the 95% confidence level, decreasing from
a period of 896 seconds to a period of 257 seconds in the sec�
ond half of the time series, but it does not last for more than
two cycles in any of the periods. There is, however, a peak in
the power spectrum at around 1123 seconds that lasts for 63
minutes, with the first two cycles well outside the cone of in�
fluence. Simultaneous to that we can identify a band with short
periods (306 -- 236 seconds), two and four cycles long each.
We checked that the running average is not introducing false
frequencies by removing the trend and longer periods in a dif�
ferent way. We applied the wavelet analysis to the un�detrended
light curve and reconstructed it using the power spectrum with
only the shorter periods taken into consideration. We find simi�
lar results, which give us confidence that the peaks in the power
spectrum have a solar origin.
The analysis of the O # light curve shows a power spectrum
with certain similarities to He #. A band with a confidence level
over the 95%, with periods ranging between 944 and 794 sec�
onds, appears at the same time as the diagonal band in He #,
suggesting a connection between the events. In this case, the
peak in the power spectrum at 944 seconds can host around 3
oscillatory cycles. The same occurs for the peak at 364 seconds
simultaneous to the one found for He # at 306 seconds.
In the case of Mg ###/## we can clearly identify, at the bot�
tom panel of Fig. 6, the band peaking at 1123 seconds that we
found in the He # data. It is long enough to host 4 complete
cycles, 3 of them outside the cone of influence. The shorter
periods, 216 seconds, barely seen in the figure can also be de�
tected in this case using the reconstruction technique, but only
marginally and for less than two cycles over the 95% confi�
dence level. Somehow, the fact that they are present at the same
location as in the cooler lines suggests that they could be real
and therefore could be part of the coronal evolution of the BP.
The lower signal to noise of this spectral line and the mix�
ture of di#erent structures with the current spatial resolution
are all possible causes that this oscillation is partially hidden in
Mg ###/##. Finally there is another period peaking at 612 sec�
onds, close to the end of the temporal sequence.
4. Magnetic field evolution
It was noted in the introductory section the importance of the
evolution of the magnetic fragments and the associated mag�
netic flux in relation to the BP appearance (see also Ugarte�
Urra et al. 2004). We feel that the present discussion will ben�
efit if we discuss the EUV brightness fluctuations in the con�
text of the changes experienced by the magnetic field. We do
it, therefore, for one of the BPs for which we have MDI high
resolution magnetograms.
The MDI high resolution magnetograms (0.6 ## /pixel) are
available for the BP observed on January 16. The observations
cover the whole CDS observing time with a cadence of one
magnetogram every minute. In order to improve the signal to
noise we produced new magnetograms by averaging five con�
secutive images, which left us a final cadence of five minutes.
The noise in the dataset can be obtained from the core of the
distribution function of flux densities, # (Hagenaar 2001). This
core is well described by a gaussian fit, which reveals in our
case a level of noise of # = 11.1 Mx cm -2 .
The magnetic configuration of the BP area is visible in
Fig. 5 (right panel). The left image in Fig. 7 shows a close�up of
the region centered approximately at the o#set coordinates of
(45 ## ,120 ## ) in Fig. 5, with the Mg ###/## and O # isocontours to
make the comparison easier. The alignment between CDS and
MDI was done using the header coordinates. The uncertainty
is not larger than 5 ## . There is a main negative polarity made of
several fragments that dominates and three satellite positive po�
larities around it, one of them stronger than the others, see A in
Fig. 7. The corresponding O # and He # images show three net�
work elements associated with the three points where opposite
polarities are confronted. Attending to the evolution of these
satellite polarities we find in the images from 2 to 4 network
elements associated with the coronal emission, which mainly
lies over the left interaction point, A. The movie, made with
the collection of images, shows an approaching movement of
the two polarities, A and B, at this location, as well as a disap�
pearance or cancellation of the lower positive fragment.
We wanted to inspect this issue in order to see the influ�
ence of the magnetic evolution on the EUV emission, which
becomes more variable and with oscillations more evident in
the last part of the observing period. Following the method de�
scribed by Parnell (2002), we calculated the survival functions
of the magnetic flux density distribution and the gaussian fit
representing the noise. The survival function gives the proba�
bility for a certain variate to be greater than a certain value. In
this case the variate is the magnetic flux density. From the ra�
tio of the two survival functions we obtained the proportion of
pixels associated with noise for the di#erent flux density val�
ues. At 21 Mx cm -2 less than 1% of the pixels are associated
with noise. We adopt this as our working value. We plot in the
center & right panels of Fig. 7 the BP magnetic configuration
considering only those pixels with |#| # 21 Mx cm -2 .
We determined then the distance between A and B all along
the time�series in two ways. First of all, we calculated the dis�
tance between the two closest pixels with a magnetic flux that
we know for certain it is not associated with noise, i.e. those
pixels with |#| # 21 Mx cm -2 . Secondly, we calculated the
distance between the centroids of both polarities at di#erent
instances of the time�series. Both results are shown in Fig. 8,
together with the temporal changes in the number of counts
(photon�events) in the transition region line. Asterisks repre�
sent centroid distances, while the single solid line accounts for

8 I. Ugarte�Urra et al.: CDS wide slit time�series of EUV coronal bright points
Fig. 6. Wavelet analysis results of the BP observed on January 16, 1997 at 08:02 UT. Top left panel corresponds to the He # 584.34 � time�series,
top right to the O # 629.73 � ones and bottom to the Mg ###/## 368 � blend.

I. Ugarte�Urra et al.: CDS wide slit time�series of EUV coronal bright points 9
Fig. 7. Section of a five minute averaged MDI high resolution magnetogram showing the photospheric magnetic configuration associated to the
Mg ###/## BP observed on January 16, 1997. On the left panel, the solid and dotted contours locate the Mg ###/## and O # emission, respectively.
The section is a close�up of the region centered at approx. (45 ## ,120 ## ) in Fig. 5. White accounts for positive magnetic flux density, #, and black
for negative. In the center and right panels we plot only those pixels with |#| # 21 Mx cm -2 . The right panel shows the maximum change in
orientation (#20 # ) of the straight line joining the centroid positions of the two opposite polarities for two di#erent images. White/black contours
correspond in this case to the negative/positive polarity.
the distance between closest pixels. The approaching trend is
clear in both curves. The latter shows two di#erent sections
of decreasing distance joined by a steep increase at around 80
minutes after the start of the observing period. This increase
is due to the cancellation of the positive fragments that we see
detached from the main positive polarity in the left and center
panels of Fig. 7. After the disappearance of these fragments,
the closest pixel belongs to the largest positive fragment in
the figure, which is further apart. The separation between cen�
troids shows a similar profile due to the same phenomenon.
This is the distance between the strongest flux concentrations
since we only consider pixels with a flux over 21 Mx cm -2 .
The real distance between opposite polarity fragments would
probably be less than that. A comparison with the CDS emis�
sion reveals that the approach of the polarities is followed by
an increase in the transition region flux. Interestingly, the flux
increase reaches a 30 minute `plateau' when the distance be�
tween polarities experiences the steep increase, and as soon as
it starts to decrease again, the flux rises. Another characteristic
is that the amplitude of the O # brightenings becomes larger
with time. The fact that the closest distance between polarities
is reached just before the `plateau' phase and the amplitudes
are not the largest at this moment, suggests that the proximity
of the centroid, or in other terms the bulk of the positive mag�
netic flux, is more important in terms of energy release. From
these plots we can infer the approach velocity. From a linear
fit (Fig. 8) to the second section of the curve that gives the
distance between closest pixels (i.e. when the small fragments
have already been canceled), we find a velocity of approach of
0.27 � 0.02 km s -1 . In the case of the centroids, we obtain a
value of 0.15 � 0.06 km s -1 using all the points. If we just fit
the points after the small fragments have canceled the result is
0.26�0.09 km s -1 . These results are inside the range of typical
values (0.1 -- 0.3 km s -1 ) quoted by Priest et al. (1994).
Fig. 8. Comparison between the temporal variation in the transition
region emission of O # 629.73 � (top) and the distance between oppo�
site polarities A and B (bottom). The asterisks account for the distance
between centroids and the single solid line represents the distance be�
tween the closest pixels. The dashed line is a linear fit to the second
distance decrease section used to determine the approaching velocity
of 0.27 km s -1 .
Once a reasonable threshold has been chosen to delimit the
polarities, what we want to do is to compare the evolution of
the emission in the di#erent spectral lines with the evolution of
the integrated magnetic flux of the polarities. This comparison
is shown in Fig. 9. The top three panels represent the BP's de�
trended flux variations in Mg ###/##, O # and He #. The bottom
two panels show the evolution of the positive magnetic flux in
A, # + , and the negative magnetic flux in B, # - . The two values
are obtained from the integration for each image of those pixels
with |#| # 21 Mx cm -2 in the region containing A and B. From
this comparison the first thing that we notice is that the mono�
tonic increase of the EUV emission, visible in Fig. 8 for O #

10 I. Ugarte�Urra et al.: CDS wide slit time�series of EUV coronal bright points
Fig. 9. Comparison between the EUV variability of the BP detrended
emission, as obtained from the CDS wide slit images, and the mag�
netic flux of the two opposite polarities. Shaded areas point out loca�
tions where we see a decrease in the positive flux preceding a bright�
ening in the EUV lines. The local minimum is given by a dashed line.
and similar for the other two lines, is associated in Fig. 9 with
a decrease in the positive magnetic flux, which is an order of
magnitude larger than the negative one. On smaller timescales,
the plot suggests that sudden decreases in the positive magnetic
flux (shaded areas) seem to take place just before (or during the
rising phase) a brightening is observed in the transition region
or the corona, and sometimes both. The location of the local
minimum in the magnetic flux is given by a dashed line. These
results could suggest that magnetic flux cancellation are associ�
ated to the stronger brightenings and give support to the picture
that proposes reconnection as the main process producing the
appearance of BPs and its characteristic variability (Priest et al.
1994; Longcope 1998). However, we feel that more observa�
tions are needed to confirm a definite correspondence between
decreases in the magnetic flux of this order and the brighten�
ings. The negative flux does not show as good correlation as
the positive one, but this could be explained by the fact that
fragment B is weaker in flux and also belongs to a more com�
plex structure, where the surrounding positive fragments (see
Fig. 7) could play an active role in the evolution.
5. Discussion
It is clear from the present work and that of Madjarska et al.
(2003) and Ugarte�Urra et al. (2004) that BPs have intermittent
(few cycles), but clearly defined periods of oscillation ranging
from around 400 s up to 1100 s (0.9 -- 2.5 mHz) in upper chro�
mospheric, transition region and sometimes coronal plasma. In
the case of one BP, periods as short as 236 s (4.2 mHz) were
also detected. Apart from the clear oscillatory cases, we have
found several other cases with similar periods where the prob�
ability estimates do not reach the required 95%, but they are
very close. We believe that some of these oscillations are real,
in particular those detected simultaneously in several spectral
lines, but are hidden in the low spatial resolution of the inte�
grated emission.
Oscillations in the quiet Sun chromosphere have been
known for a long time and is a well documented phenomenon.
Two regions showing two di#erent characteristic frequency
ranges are normally di#erentiated. On one side the #5.5 mHz
(3 minutes) oscillations seen in the chromospheric internet�
work (Dame et al. 1984; Carlsson et al. 1997; Curdt & Heinzel
1998; Krijger et al. 2001; Wilhelm & Kalkofen 2003, and ref�
erences therein); on the other, the network lanes and the net�
work bright points (NBPs), which show a wider range of fre�
quencies (0.8 -- 3.3 mHz), 5 -- 20 minutes periods according
to Lites et al. (1993) (but see also Dame et al. 1984; Curdt &
Heinzel 1998; Doyle et al. 1999; Banerjee et al. 2001; McAteer
et al. 2002, 2003). The four BPs analyzed here lie on the net�
work lanes and their O # and He # counterparts are network
elements generally indistinguishable from other elements with
no coronal counterpart. The first thing that we notice is that the
frequencies found for these BPs lie in the frequency range of
the chromospheric network oscillations. One characteristic of
NBPs is that they are associated with photospheric magnetic
flux elements (Cauzzi et al. 2000). This can be clearly seen in
Fig. 2 of McAteer et al. (2003), which shows the appearance
of several NBPs at di#erent temperatures in the solar atmo�
sphere, from the chromosphere, through the transition region
to the corona, co�aligned with a magnetogram. It is worth not�
ing that McAteer et al. (2003) NBP 5 shows the characteristic
size and enhanced coronal emission overlying two opposite po�
larity magnetic fragments of a coronal bright point (with a peak
emission around 2 times higher than the surrounding corona, as
checked from the original image). In the transition region im�
age we see two bright features at the location where the foot�
points of the coronal loop would be, connected by what looks
like a cooler loop structure; these footpoints can be traced into
the chromosphere and the two opposite polarities in the mag�
netogram. Only the brightest one was the subject of the NBP
study by the authors. Fig. 2 in Cauzzi et al. (2000) shows a
similar configuration with two opposite magnetic fragments of
reasonable strength for a coronal bright point, separated by less
than 10 ## , below the chromospheric emission of two NBPs. No
coronal image is available in this case. More images can be
compared in Nindos & Zirin (1998).
This leads us to the conclusion that a fraction of the NBPs
(and the associated magnetic field) are most likely the foot�
points of the tiny loops which comprise coronal bright points
(Sheeley & Golub 1979); the cool footpoints of loops which
rise to greater temperatures (Cook et al. 1983). This would
explain why we see the same periodicities in both features.
Recently, McAteer et al. (2004), analyzing quiet Sun network

I. Ugarte�Urra et al.: CDS wide slit time�series of EUV coronal bright points 11
and internetwork TRACE oscillations, finding that the network
has a peak occurrence rate periodicity of 231 -- 346 s, with a
significant tail of higher period oscillations (as long as #1000
seconds). The peak decreases in occurrence rate in the high�
est formation temperature passband (# 1 � 10 5 K) suggesting
that the waves may dissipate, shock, move away from the net�
work or change frequency. Gouttebroze et al. (1999) did not
find signature of 2.5 -- 7 mHz (period of 143 -- 400 s) oscilla�
tions in lines with a temperature formation # 5 � 10 4 K. If we
bear in mind that BP's studies (Ugarte�Urra et al. 2004, and the
present one) show oscillations with typical periods of around
400 -- 1100 s, with only one case showing shorter periods, all
this seems to be consistent with a picture where only the longer
periods observed in the chromospheric NBP, assumed here as
the footpoints of hotter loops, make it through to the transition
region and corona. However, more observations with simulta�
neous chromospheric and coronal counterparts of the BP will
be needed to confirm that.
Di#erent interpretations are given for these long frequency
oscillations in the chromospheric network, which show no sign
of distinctive peaks in the power spectrum (Krijger et al. 2001);
namely, in terms of formation of standing waves by inter�
nal gravity waves (Deubner & Fleck 1990; Lou 1995) or up�
ward propagating MHD waves excited by granular motions
(Kalkofen 1997; Hasan & Kalkofen 1999), although more dis�
cussion is still needed (Musielak & Ulmschneider 2003).
The modulation seen in Ugarte�Urra et al. (2004) and
our results is more pronounced at lower layers of the atmo�
sphere: in the emission lines associated with the temperatures
30 000 K, 250000 K and 1 000 000 K (the corresponding sound
speeds are 27 km s -1 , 76 km s -1 and 152 km s -1 , respectively).
Although we should note that the formation of He # line is
still under discussion (Andretta et al. 2003). We see periods
of around 600 s in O # and He # overlapping. The January 16
BP is the clearest case with certain periods shared by two of
the three lines in the sample (He #, O # and Mg ###/##). How are
these oscillations related to the heating of BPs and how do they
merge with the idea that reconnection plays a dominant role in
the evolution of BPs ?
One idea is that the observed intensity oscillations are asso�
ciated with compressible waves, e.g. the acoustic mode which
has been identified in coronal structures as both a propagating
wave (e.g. Nakariakov et al. 2000) and standing waves (e.g.
Ofman & Wang 2002) or the fast mode. An alternative inter�
pretation, as a kink mode observed through the modulation of
the observed width of the structure guiding the wave (Cooper
et al. 2003), may be excluded from consideration as it is not
likely that the kink mode, practically incompressible, has a no�
ticeable amplitude in the chromosphere and the transition re�
gion.
The mode oscillates between the footpoints of the magnetic
field line. Along the field line, the sound speed changes, but for
the global mode (or any other lower harmonics) it is the aver�
age sound speed which determines the period, not the local one,
hence this would mean that the oscillations have a collective na�
ture, e.g., similar to the global acoustic mode of a coronal loop
(Ofman & Wang 2002) or their second harmonics (Nakariakov
et al. 2004). The common feature of these oscillations is the
positioning of nodes of the velocity perturbations and maxima
of density perturbations at the lower layers of the atmosphere.
In particular, in the structure of the global acoustic mode is
V z (s, t) # cos # #C s
L t # cos # #
L s # , (1)
#(s, t) # sin # #C s
L t # sin # #
L s # , (2)
where C s is a speed of sound, A the wave amplitude, L the loop
length, and s is a distance along the loop with the zero at the
loop top. According to Eq. (1) and (2), the oscillation period
is given by the expression 2L/C s . As the magnetic field in a
coronal bright point is closed, it may be considered as a small
loop with the oscillation modes described by these equations.
On the other�hand, if the magnetic topology of the bright point
is more complicated than the loop geometry, its ``building el�
ements'' may be modelled as individual loops. In addition, as
the acoustic wave considered here propagated strictly along the
magnetic field, it does not feel the magnetic field complexity,
simply following the local line.
Suppose the length of the loop forming an oscillating bright
point is 30 Mm (corresponding to a 10 Mm radius), the global
acoustic mode with periods of about 600 s would require the
sound speed to be 100 km s -1 . This value of the sound speed
is certainly reasonable. The period of the second harmonics of
this loop is 300 s, also in the observed range. Di#erent periods
can be easily explained by di#erent loop lengths and by slightly
di#erent profiles of the sound speed along the loop (growing
from the chromospheric 30 km s -1 to coronal 150 km s -1 ).
Also, a lower lying loop would in general have lower sound
speed.
The domination of the oscillations in chromospheric and
TR emission lines, is consistent with this interpretation too: ac�
cording to Eq. (2) the density perturbations are stronger near
the loop footpoints. At the coronal heights, the oscillations are
not so pronounced because the density oscillations have a node
at the loop apex.
Another possible interpretation of these oscillations is that
they are connected with vertically propagating fast magneto�
acoustic waves trapped between the dense chromosphere and
the upper boundary of the bright point magnetic structure,
where the Alfv�en speed experiences a sharp increase. The
wavelength of such waves is about double the radius of the
closed magnetic field lines, and the characteristic speed is sev�
eral sound speeds (the lowest possible value is one and a half
sound speed, in the case of # = 1). Taking the values from
the previous example, we obtain the longest characteristic pe�
riod of about 130 s (for the radius 10 Mm and the sound speed
100 km s -1 ). This value is less consistent with the observational
findings than the periods given by the acoustic model.
This simple model then suggests that the bright point os�
cillations could be associated with the global acoustic modes
of closed magnetic structures associated with the bright point.
However, this does not address the important question of how
these waves are generated. They could be related to the super�
granular motions or may be excited by the energy deposition
from the reconnection process. This second scenario would be

12 I. Ugarte�Urra et al.: CDS wide slit time�series of EUV coronal bright points
supported by the fact that, in the case of the BP observed on
January 16, we detect sudden decreases in the magnetic flux at
the same time as the oscillations in the EUV flux take place.
If reconnection is important in BPs (as commonly assumed)
it has to be a slow process as the observed Doppler shift of tran�
sition region lines is at most 10 km s -1 red�shifted (Madjarska
et al. 2003; Xia et al. 2003; Popescu et al. 2004) compared to
a red�shift of 10 -- 25 km s -1 for blinkers (Madjarska & Doyle
2003) and �100 to 200 km s -1 (Madjarska & Doyle 2002) for
bi�directional jets. It is more in the order of the quiet Sun net�
work (Teriaca et al. 1999) which has a red�shift of 8 -- 10 km/s
at these temperatures.
In this respect we may have a problem. As shown in �4,
the converging speed of the two polarities was measured as
0.27 km s -1 which is an excellent agreement with a study by
Chae et al. (2002a) who measured the approach for two di#er�
ent observations of converging polarities. In one instance the
speed was 0.35 km s -1 , while the other was 0.27 km s -1 . In
follow�up work, Chae et al. (2002b) considered two reconnec�
tion models, the Sweet�Parker model and the faster reconnec�
tion Petschek model. They showed that the Sweet�Parker model
does not have inflow speeds greater than 0.08 km s -1 while
in the Petschek model, the inflow speed equals the observed
speed just below the temperature minimum. Higher up in the
atmosphere, e.g. the chromosphere and transition region, the
speeds are in the range 20 to 100 km s -1 . Thus, although the
Petschek looks applicable around the temperature minimum it
is clearly not correct in the upper atmosphere as these velocities
are much greater than the maximum observed Doppler shifts.
In fact, these rates would be more applicable to bi�directional
jets. One possible way of resolving the Sweet�Parker rates in
the lower atmosphere is to change some of the assumptions
in the model. For example, taking a very small magnetic fill�
ing factor (#0.05) and a small electric conductivity of 10 10 s -1
yields an inflow speed of 0.27 km s -1 which would be consis�
tent with the observations (Chae et al. 2002a,b). However, even
the Sweet�Parker model gives large velocities in the upper at�
mosphere. Thus this would rule out both of these as possible
reconnection models for BPs.
Binary reconnection, an interaction of pairs of unbalanced
sources has been proposed (Priest et al. 2003) as a fundamen�
tal mechanism to produce coronal heating. One aspect of this
mechanism is that the relative motions of the sources are likely
to build�up twisted force�free structures that relax by turbulent
reconnection resulting in heating. We have tested this possible
scenario against the BP observed on January 16. The sources
show a relative motion of approach, as described earlier, but
also of rotation (right panel of Fig. 7). We have measured the
angle that the line joining the two centroids makes with the
S olar Y = 0 (Fig. 10). The figure shows how one source ro�
tates with respect to the other reaching a maximum of 41 #
and then returning to the original configuration. The error bars
come from the uncertainty in the determination of the centroid,
assumed as 1 pixel. The energy injected by this rotation can
be determined using #E = F 2 ## 2 /(24# 2 �L e f f ) (Priest et al.
2003) with F the magnetic flux of the weakest polarity, ## the
angle change, � the magnetic permeability and L e f f the e#ec�
tive length of the magnetic field lines. The maximum change
Fig. 10. Variation in time of the angle between the line joining the
centroids of the two polarities of the January 16 BP and S olar Y = 0.
in orientation is ## #20 # and takes 122 minutes. Choosing
L e f f as 10.2 Mm (15 MDI pixels for the footpoint separation
and semicircular loop shape) and a range of magnetic flux val�
ues ranging from 7.8 � 10 17 to 7.2 � 10 18 Mx (that depend on
the threshold of integration) for the weakest polarity, we ob�
tain an energy injection per second of #10 18 -- 10 20 ergs s -1 ,
much smaller than the typical values (10 23 -- 10 24 ergs s -1 )
of energy loss in BPs (Habbal & Withbroe 1981; Priest et al.
1994; Longcope et al. 2001). A magnetic flux of 1.4 � 10 20 Mx
would be required under these conditions, an order of magni�
tude larger than the positive polarity flux. The uncertainty in
the di#erent values can not explain the disagreement, therefore
we conclude that the heating of this BP can not be explained
by a helicity injection process.
6. Conclusions
We have studied CDS wide slit movies of four coronal
bright points (BPs) in three spectral lines: He # 584.34 �,
O # 629.73 � and Mg ###/## 368 �. The BPs can be clearly
identified in the Mg ###/## images where they have a peak emis�
sion between 1.8 -- 2.7 the flux of the surrounding corona. This
identification is not straightforward for the cooler lines, where
the emission and appearance can be similar to that of other net�
work elements. The BPs show flux fluctuations with an ampli�
tude generally ranging 4 -- 20% in He # 584.34 �, 10 -- 20% in
O # 629.73 � and 2 -- 10% in Mg ###/## 368 �, although there
are brighter events where the percentage could reach the 50%
increase.
The wavelet analysis of the time�series confirms the oscil�
lating nature of the BP's emission in EUV lines (Ugarte�Urra
et al. 2004). We find wavetrains with periods ranging between
600 -- 1100 seconds in all three lines. We find as well, in one of
the BPs, periods as short as 236 seconds. We notice that these
periods lie in the period range of the chromospheric oscillations
and we conclude that a fraction of the network bright points that
are observed in the chromospheric network are most likely the
cool footpoints of the loops that coronal BPs are made of.

I. Ugarte�Urra et al.: CDS wide slit time�series of EUV coronal bright points 13
In the case of one BP, for which MDI high resolution mag�
netograms were available, we have found that the increase
in the EUV flux and its variability takes place as the mag�
netic polarities approach and the dominant (positive) flux de�
creases. Sudden magnetic flux decreases are observed at the
same time as the oscillations in the EUV emission become sig�
nificant. This would support a scenario where the waves, in�
terpreted here in terms of global acoustic modes, are driven
by magnetic reconnection events. However, Sweet�Parker and
Petschek models do not seem to explain the typical small
Doppler shift values characteristic of BPs. We believe that a
new model is needed in order to explain these Doppler shift
values in the context of magnetic reconnection, where waves
play also a very important role.
Acknowledgements. Research at Armagh Observatory is grant�aided
by the N. Ireland Dept. of Culture, Arts and Leisure. This work was
supported by PPARC grant PPA/V/S/1999/00668. CDS, EIT and MDI
are instruments onboard SoHO which is a project of international co�
operation between ESA and NASA. We would like to thank the ref�
eree for useful comments and suggestions that helped us improve the
presentation of the paper.
References
Andretta, V., Del Zanna, G., & Jordan, S. D. 2003, A&A, 400,
737
Aschwanden, M. J. 2003, in NATO Advanced Research
Workshops: Turbulence, Waves and Instabilities in the Solar
Plasma, 215--237
Banerjee, D., O'Shea, E., Doyle, J. G., & Goossens, M. 2001,
A&A, 371, 1137
Banerjee, D., O'Shea, E., Goossens, M., Doyle, J. G., & Poedts,
S. 2002, A&A, 395, 263
Carlsson, M., Judge, P. G., & Wilhelm, K. 1997, ApJ, 486,
L63+
Cauzzi, G., Falchi, A., & Falciani, R. 2000, A&A, 357, 1093
Chae, J., Choi, B., & Park, M. 2002a, Journal of Korean
Astronomical Society, 35, 59
Chae, J., Moon, Y., Wang, H., & Yun, H. S. 2002b, Sol. Phys.,
207, 73
Chae, J., Wang, H., Qiu, J., et al. 2001, ApJ, 560, 476
Cook, J. W., Brueckner, G. E., & Bartoe, J.�D. F. 1983, ApJ,
270, L89
Cooper, F. C., Nakariakov, V. M., & Williams, D. R. 2003,
A&A, 409, 325
Curdt, W. & Heinzel, P. 1998, ApJ, 503, L95+
Dame, L., Gouttebroze, P., & Malherbe, J.�M. 1984, A&A,
130, 331
De Moortel, I. & Hood, A. W. 2000, A&A, 363, 269
Deforest, C. E. & Gurman, J. B. 1998, ApJ, 501, L217+
Deubner, F.�L. & Fleck, B. 1990, A&A, 228, 506
Doyle, J. G., van den Oord, G. H. J., O'Shea, E., & Banerjee,
D. 1999, A&A, 347, 335
Egamberdiev, S. A. 1983, Soviet Astronomy Letters, 9, 385
Gallagher, P. T., Phillips, K. J. H., Harra�Murnion, L. K., &
Keenan, F. P. 1998, A&A, 335, 733
Gouttebroze, P., Vial, J.�C., Bocchialini, K., Lemaire, P., &
Leibacher, J. W. 1999, Sol. Phys., 184, 253
Habbal, S. R. & Withbroe, G. L. 1981, Sol. Phys., 69, 77
Habbal, S. R., Withbroe, G. L., & Dowdy, J. F. 1990, ApJ, 352,
333
Hagenaar, H. J. 2001, ApJ, 555, 448
Harrison, R. A., Sawyer, E. C., Carter, M. K., et al. 1995,
Sol. Phys., 162, 233
Hasan, S. S. & Kalkofen, W. 1999, ApJ, 519, 899
Kalkofen, W. 1997, ApJ, 486, L145+
Krijger, J. M., Rutten, R. J., Lites, B. W., et al. 2001, A&A,
379, 1052
Lites, B. W., Rutten, R. J., & Kalkofen, W. 1993, ApJ, 414, 345
Longcope, D. W. 1998, ApJ, 507, 433
Longcope, D. W., Kankelborg, C. C., Nelson, J. L., & Pevtsov,
A. A. 2001, ApJ, 553, 429
Lou, Y. 1995, MNRAS, 274, L1
Madjarska, M. S. & Doyle, J. G. 2002, A&A, 382, 319
---. 2003, A&A, 403, 731
Madjarska, M. S., Doyle, J. G., Teriaca, L., & Banerjee, D.
2003, A&A, 398, 775
Marsh, M. S., Walsh, R. W., & Bromage, B. J. I. 2002, A&A,
393, 649
Mazzotta, P., Mazzitelli, G., Colafrancesco, S., & Vittorio, N.
1998, A&AS, 133, 403
McAteer, R. T. J., Gallagher, P. T., Bloomfield, D. S., et al.
2004, ApJ, 602, 436
McAteer, R. T. J., Gallagher, P. T., Williams, D. R., et al. 2003,
ApJ, 587, 806
---. 2002, ApJ, 567, L165
Musielak, Z. E. & Ulmschneider, P. 2003, A&A, 406, 725
Nakariakov, V. M., Tsiklauri, D., Kelly, A., Arber, T. D., &
Aschwanden, M. J. 2004, A&A, 414, L25
Nakariakov, V. M., Verwichte, E., Berghmans, D., &
Robbrecht, E. 2000, A&A, 362, 1151
Nindos, A. & Zirin, H. 1998, Sol. Phys., 179, 253
Ofman, L. & Wang, T. 2002, ApJ, 580, L85
O'Shea, E., Banerjee, D., Doyle, J. G., Fleck, B., & Murtagh,
F. 2001, A&A, 368, 1095
Parnell, C. E. 2002, MNRAS, 335, 389
Pauluhn, A., R�uedi, I., Solanki, S. K., et al. 1999, Appl. Opt.,
38, 7035
Popescu, M. D., Doyle, J. G., & Xia, L. 2004, A&A, in press
Priest, E. R., Longcope, D. W., & Titov, V. S. 2003, ApJ, 598,
667
Priest, E. R., Parnell, C. E., & Martin, S. F. 1994, ApJ, 427, 459
Sheeley, N. R. & Golub, L. 1979, Sol. Phys., 63, 119
Teriaca, L., Banerjee, D., & Doyle, J. G. 1999, A&A, 349, 636
Thompson, W. T. 2000, CDS Software Note N o 49
Torrence, C. & Compo, G. P. 1998, Bull. Amer. Meteor. Soc.,
79, 61
Ugarte�Urra, I., Doyle, J. G., Madjarska, M. S., & O'Shea, E.
2004, A&A, 418, 313
Wilhelm, K. & Kalkofen, W. 2003, A&A, 408, 1137
Xia, L. D., Marsch, E., & Curdt, W. 2003, A&A, 399, L5
Young, P. R., Del Zanna, G., Landi, E., et al. 2003, ApJS, 144,
135
Zhang, J., Kundu, M. R., & White, S. M. 2001, Sol. Phys., 198,
347