Vous êtes sur la page 1sur 5

New Astronomy 13 (2008) 9397

www.elsevier.com/locate/newast

Wavelet analysis of solar macro-spicule recurrences


a,b,*
A. Ajabshirizadeh , E. Tavabi a, S. Koutchmy c

a
Department of Theoretical Physics and Astrophysics, Tabriz University, 51664 Tabriz, Iran
b
Research Institute for Astronomy & Astrophysics of Maragha (RIAAM), 55134-441 Maragha, Iran
c
Institut dAstrophysique de Paris & UPMC, 98 Bis Boulevard Arago, F-75014 Paris, France

Received 22 January 2007; received in revised form 8 June 2007; accepted 9 July 2007
Available online 18 July 2007
Communicated by M. van der Klis

Abstract

We looked at the temporal behavior of intensity variations of spicules and macro-spicules observed at the solar limb by TRACE (tran-
sition region and coronal explorer) EUV telescope in the 1600 A channel. Using wavelet analysis technique we got evidence of spicule
oscillations.
The timefrequency analysis provided by the wavelet analysis shows a temporal behavior of spicules with recurrences at periods of
about 210260 s with a typical lifetime of 10 min. Finally, we shortly discuss two scenarios regarding the possible origin of spicule
oscillations.
 2007 Elsevier B.V. All rights reserved.

PACS: 96.60.j

Keywords: Sun; Spicule oscillations-method; Wavelet analysis

1. Introduction (for a recent review, see Sterling, 2000) and are observed in
many optical lines, such as Ha, Hb, H&K of Ca II and
Sandwiched layers between the solar photosphere and EUV. Temperatures of spicules are dierent, from 104 to
the corona are the chromosphere and the transition region 105 (Xia et al., 2005). Spicules are usually seen all around
(TR), an irregular layer of hot plasma at temperatures of but well above the limb of the Sun. On the disk they were
600060,000 K. Some chromospheric radiations in quiet historically called mottles in quiet Sun regions but they
Sun regions come from spicules and macro-spicules, which should not be confused with the brils of lower altitudes
are jet-like chromospheric structures that erupt from the (see Athay, 1986). They usually show a group behavior
lower chromosphere at speeds of around 2550 km/s. and are concentrated between supergranule cells (see e.g.
Before they were called spikes (Koutchmy and Stellmacher, the review of Sterling (2000)). At any given time, there
1976). They reach heights of 650020,000 km before fading are at least 60,00070,000 spicules on the solar surface
out of view or falling back towards the solar surface. They (Zirin, 1988) although Athay (1986) gave a much larger
have diameters, all along their length, of only 200500 km number. Typical lifetime of spicules are 515 min and
comparable to the photospheric granules lifetime. Macro-
spicules or spikes appear to be giant spicules, these
*
Macro-spicules seem to be comprised of a group of high
Corresponding author. Address: Department of Theoretical Physics spicules (Xia et al., 2005).
and Astrophysics, Tabriz University, 51664 Tabriz, Iran. Fax:
+984113858959.
Macro-spicules were discovered from the Skylab He II
E-mail addresses: a-adjab@tabrizu.ac.ir (A. Ajabshirizadeh), tavabi@ (304 A) imaging experiments. Their temperature struc-
tabrizu.ac.ir (E. Tavabi), koutchmy@iap.fr (S. Koutchmy). ture (log T > 4) extends into the transition region (TR)

1384-1076/$ - see front matter  2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.newast.2007.07.001
94 A. Ajabshirizadeh et al. / New Astronomy 13 (2008) 9397

temperatures and their heights extend into the corona E(n, m) of unknown nature added. We intend to reduce the
(Bohlin et al., 1975). noise from A(n, m) by wavelet thresholding to recover a
The mechanism of spicule formation and evolution is not better image.
well understood (for the mechanisms of formation, see This is a non parametric method; this means that there is
reviews of Sterling, 2000 and also Lorrain and Koutchmy, no particular assumption about the data; the only assump-
1996; Filippov et al., 2007). The investigation of solar spic- tion is that the noise nature remains the same after the
ules is necessary in order to understand TR and coronal wavelet transform. As we know dierent wavelets have dif-
heating (Kudoh and Shibata, 1999). The magneto-acoustic ferent performances for the images and there is no univer-
waves may get dissipated after transforming to smaller spa- sally best wavelet transform for removal noise for all the
tial scales through phase mixing or resonant absorption dierent kinds of image because the performance of wave-
(Walsh and Ireland, 2003; Kudoh and Shibata, 1999). The let lter depends principally on inherent statistical informa-
propagation of MHD waves through the chromosphere tion of the image then we intent to emphasize the most
with shorter period than the spicule lifetime can be traced important features in the images (for spicules to detect
through spicule observation (Kukhianidze et al., 2006). the edges and increase the visibility of very small scale
The existence of 5 min oscillations in spicules, reported rst features).
by Kulidzanishvili and Nikolsky (1978) and more recently Briey, this procedure involves three steps:
the existence of 35 min oscillations in dark mottles reported
by De Pontieu et al. (2003) and in spicules by Xia et al. 1. Choose a wavelet; choose a level N. Compute the wave-
(2005), seem to agree with periods that we obtained in this let decomposition of input image at level N.
article using a dierent method. The diagnostic can be 2. Threshold detail coecients for each level from 1 to N.
improved by the use of the wavelet analysis, which is a Select a threshold and apply this to the detail coecients.
powerful tool to obtain information on the temporal behav- 3. Compute wavelet reconstruction using the original
ior of an oscillatory signal (see Baudin et al., 1996). approximation coecients of level N and the modied
detail coecients of levels from 1 to N.
2. Observations and image processing
For the details of calculations involved, see the descrip-
The data used in this work consists of a rather long tion given on the O. Koutchmy Website: (http://www.ann.
sequence of observation of spicules from the Transition jussieu.fr/~koutchmy/debruitage/spicul~2E.html).
Region and Coronal Explorer (TRACE) EUV telescope In our case, we prefer to choose Daubechies-4 (these
done in part in the 1600 A UV continuum channel, made wavelets have no explicit expression except for Daube-
on July 6, 2000. The 1600 A channel has a temperature sen- chies-1, the rst and simplest member of this family, which
sitive range of 4.010 103 K (Handy et al., 1999). The is the Haar wavelet and resembles a step function) and glo-
duration of observation was 58 min from 7:09 to 8:07 bal universal threshold, after testing dierent mother wave-
UT at location, in solar coordinate center, (Xcen =  29300 , let functions and noise estimation, in the following from
p
Ycen =  94300 ); the pixel spatial resolution is 000 .5. The time thr q 2 lognm where [n, m] is the image size and q is
interval between successive frames was 15 s. The images the noise standard deviation. The initial image was decom-
can be found at the TRACE Web site: (http://trace.lmsal. posed at level 3. We obtain a threshold number thr = 7 (see
com/trace_cat.html). Fig. 1).
In this section, we give a brief outline of some of the
main features of image processing by wavelet analysis that 3. Time series analysis using the wavelet transform
will be used.
There exist various extensions of the one dimensional A wavelet analysis is capable of revealing the time/fre-
wavelet transform to two-dimensional and indeed multi- quency aspects of the data that other signal analysis tech-
dimensional wavelet. De-noising of image is an interesting niques miss (such as Fourier analysis). As we know the
application of wavelet analysis. wavelet is ideal for analyzing non-stationary time series
The underlying model for the noisy image is basically of or time series where one expects localized variations of
the following form: power, since it allows a local decomposition of timescales
in the time series. The temporal information is obtained
An; m Bn; m rEn; m 2-1
by computing the correlation of input signal f(t) with a
where n, m is the dimensions of the image. mother wavelet function w, which is a function of nite
In the simplest mode we suppose that E(n, m) is a Gauss- duration in time (where duration has to be precisely
ian white noise and the noise level r is supposed to be equal dened, see Baudin et al., 1994; Christopoulou et al., 2003):
to 1 (many other statistical noise estimations are proposed Z t  t 
0
such as: histogram, median, minimax threshold, Birge CWTs; t0 f tw dt 3-1
s
Masart threshold, global universal threshold, etc.).
We assume that each image A(n, m) represents an obser- where s is the wavelet scale and variation of t allows the
vation of true astrophysical image, B(n, m) with a noise translation of the analyzing wavelet in time. The variation
A. Ajabshirizadeh et al. / New Astronomy 13 (2008) 9397 95

A wavelet coecient CWT (continue wavelet transforms)


depends on two parameters, the time t and the scale s. Thus
it will provide information around time t and around fre-
quency t:
p
k 2 K2
t : 3-3
4ps
The global wavelet spectrum is an averaged power spec-
trum for all the scales and frequencies. It is similar to a
smooth Fourier power spectrum and it is dened as the
squared modulus proportional to the energy of the signal.
This information will be displayed in two-dimensional
plots, showing variations of the energy with time and fre-
quency. One of the main problems of this method is to
optimize the resolution in both time and frequency do-
mains; the parameter k of Eq. (3-2) of the wavelet allows
the resolution to be set to the required values. For larger
values of k, we get a better frequency resolution but poorer
time localization, in agreement with the Heisenbergs rela-
tion (Baudin et al., 1994). We choose k = 6 (following De
Moortel et al. (2004)) which is a good compromise between
time and frequency resolution (t  1/s), enabling the study
of the temporal variation with satisfactory time localiza-
tion and a signicantly accurate frequency resolution.
Thus, with suitably optimized parameters, the time/fre-
quency analysis will be able to give signicant information
in both domains.

4. Discussion-results

In this section we analyze the time behavior of the inten-


sity variations. In Eq. (3-1), f is the input signal, it is the
integrated intensity over the selected spicule (we select a
box and sum the emission within that box).
The results of wavelet power analysis for two dierent
spicules are shown in Fig. 2a and b. In this gure the left
plots show wavelet power: the horizontal axis is the time
in seconds, while the vertical axis is the period in seconds.
Darker shades correspond to higher power (95% con-
dence) of the wavelet coecients. The contours determine
Fig. 1. Top: A Trace 1600 A image take from a time series from 7:09 to the 95% condence level which was calculated by assuming
8:07 UT on 2000 July 6. Bottom: The selected region by box and shows jet-
a white noise background spectrum. Wavelet transform
like structures in the chromosphere, before (a) and after (b) denoising
using Daubechies-4 and after (c) using the unsharp mask (from Christo- suers from the wraparound errors at both edges of a
poulou et al., 2001). In (c) we can see spicules and macro-spicules with time series (of nite length). The regions in which these
possible small loops at their feet. In (c) black crosses mark location where eects are important are dened by the cone of inuence
we found evidence of oscillations with signicant (95%) wavelet power (COI; see Torrence and Compo, 1998).
for three cycles, see Fig. 2a and b.
The COI region, or in other words the region that suers
from edge eects, is chosen as the e-folding time of the
of these two parameters leads to the time/frequency local- Morlet wavelet and it is presented as a cross-hatched in
ization. Several dierent mother wavelet functions can be each plot.
selected, but when applied to an oscillatory signal the best The right plots show, from a precise investigation of the
choice is the Morlet wavelet (Grossmann and Morlet, input signal (f) variations. We have investigated 20 dierent
1984). The Morlet wavelet is just a sine wave modulated spicules and obtained that nearly 30% of them show (quasi)
by a Gaussian window: periodic behavior.
t      Our results verify that the input signal (f) shows distinct
ikt 1  t 2 peaks for periods of about 210260 s; with frequency
w exp exp  ; 3-2
s s 2 s around 4.5 mHz (spicules show typically a range of
96 A. Ajabshirizadeh et al. / New Astronomy 13 (2008) 9397

Fig. 2. Left plots show time/frequency variation of the power spectrum for two dierent spicules. Cross-hatched regions indicate the cone of inuence
(COI). The darkest regions mean higher power, and the contours correspond to the 95% condence level (see text and reference for details). Right plots:
The input signal (f) curves (f vs. time), correspond to the location of black crosses in Fig. 1c bottom panel, and show that spicule oscillations are not
steady harmonic wave, but they are quasi-periodic features, or in other words wave trains of nite duration.

lifetimes from 500 to 600 s, for three cycles). Spicules rise ules allow signicant leakage and propagation of p-modes
rapidly and reach their maximum height (after about (periods of 5min) from the photosphere into the chromo-
12 min), remain about 1 min in this phase, and then start sphere and TR when inclined motion is considered (Suema-
to fall back down. tsu, 1990).
An important question is: What is the cause of these Several authors have studied the temporal variation of
periodic ows? Based on this consideration, our suggested the intensity of the chromospheric and TR spectral lines.
scenarios are the following: Doyle et al. (1998) found that the intensities of the TR
spectral line (O V 626.73 A) show oscillations with domi-
(i) The periodic upward and downward motion of spic- nating periods around 250 s with a lifetime approximately
ules could be due to the nonstationary character of 1015 min and for the chromospheric spectral line (HeI
granular motion (see Baudin et al., 1996), i.e., spic- 584 A) oscillations appear with periodicity of 220 s and
ules oscillated when granules move with respect to typical lifetimes were about 1020 min. Moreover, slightly
each other. dierent values of periods and lifetimes of the intensity
(ii) These oscillations seem to be related to p-modes, but oscillations for the transition region and chromosphere
it is evident that if spicules are driven by p-modes, have been reported by Hansteen et al. (2000), Banerjee
crucial details about their formation are still missing. et al. (2001), and more recently by Gomory et al. (2006).
Clearly, not all spicular ows are periodic, whereas However, our results suggest that some of the chromo-
most photospheric oscillations are. In addition, the sphere and TR oscillations may be associated with periodic
horizontal scale for amplitude coherence of p-modes ows in spicular structures, because their periods and life-
(8000 km) is well beyond the width of brils (De times are equal, but the relationship of the chromospheric
Pontieu et al., 2003). and TR oscillations with spicule oscillations is less clear.
This could be indicative of a lack of relationship or perhaps
However, we should remark the increasing of local some kind of nonlinear relationship between these oscilla-
acoustic cut-o period (Pc) inside the spicule because of tions. A detailed discussion of this subject needs more con-
the higher temperatures and, more importantly, the spic- siderations. A last remark concerns the distinction to be
ules inclination. Pc dramatically increases from 180 s, in made between the rather cool spicules and the rather hotter
the upper photosphere, to more than 300 s inside spicules macro-spicules not to be confused with the SXR coronal
(see De Pontieu et al., 2004). So, spicules and macro-spic- jets.
A. Ajabshirizadeh et al. / New Astronomy 13 (2008) 9397 97

Acknowledgements De Pontieu, B., Erdelyi, R., De Wijn, A.G., 2003. ApJ 595, L63.
De Pontieu, B., Erdelyi, R., James, S.P., 2004. Nature 430, 536.
Doyle, J.G., Van den Oord, G.H.J., OShea, E., Banerjee, D., 1998. Solar
TRACE is a mission supported by NASA and Lock- Phys. 181, 51.
heed. The authors are most grateful to F. Baudin for useful Filippov, B., Koutchmy, S., Vilinga, J., 2007. Astron. Astrophys. 464,
discussions and critical reading of the manuscript. Image 1119.
processing wavelet software was provided by O. Koutchmy Gomory, P., Rybak, J., Kucera, A., Curdt, W., Wohl, H., 2006. Astron.
http://www.ann.jussieu.fr/~koutchmy/index_newE.html, and Astrophys. 448, 1169.
Grossmann, A., Morlet, J., 1984. SIAM J Math Anal 15, 723.
wavelet analysis software was provided by Torrence and Handy, B.N. et al., 1999. Solar Phys. 187, 229.
Compo (http://atoc.colorado.edu/research/wavelets), this Hansteen, V.H., Betta, R., Carlsson, M., 2000. Astron. Astrophys. 360,
work has been supported partly by RIAAM. 742.
Koutchmy, S., Stellmacher, G., 1976. Solar Phys. 49, 253.
References Kudoh, T., Shibata, K., 1999. ApJ 514, 493.
Kukhianidze, V., Zagarashvili, T.V., Khutsishvili, E., 2006. Astron.
Astrophys. 449, L35.
Athay, G., 1986. In: Sturrock, P.A. (Ed.), Physics of the Sun, vol. II.
Kulidzanishvili, V.I., Nikolsky, G.M., 1978. Solar Phys. 59, 21.
Reidel pub., p. 51.
Lorrain, P., Koutchmy, S., 1996. Solar Phys. 165, L115.
Banerjee, D., OShea, E., Doyle, J.G., Goossens, M., 2001. Astron.
Sterling, H.C., 2000. Solar Phys. 196, 79.
Astrophys. 371, 1137.
Suematsu, Y., 1990. In: Osaki, Y., Shibahashi, H. (Eds.), Proceedings of
Baudin, F., Gabriel, A., Gilbert, D., 1994. Astron. Astrophys. 285, L29.
the Oji International Seminar Held at Hakone, Japan, 1114 Decem-
Baudin, F., Bocchialini, K., Koutchmy, S., 1996. Astron. Astrophys 314,
ber, 1989. Lecture Notes in Physics, vol. 367. Springer-Verlag, Berlin,
L9.
Heidelberg, New York, p. 211.
Bohlin, J.D., Vogel, S.N., Purcell, J.D., Sheeley, N.R., Tousey, R.,
Torrence, C., Compo, G.P., 1998. Bull. Amer. Meteor. Soc. 79, 61.
Vanhoosier, M.E., 1975. ApJ 197, L13.
Walsh, R.W., Ireland, J., 2003. Astron. Astrophys. Rev. 12, 1.
Christopoulou, E.B., Georgakilas, A.A., Koutchmy, S., 2001. Solar Phys.
Xia, L.D., Popescu, M.D., Doyle, J.G., Giannikakis, J., 2005. Astron.
199, 61.
Astrophys. 438, 1115.
Christopoulou, E.B., Skodras, A., Georgakilas, A.A., Koutchmy, S., 2003.
Zirin, H., 1988. Astrophysics of the Sun. Cambridge University Press,
ApJ 591, 416.
Cambridge.
De Moortel, I., Munday, S.A., Hood, A.W., 2004. Solar Phys. 222, 203.

Vous aimerez peut-être aussi