High-Frequency Oscillations in a Solar Active Region observed with the Rapid Dual Imager

Abstract

High-cadence, synchronized, multiwavelength optical observations of a solar active region (NOAA 10794) are presented. The data were obtained with the Dunn Solar Telescope at the National Solar Observatory/Sacramento Peak using a newly developed camera system : the Rapid Dual Imager. Wavelet analysis is undertaken to search for intensity related oscillatory signatures, and periodicities ranging from 20 to 370 s are found with significance levels exceeding 95%. Observations in the H-alpha blue wing show more penumbral oscillatory phenomena when compared to simultaneous G-band observations. The H-alpha oscillations are interpreted as the signatures of plasma motions with a mean velocity of 20 km/s. The strong oscillatory power over H-alpha blue-wing and G-band penumbral bright grains is an indication of the Evershed flow with frequencies higher than previously reported. Comment: 9 pages, 9 figures

Full text

Page 1

arXiv:0707.2716v2 [astro-ph] 1 Oct 2007
Astronomy & Astrophysics manuscript no. 7142
February 1, 2008
c ? ESO 2008
High-Frequency Oscillations in a Solar Active Region
observed with the RAPID DUAL IMAGER
D.B. Jess1,2, A. Andi´ c1, M. Mathioudakis1, D.S. Bloomfield3, and F.P. Keenan1
1Astrophysics Research Centre, School of Mathematics and Physics, Queen’s University, Belfast, BT7 1NN, Northern Ireland, U.K.
2NASA Goddard Space Flight Center, Solar Physics Laboratory, Code 612.1, Greenbelt, MD 20771, USA
3Max-Planck-Institut f¨ ur Sonnensystemforschung, Max-Planck-Str. 2, 37191 Katlenburg-Lindau, Germany
Received 22 January 2007 / Accepted 16 July 2007
ABSTRACT
High-cadence, synchronized, multiwavelength optical observations of a solar active region (NOAA 10794) are presented. The data were ob-
tained with the Dunn Solar Telescope at the National Solar Observatory/Sacramento Peak using a newly developed camera system : the RAPID
DUAL IMAGER. Wavelet analysis is undertaken to search for intensity related oscillatory signatures, and periodicities ranging from 20 to 370 s
are found with significance levels exceeding 95%. Observations in the H-α blue wing show more penumbral oscillatory phenomena when
compared to simultaneous G-band observations. The H-α oscillations are interpreted as the signatures of plasma motions with a mean velocity
of 20 km/s. The strong oscillatory power over H-α blue-wing and G-band penumbral bright grains is an indication of the Evershed flow with
frequencies higher than previously reported.
Key words. Instrumentation: miscellaneous – Waves – Sun: chromosphere – Sun: oscillations – Sun: photosphere – Sun: sunspots
1. Introduction
Since the discoveryof solar oscillations in the 1960s (Leighton
1960), and their subsequent confirmation by Deubner (1975),
there has been a multitude of observational evidence presented
verifying the existence of oscillations in the solar atmosphere
(Stein & Leibacher 1974). Oscillations have been suggested as
candidates to explain one of the main unanswered questions in
solar physics – Why is the outer solar atmosphere hotter than
its surface? Acoustic oscillations as a mechanism to support
atmospheric heating in the form of wave dissipation was pos-
tulated by Schwarzschild (1948) and Biermann (1948). Theory
suggests that this form of magnetohydrodynamic(MHD) wave
can either propagate upwards from the lower solar atmo-
sphere or be induced in active regions by reconnection events
(Hollweg 1981). Early work concerned with oscillatory phe-
nomena in highly magnetic structures (e.g. sunspots), have val-
idated the detection of low-frequency oscillations which are a
response of the umbral photosphere to the 5-minute p-mode
global oscillations (see review by Lites 1992). Furthermore,
Bloomfield et al. (2007) have examined wave modes within
sunspot penumbra and have determined that modified p-mode
waves exhibit the best agreement with current observations.
Similarly,Marcu&Ballai (2005)havestudiedoscillationswith
reference to the propagation of compressional MHD waves in
penumbral filamentary structures in the photosphere.
Send offprint requests to: D.B. Jess, e-mail: djess01@qub.ac.uk
Extensive analyses of low-frequency(≤ 3.6 mHz) oscilla-
tions has shown evidence for mode coupling in the lower so-
lar atmosphere (McAteer et al. 2003, Bloomfield et al. 2004).
Higher-frequency acoustic oscillations (3.3 - 33 mHz) have
been detected in the chromosphere (White & Athay 1979a,
1979b and Athay & White 1978, 1979a, 1979b), but it
was found that the total energy flux of the oscillations was
two orders of magnitude lower than that required to bal-
ance the radiative losses from the chromosphere. Further
work on high-frequency oscillations has been performed by
Williams et al. (2001, 2002), who interpret 6 s intensity os-
cillations as longitudinal magneto-acoustic waves in an active
region coronal loop, and by Andi´ c (2007) who investigated the
radial component of chromospheric oscillations with periodic-
ities as short as 45 s.
Following analysis of TRACE data, Fossum & Carlsson
(2005a) were unable to detect sufficient power in high-
frequency waves and concluded that these waves cannot
constitute the dominant heating mechanism of the chromo-
sphere. However, this study is limited by the cadence TRACE
can achieve and the onboard filter transmissions (Fossum &
Carlsson 2005b). This means that physically small oscilla-
tion sites with short coherence lengths may be smeared out
by the coarse sampling. In addition this method may over-
look dynamic patterns created on spatial scales too small
to be resolved with TRACE (Wedemeyer-B¨ ohm et al. 2007,
Jefferies et al. 2006).

Page 2

2 D.B. Jess et al.: High-Frequency Solar Oscillations
Another interesting area incorporating oscillatory phe-
nomena is the work concerned with Evershed flows (see
Solanki2003fora review).Pioneeringworkby St. John(1913)
states that the Evershed velocity flow in sunspots induces
plasma behaviour which is detectable both in the photosphere
and chromosphere. This flow of plasma is readily observed in
H-α wings and has an associated flow speed of up to 20 km/s.
To date, there has been a multitude of observations under-
taken to search for oscillations in the upper solar atmosphere
(chromosphere, transition region and corona). However, very
little work has been carried out to search for high-frequency
oscillations in the photosphere, where the building blocks of
wave generation and propagation may exist.
Van Noort & Rouppe van der Voort (2006) have verified
the existence of highly dynamic structures, including propa-
gating waves with velocities exceeding 200 km/s, in the chro-
mosphere. This, coupled with the detection of fast fluctuations
of H-α emission, from a flare kernel on timescales of 0.3-0.7 s,
by Wang (2005) demonstrates the need for high-cadence so-
lar imaging. Furthermore, using typical chromospheric plasma
parameters, the cooling time of the chromosphere is approxi-
mately 1 s (Heinzel 1991). Therefore, in order to study such
rapidatmosphericvariations,it is necessaryto implementhigh-
cadence imaging techniques. When searching for rapid solar
variability, in particular, high-frequency oscillations, there are
many obstacles which first must be overcome. The search for
rapid, often low-amplitude intensity variations requires a suit-
ably high cadence to satisfy the Nyquist parameter, a highly
sensitive camera system providing accurate, sustained frame
rates, good seeing and a minimally weakened signal caused
by the convolution of wave perturbations with a wide response
function.
Here we report intensity oscillations detected within the
lower solar atmosphere using a new camera system devel-
oped at Queen’s University Belfast – the RAPID DUAL IM-
AGER (RDI). In § 2 we provide a brief background to the RDI
system used for our observations, which are described in de-
tail in § 3. In § 4 we discuss the methodologies used during
the analysis of the observations and the search for reliable os-
cillatory signatures. A discussion of our results in the context
of Evershed oscillations and penumbral waves is in § 5, and
finally, our concluding remarks are given in § 6.
2. Rapid Dual Imager
The RDI was developed as a follow on to the highly successful
SECIS (Phillipset al. 2000)camerasystem andconsists oftwo
Basler A301b cameras – one ‘master’ and one ‘slave’ channel
– connected to a custom built PC, and controlled by software
designed and developed by 4C’s of Somerset, UK. The cam-
eras have a 502 × 494 pixel2CCD, with a square pixel size of
9.9 µm and can operate at a maximum speed of 80 fps. Their
lightweight, robust design and C-mount housing make them
ideal for a portable system such as RDI. The camera mounts act
as passive heatsinks, ensuring that the cameras are kept within
the recommended operating temperatures.
Each camera is controlled via a PCI interface card. A
lossless compression algorithm loaded into an on-board field-
Table 1. Details of image acquisition.
ObservationCentral
Wavelength
(˚ A)
4305.50
6561.51
Filter
Bandpass
(˚ A)
9.20
0.21c
Height of
Formation
(km)
< 250a,b
< 200d,e
G-band
H-α blue wing
a: Uitenbroek & Tritschler (2006)
b: Rimmele (2004)
c: Beckers et al. (1975)
d: Ding et al. (2001)
e: Leenaarts et al. (2006)
programmable gate array increases the data throughput rate to
the PC. The cameras are attached to the control cards via 5-
metre-long high-bandwidthcables which carry data, command
signals (via a standard RS-232 serial link) and power, and are
synchronized via a ribbon connecting the interface cards. The
master card’s clock is set from the PCI bus clock; the slave
clock is in turn synchronized to the master clock so that both
channels are truly synchronous. RDI’s high rate of data record-
ing is made possible using an IDE RAID array of two pairs of
‘striped’ disks, each pair acting as one data storage area, as-
signed to a particular camera. While images are streamed to
two preallocated files, sample images from the stream are dis-
played on-screen to allow the observers to check factors such
as image quality and seeing conditions.
The preallocated space is filled at a semi-compressed rate
of 195 kB per image per camera. For maximum window size
at a frame rate of 20 Hz, this corresponds to an uninterrupted
data run of over 7 hr using the full available 200 GB of hard-
disk space. The data are then converted to FITS format us-
ing the built in conversion software and backed up onto tape
drivesor externalhard-drives. RDI is equippedwith a multitude
of additional features, such as the ability to window specific,
user-defined regions of the field of view and also to bin multi-
ple groups of pixels together for increased count-rates. RDI re-
mains stationed at the National Solar Observatory, Sacramento
Peak as a common-user instrument.
3. Observations
The data presented here are part of an observing sequence ob-
tained on 2005 August 10, with the Richard B. Dunn Solar
Telescope (DST) at Sacramento Peak. The optical setup al-
lowed us to image a 50.4′′× 49.2′′region surrounding active
regionNOAA10794completewithsolarrotationtracking.The
active region under investigation was located at heliocentric
co-ordinates (770′′,−254′′), or S12W56 in the solar NS-EW
co-ordinate system. A Zeiss universal birefringent filter (UBF;
Beckers et al. 1975) was used for H-α blue-wing (H-α core -
1.3˚ A) imaging with one of the RDI CCD detectors. In addition,
a G-band filter was employed with the second RDI camera to
enable synchronized imaging in the two wavelengths. During
the observations presented here, low-order adaptive optics was
implemented.

Page 3

D.B. Jess et al.: High-Frequency Solar Oscillations3
Fig.1. The G-band (left) and H-α blue-wing (right) field of view. The white contours mark areas where oscillations with a 26 s periodicity
are detected for a particular instant in time. The scale is in heliocentric co-ordinates (1 pixel = 0.1′′) and all contoured oscillation sites have
confidence levels above 99% and abide by the criteria enforced in § 4.1. The arrow indicates the direction of disc centre.
The observations employed in the present analysis consist
of 72000 images in each wavelength, taken with a 0.05 s ca-
dence, providing one hour of uninterrupted data. The acquisi-
tion time for this observing sequence was early in the morning
and seeing levels were good with minimal variationthroughout
the time series. Table 1 lists the wavelengths and filter pass-
bands used in the image acquisition, alongside their approxi-
mate heights of formation. The height of formation of the H-α
line profilehas long been a topic of key interest amongground-
based solar observers. Leenaarts et al. (2005) have carried out
magneto-convection simulations and show that the formation
heightoftheH-α blue wing(core- 0.8˚ A)is well under500km.
Thus, our use of the far H-α blue wing will provide a forma-
tion height lower than this, forming at a height of 200 km or
less (Leenaarts et al. 2006). The H-α line, however, is very
sensitive to solar activity. Indeed, during flares the formation
height can be pushed into the chromosphere up to a height
of approximately 1500 km (Ding et al. 2001). During the ob-
servations discussed here, solar activity was minimal with no
flare activity registered within ± 4 days. Utilizing the models
of Leenaarts et al. (2005, 2006) we therefore conclude that the
formation height of the H-α blue wing in our observations is
less than 200 km.
A comparison of Figs. 1 and 2 reveals that many solar fea-
tures, (sunspot, bright points, granulation etc.), are co-spatial
and of similar proportions, thus indicating that the H-α blue-
wing (core - 1.3˚ A) forms at an atmospheric height similar to
the G-band. Indeed, Uitenbroek & Tritschler (2006), through
the use of high-resolution synthetic CH- and CN-band filter-
grams, compute the formation height of the G-band to be ap-
proximately200km.Severaldifferences,however,betweenthe
two filter bandpasses exist. For example, magnetic elements in
the H-α blue wing appear to be much brighter and granulation
contrast much weaker when compared to simultaneous G-band
images. Thus, the similar height formation of the two chosen
bandpasses, added to their subtle imaging differences, provide
us with an interesting platform to study processes occurring in
the photosphere. The acquired images have a sampling of 0.1′′
per pixel to match the telescope’s diffraction limited resolution
in the H-α blue wing to that of the CCD. The optical path to
both cameras were identical meaning the G-band camera was
slightly undersampled.This is desirableto keep the dimensions
of the field of view the same for both cameras.
4. Data Analysis
4.1. Initial Image Processing
After implementing temporal Fourier analysis on 2400 succes-
sive dark images, a moving-pattern noise was revealed, show-
ing up as an intense Fourier power with a variable periodic-
ity between 3 and 9 s. This banding was identical in magni-
tude and periodicity on both cameras. To insure this camera
induced moving-pattern noise did not interfere with the anal-
ysis, all periodicities less than 20 s were neglected. Arrays of
dark-subtracted and flat-fielded data were saved for each cam-
era for subsequent analysis. To compensate for camera jitter
and large-scale air-pocket motions, all data was subjected to
a Fourier co-aligning routine commonly available in the SSW
tree of IDL. Thisroutineutilizes cross-correlationtechniquesas
well as squared mean absolute deviations to provide sub-pixel
co-alignmentaccuracy.After 5 successive co-alignmentsof the
data, the maximum x- and y-displacements, over the entire du-
ration of the dataset, are both less than one tenth of a pixel.
Since the observing sequence was obtained in the early hours
of the morning, when image warping is particularly strong, all
data were de-stretched relative to simultaneous, high-contrast
G-band images. We use a 40 × 40 grid, equating to a 1.25′′
separation between spatial samples, to evaluate local offsets
between successive G-band images. Due to both cameras shar-
ing the same pre-filter optical path, all determined local offsets
are applied to simultaneous narrowband images to compen-

Page 4

4 D.B. Jess et al.: High-Frequency Solar Oscillations
Fig.2. Speckle reconstructed G-band (left) and H-α blue-wing (right) images. The white contours mark areas where oscillations with a 33 s
periodicity are detected. The scale is in heliocentric co-ordinates (1 pixel = 0.1′′) and all contoured oscillation sites have confidence levels
above 99% and abide by the criteria enforced in § 4.1. The arrow indicates the direction of disc centre.
sate for spatial distortions caused by atmospheric turbulence
and/or air bubbles crossing the entrance aperture of the tele-
scope. The fine destretching grid implemented in this process
allows small-scale seeing conditions, of 1′′to 2′′, to be com-
pensated for.
After successfulco-alignmentanddestretching,lightcurves
were created for each pixel of each camera beforebeing passed
into Fast Fourier Transform (FFT) and wavelet analysis rou-
tines. While a FFT searches for periodic signatures by decom-
posing the input signal into infinite length sinusoidal wave-
trains using a basic exponential function, wavelet analysis uti-
lizes a time localised oscillatory function continuous in both
frequency and time (Bloomfield et al. 2004) and is there-
fore highly suited in the search for transient oscillations. The
wavelet chosen for this study is known as a Morlet wavelet
and is the modulation of a sinusoid by a Gaussian envelope
(Torrence & Compo 1998). Strict criteria were implemented
duringwaveletanalysistoinsurethatoscillatorysignaturescor-
respond to real periodicities. The first is a test against spurious
detectionsofpowerthatmaybeduetoPoissonnoise,wherethe
input lightcurve is assumed to be normally distributed (consis-
tent with photon noise) and following a χ2distribution with
two degrees of freedom. A 99% confidence level is calculated
by multiplying the power in the background spectrum by the
values of χ2corresponding to the 99th percentile of the distri-
bution (Torrence & Compo 1998, Mathioudakis et al. 2003).
The second criterion applied is a comparison of the input
lightcurve with a large number (1500) of randomized time-
series with an identical distribution of counts. The probabil-
ity, p, of detecting non-periodic power is calculated for the
peak power at each timestep by comparing the value of power
found in the input lightcurve with the number of times that the
power transform of the randomized series produces a peak of
equal or greaterpower. A percentageconfidenceis attributedto
the peak power at every time step in the wavelet transform by
(1 − p) × 100, such that a high value of p implies that there is
no periodic signal in the data, while a low value suggests that
the detected periodicity is real (see Banerjee et al. 2001).
Our final wavelet criterion is the exclusion of oscillations
which last, in duration, less than 1.41 cycles. This is con-
sistent with the decorrelation time defined by Torrence &
Compo (1998). One can distinguish between a spike in the data
and a harmoniousperiodiccomponentat the equivalentFourier
frequency by comparing the width of a peak in the wavelet
power spectrum with the decorrelation time. From this, the os-
cillation lifetime at the period of each power maximum is de-
fined as the interval of time from when the power supersedes
95% significance to when it subsequently dips below 95% sig-
nificance (McAteer et al. 2004). The lifetime was then divided
by the period to give a lifetime in terms of complete cycles
(Ireland et al. 1999). Any oscillations which last for less than
thisminimumdurationwerediscardedas theymayhavesimply
been a spike in the lightcurve.
Four-dimensional maps containing spatial information as
well as wavelet power and oscillatory period were saved as
outputs of wavelet analysis for detected oscillations which lay
above the 95% significance level set by the criteria above.
4.2. Speckle Reconstructed Data
Small-scale turbulent seeing in the Earth’s atmosphere means
that even high-order adaptive optics cannot compensate for all
rapid air motions, and speckle reconstruction is often used as a
powerfulpost-processingroutinedesignedtorestorethedatato
diffraction-limited resolution. Here we implement the speckle
masking method of Weigelt & Wirnitzer (1983), adapted for
solar imaging by von der L¨ uhe (1993) and further improved
by de Boer (1995). By observing at a high cadence, the short
exposuretimes acquiredessentially freeze out atmosphericdis-
tortions and maintain signals at high spatial frequencies, albeit
with statistically disturbed phases (S¨ utterlin et al. 2001). It is
possible to recover the true amplitudes and phases in Fourier
space by taking a large number of such short exposure images,

Page 5

D.B. Jess et al.: High-Frequency Solar Oscillations5
Fig.3. The top left shows a frame taken from the H-α blue-wing datacube followed by a zoomed in portion of the penumbra in the top right.
The pixel location from which the lightcurve used in wavelet analysis was created is indicated by the cross. The bottom diagram shows the
original pixel lightcurve in a). The wavelet power transform along with locations where detected power is at, or above, the 99% confidence level
are contained within the contours in b). Plot c) shows the summation of the wavelet power transform over time (full line) and the Fast Fourier
power spectrum (crosses) over time, plotted as a function of period. Both methods have detected a well pronounced 35 s oscillation. The global
wavelet (dotted line) and Fourier (dashed dotted line) 95% significance levels are also plotted. The cone of influence (COI), cross-hatched
area in the plot, defines an area in the wavelet diagram where edge effects become important and as such any frequencies outside the COI are
disregarded. Periods above the horizontal line (dotted) fall within the COI. The probability levels (1−p)×100 as discussed in § 4.1 are plotted
in d).
called a “Speckle Burst”, and utilizing an elaborate statistical
model.
Eighty raw data frames were used for each speckle recon-
struction producing a new effective cadence of 4 s. This pro-
vides a Nyquist frequency of 125 mHz and is therefore suit-
able for the search of oscillations with periods longer than 8 s.
Typical Fried parameters obtained prior to speckle reconstruc-
tion were r0 ≈ 10 cm, indicating good post-speckle image
quality. However, to strengthen the reliability of our detections
and to remain consistent with the analysis performed in § 4.1,
onlyoscillationswithperiodicitiesgreaterthan20swereincor-
porated into the analysis. The above mentioned dark subtrac-
tion was performed identically prior to Speckle reconstruction,
after which co-aligning and pixel-by-pixel lightcurve analysis
was again implemented. Due to the increased spatial resolu-
tion provided by the implementation of Speckle reconstruc-
tion, only 3 successive runs of the co-alignment software was
required to provide sub-pixel shifts. Again, four-dimensional
maps containing spatial information as well as wavelet power
andoscillatoryperiodweresavedasoutputsofwaveletanalysis
for detected oscillations which lay above the 95% significance
level.

Page 6

6 D.B. Jess et al.: High-Frequency Solar Oscillations
Fig.4. The duration of oscillations is shown as a function of the pe-
riod and this plot demonstrates the high occurrence of H-α blue-wing
oscillatory phenomena. The minimum oscillatory duration plotted is
1.41 cycles as described in § 4.1. Each detected period is placed into
bins corresponding to its cycle duration. For example, a lifetime of
2.65 cycles is placed in the 2–3.99 cycle bin. Cycle duration bins are
placed along the x-axis beginning with 1.41–1.99, 2–3.99, 4–5.99, 6–
7.99 etc., while the detected period range is placed along the y-axis.
Evaluated periods are again placed into bins which are 10 s in size
beginning with a 0–9.99 s period bin. All oscillations with periodici-
ties less than 20 s were not studied in order to fully satisfy the data’s
Nyquist parameter statedin§4.2.Thecolour scalerepresentsthenum-
ber of pixels detecting oscillatory phenomena throughout the entire
time series and over the whole field of view which are consistent with
the criteria enforced in § 4.1.
5. Results and Discussion
Both processed and speckle reconstructed data reveal signa-
tures of high-frequency oscillations in the G-band and H-α
blue wing (Fig. 1). The well-established 5 minute global os-
cillation can be seen predominantly in regions away from the
sunspot (i.e. quiet sun), and indeed the 3 minute umbral os-
cillation can be detected. This is consistent with the work of
Brynildsen et al. (2002) who report oscillations with a period
of 5 minutes are observed in sunspot umbrae, but with con-
siderably less power than locations away from the sunspot.
However, the oscillations we will concentrate on are those of
much higher frequency. We have detected high-Fourier power
at periodicitiesbetween20and370s in boththe G-bandandH-
α bluewing,with a significantconcentrationof high-frequency
(> 20 mHz) activity in the sunspot penumbra (Fig. 2). The de-
tected oscillations have highly-correlatedglobal wavelet power
and Fourier power with the additional support of significance
levels over 95%. Figure 3 shows the detection of a 35 s oscilla-
Fig.5. Same as Fig. 4, but for the G-band. Notice the reduction in de-
tectedoscillations at higher frequencies whencompared toFig.4. This
can also be viewed via the spatial representation in Figs. 1, 2 and 6.
tion originating from a dark penumbral filament within images
acquired in the H-α blue wing.
Figures 4 and 5 show histograms relating to the number of
detected oscillations found, in the speckle reconstructed data,
using the pixel by pixel wavelet analysis technique outlined in
§4.1.Itisclearthatalargenumberofoscillationshavebeende-
tected, and that the occurrence of higher frequency oscillations
is greater in the H-α blue wing. To further test the integrity
of our results we binned 9 (3 × 3) spatial pixels together and
re-computed the behaviour of oscillatory phenomena for both
the G-band and H-α blue-wing data. This more coarse sam-
pling will help prevent small-scale pixel noise below 3 pixels
in amplitude from registering as Fourier power during wavelet
analysis. From Figure 6, it is clear that the results discussed
above are unchanged even with coarse sampling.
In this work, we detect significant amounts of oscillatory
poweroriginatingfrombrightpenumbralgrains(Figs.7and8).
We propose that this power, detected in both G-band and H-
alpha blue-wing observations, is closely linked to Evershed
flow. A mechanism promoting Evershed flow was proposed by
Schlichenmaier et al. (1998), who modelled the dynamic evo-
lution of a thin flux tube inside the penumbra. A flux tube ini-
tially positioned at the magnetopause becomes buoyant due to
radiativeheating and rises. Pressure differenceswithin the loop
are created due to radiative cooling at the photosphere which
drives an outward flow along the flux tube as it rises through
the penumbra (Schlichenmaier et al. 1998). This model also
predicts a filamentary structure characterized by a hot, nearly
verticalupflowofplasmaat thefootpointofthe filament,which

Page 7

D.B. Jess et al.: High-Frequency Solar Oscillations7
Fig.6. The speckle reconstructed G-band (left) and H-α blue-wing (right) fields of view with 33 s oscillations overplotted. In both images
the spatial sampling has been degraded by binning 9 (3 × 3) pixels together as discussed in § 5. It is clear to see that penumbral oscillations
still dominate in the H-α blue wing when compared to simultaneous G-band observations. The scale is in heliocentric co-ordinates and all
contoured oscillation sites have confidence levels above 99% and abide by the criteria enforced in § 4.1
Fig.7. Speckle reconstructed H-α blue-wing image of the penumbral
structure overplotted with locations of detected 33 s periodicity. The
scale is in heliocentric co-ordinates and all contoured oscillation sites
have confidence levels above 99% and abide by the criteria enforced
in § 4.1.
within several 100 km turns into a horizontal filament. We pro-
pose that the oscillations detectedoverbright penumbralgrains
correspond to the Evershed plasma flow along flux tubes an-
choredinthe photosphere.Indeed,Schlichenmaieret al.(1998)
state that footpointsof flux tubes creatingEvershedflow within
the penumbra could be identified by bright penumbral grains
which reiterates our belief that such bright penumbralgrain os-
cillations are associated with the Evershed flow.
In order to quantify our results, we determine the percent-
age of penumbral bright grains, both in the G-band and in the
H-α blue wing,whichdemonstrateoscillatoryphenomenaover
a range of periodicities. By imposing a minimum bright grain
intensity threshold (G-band : penumbral median + 5σ, H-α
blue wing : penumbral median + 8σ) we were able to mask
out all regions excludingpenumbralbrightgrains. By overplot-
ting co-temporal locations of oscillations in the period range
28–33 s, we were able to determine the percentage of bright
grains undergoing oscillatory phenomena. After examination
of the entire time series, the average number of penumbral
bright grains found in the G-band and H-α blue wing, respec-
tively, were 17 and 21, using the minimum intensity threshold
values stated above. With comparison to the location of high-
frequency oscillatory power, we determine that 70% of the G-
band bright grains exhibit co-spatial oscillations, while 79% of
theH-α blue-wingbrightgrainsdemonstrateoscillatorypertur-
bations. This reiterates our belief that the oscillations detected
in penumbral bright grains are indicative of the Evershed flow,
as described by Schlichenmaier et al. (1998).
From Figs 1, 2 and 6 it is clear that we detect more oscil-
latory signatures in the H-α blue-wing penumbra compared to
the G-band penumbra. Since the formation heights of the H-α
blue wing and G-band are similar, we are unable to directly in-
terpret these findings based solely on magneto-acoustic oscil-
lations. Instead, we believe that this phenomena is a physical
signature of velocity oscillations in the penumbra,whereby the
H-α line profile is periodically shifted due to plasma velocity
flows. In this regime, wavelet analysis will detect the resulting
periodic deviation in intensity as the H-α line profile is moved
across the filter’s 0.21˚ A bandpass. Contrarily, the G-band sam-
ples continuum wavelengths and has a broad bandpass (9.2˚ A)
and is therefore only sensitive to intensity rather than velocity
variations. After investigating the amplitude of intensity varia-
tions in the H-α blue-wing penumbra, we find that the average
intensity variation detected during oscillations, with respect to

Page 8

8 D.B. Jess et al.: High-Frequency Solar Oscillations
Fig.8. Speckle reconstructed G-band image of the penumbral struc-
ture overplotted with the locations of 33 s oscillations. The scale is in
heliocentric co-ordinates and all contoured oscillation sites have con-
fidence levels above 99% and abide by the criteria enforced in § 4.1.
Note the reduction in oscillation sites compared with Fig. 7 and how
such sites appear to be associated with bright penumbral grains.
an average quiescent value for that location on the penumbra,
is approximately 11%. Knowing the tuned wavelength of the
blue-wing in the H-α profile (6561.51˚ A), it is possible to de-
rive the necessary wavelength shift required to vary the inten-
sity profile by the 11% observed here. Consulting a solar spec-
tral atlas obtained using the Fourier Transform Spectrometer
at the McMath/Pierce Solar Telescope, Kitt Peak, Arizona, we
have established the corresponding wavelength shift required
to produce11% amplitudeoscillationsis approximately0.45˚ A.
This wavelength shift produces an estimate of the velocity as-
sociated with such plasma flows, and we derive this velocity
to be 20 km/s. Fig. 9 shows the plasma velocities associated
with the penumbra as observed in the H-α blue wing. It must
be noted, however, that since we are performing velocity anal-
ysis with narrow-band photometry rather than high-resolution
spectroscopy,therewillbeerrorsinvelocityassociatedwiththe
width of the filter’s bandpass. With the 0.21˚ A bandpass of the
UBF, this equates to velocity errors of the order of ± 5 km/s.
The use of a solar spectral atlas, for determining wavelength
shifts from percentage intensity perturbations, will introduce
some additional errors in the estimated velocities. The spectral
atlas chosen is produced from disc-centre, quiet sun observa-
tions, and as such may differ if near-limb, active region data is
used. Indeed, an 11% intensity amplitude could result in very
differentvelocities between disc-centre and near-limb observa-
tions. Thus, it is imperative to stress potential errors associated
with this form of analysis.
It has been known for quite some time that prop-
agating waves exist in the penumbra of active regions.
Zirin & Stein (1972) found evidence for running penumbral
Fig.9. Average velocitymagnitudes of penumbral plasmaflowsinthe
H-α blue wing determined from 28 s intensity oscillations. The veloc-
ities displayed represent those of the sunspot only, with all other ar-
eas masked out. The axis scales are in heliocentric co-ordinates while
the colour scale provides an indication to the plasma flow velocity in
km/s. Errors associated with thisdiagram may be larger than ± 5 km/s
as discussed in § 5.
waves (RPWs) in the chromosphere and Giovanelli (1972)
who, through the analysis of spectroscopic line profiles, dis-
covered associated velocity flows with RPWs of 20 km/s. To
date, there has been many detections of penumbral velocity
flows ranging from 10 to 70 km/s (Brisken & Zirin 1997,
Kobanov et al. 2006) and our detection of 20 km/s veloci-
ties is within this range. The photospheric Evershed flow is
usually associated with velocities of approximately 5 km/s
(Bellot Rubio et al. 2003) which is well below our 20 km/s
values. However,Schlichenmaier et al. (2002) provide a model
where photospheric Evershed flows may reach velocities up to
14 km/s, which is comparable with our findings after consider-
ation of the errors involved in our analysis.
Previous observations of Evershed oscillations have de-
tected periods ranging from 8 to 40 min (Rimmele 2004,
Makarchik & Kobanov 2001). We note that the penumbral
bright-grain oscillations reported here are of much higher fre-
quency than those observed to date. The importance of obser-
vations in the high-frequency domain, has been emphasized
by Schlichenmaier & Solanki (2003) who suggest that hot
Evershedupflowsmay significantlycontributeto the heatingof
the penumbravia the loss of energythroughradiativeprocesses
as the plasma flows along the flux tubes away from the photo-
sphere. In order to probe the viability of such heating mecha-
nisms, it is necessary to combine high spatial and high tempo-
ral resolution Dopplergrams of non-magnetic lines (to provide
a Doppler signal free from magnetic field effects) with high-
cadencevector-magnetograms.This will allow plasma flow ve-
locities as well as surrounding magnetic field strength to be
evaluated, which will enable derivations of the penumbral en-
ergy flux.

Page 9

D.B. Jess et al.: High-Frequency Solar Oscillations9
6. Concluding Remarks
We have presented direct evidence of high-frequency waves
propagating in an active region, and detected oscillations with
periodicities ranging from 20 to 370 s with significance levels
greater than 95% due to the rigorous wavelet criteria enforced
in § 4.1. More penumbral oscillatory phenomena are found to
be located in the H-α blue wing than in the G-band, and we
conclude that these oscillations are associated with penumbral
plasma flows.
H-α blue-wing and G-band oscillations appear to have a
strong spatial correspondence with penumbral bright grains.
This promotes our belief that the detected high-frequency os-
cillations are associated with Evershedflows due to similarities
with the model developed by Schlichenmaier et al. (1998).
Anotheraspectto be investigatedis the analysisof evolving
long-duration oscillatory phenomena. The four-dimensional
power maps produced during wavelet analysis will be studied
for signs of evolution and/or power variability.
Acknowledgements. This work was supported by the U.K. Particle
Physics and Astronomy Research Council. DBJ is supported by a
Northern Ireland Department for Employment and Learning stu-
dentship. DBJ additionally thanks NASA Goddard Space Flight
Center for a CAST studentship – in particular Doug Rabin and
Roger Thomas deserve special thanks for their endless help, sup-
port and scientific input. FPK is grateful to AWE Aldermaston for
the award of a William Penney Fellowship. Observations were ob-
tained at the National Solar Observatory, operated by the Association
of Universities for Research in Astronomy, Inc. (AURA), under co-
operative agreement with the National Science Foundation. We are
grateful to the anonymous referee for pointing out highly informative
references which aided in the interpretation of our results. We would
also like to thank Kevin Reardon and Gianna Cauzzi for informal dis-
cussions related to the interpretation of the findings presented in this
paper. Finally we would like to thank the technical staff at the DST
for perseverance in the face of atrocious weather conditions. Wavelet
software was provided by C. Torrence and G.P. Compo.1
Andi´ c, A., 2007, Sol. Phys., accepted
Athay, R. G., & White, O. R., 1978, ApJ, 226, 1135
Athay, R. G., & White, O. R., 1979a, ApJs, 39, 333
Athay, R. G., & White, O. R., 1979b, ApJ, 229, 1147
Banerjee, D., O’Shea, E., Doyle, J. G., & Goossens, M., 2001, A&A,
371, 1137
Beckers, J. M., Dickson, L., & Joyce, R. S., 1975, A Fully Tunable
Lyot-¨Ohman Filter (AFCRL-TR-75-0090; Bedford: AFCRL)
Bellot Rubio, L. R., Balthasar, H., Collados, M., & Schlichenmaier R.,
2003, A&A, 403, 47–50
Biermann, L., 1948, Zeitschrift fur Astrophysics, 25, 161
Bloomfield, D. S., McAteer, R. T. J., Mathioudakis, M., Williams, D.
R., & Keenan, F. P., 2004, ApJ, 604, 936–943
Bloomfield, D. S., Solanki, S. K., Lagg, A., Borrero, J. M., & Cally, P.
S., 2007, A&A, 469, 1155
Brisken, W. F., & Zirin, H., ApJ, 478, 814
Brynildsen, N., Maltby, P., Fredvik, T., & Kjeldseth-Moe, O., 2002,
ESA-SP 506
de Boer, C. R., 1995, A&AS, 114, 387
Deubner, F. -L., 1975, A&A, 44,371
1Wavelet
http://paos.colorado.edu/research/wavelets/.
softwareis availableat
Ding, M. D., Qiu, J., Wang, H., & Goode, P. R., 2001, ApJ, 552, 340–
347
Fossum, A., & Carlsson, M., 2005a, Nature, 435, 919–921
Fossum, A., & Carlsson, M., 2005b, ApJ, 625, 556–562
Giovanelli, R. G., 1972, Sol. Phys., 27, 71
Heinzel, P. A., 1991, Sol. Phys., 65
Hollweg, J. V., 1981, Sol. Phys., 70, 25
Ireland, J., Walsh, R. W., Harrison, R. A., & Priest, E. R., 1999, A&A,
347, 355
Jefferies, S. M., McIntosh, S. W., Armstrong, J. D., Bogdan, T. J.,
Cacciani, A., & Fleck, B., 2006, ApJ, 648, 151–155
Kobanov, N. I., Kolobov, D. Y., & Makarchik, D. V., 2006, Sol. Phys.,
238, 231-244
Leenaarts, J., Rutten, R. J., S¨ utterlin, P., Carlsson, M., & Uitenbroek,
H., 2006, A&A, 449, 1209–1218
Leighton, R. B., 1960, IAUS, 12, 321
Lites, B. W., 1992, Proceedings of the NATO Advanced Research
Workshop on the Theory of Sunspots, 261–302
McAteer, R. T. J., Gallagher, P. T., Williams, D. R., Mathioudakis, M.,
Bloomfield, D.S., Phillips K. J. H., & Keenan, F. P., 2003, ApJ, 587,
806–817
McAteer, R. T. J., Gallagher, P. T., Bloomfield, D.S., Williams, D. R.,
Mathioudakis, M., & Keenan, F.P., 2004, ApJ, 602, 436–445
Makarchik, D. V., & Kobanov, N. I., 2001, IAU Symposium, Vol. 203
Marcu, A., & Ballai, I., 2005, PADEU, 15, 103
Mathioudakis, M., Seiradakis, J. H., Williams, D. R., Avgoloupis, S.,
Bloomfield, D. S., & McAteer, R. T. J., 2003, A&A, 403, 1101–1104
Phillips, K. J. H., 2000, Sol. Phys., 193, 259–271
Rimmele, T. R., 2004, ApJ, 604, 906–923
St. John, C. E., 1913, ApJ, 37, 322
Schlichenmaier, R., Jahn, K., & Schmidt, H., U., 1998, A&A, 337,
897
Schlichenmaier, R., M¨ uller, D. A. N., Steiner, O., & Stix, M., 2002,
A&A, 381, 77-80
Schlichenmaier, R., & Solanki, S. K., 2003, A&A, 411, 257
Schwarzschild, M., 1948, ApJ, 107, 1
Solanki, S. K., 2003, A&A Review, 11, 153–286
Stein, R. F., & Leibacher, J., 1974, ARA&A, 12, 407
S¨ utterlin, P., Hammerschlag, R. H., Bettonvil, F. C. M., et al., 2001,
ASP Conference Series, Vol. 236
Torrence, C., & Compo, G. P., 1998, Bull. Amer. Meteor. Soc., 79, 61
Uitenbroek, H., Tritschler, A., 2006, ApJ, 639, 525–533
van Noort, M. J., & Rouppe van der Voort, L. H. M., 2006, ApJL, 648,
67
von der L¨ uhe, O., 1993, Sol. Phys., 268, 374
Wang, H., Qiu, J., Denker, C., Spirock, T., Chen, H., & Goode, P. R.,
2005, ApJ, 542, 1080–1087
Wedemeyer-B¨ ohm, S., Steiner, O., Bruls, J., & Rammacher, W., 2007,
ASP Conference Series, Coimbra Solar Physics Meeting
Weigelt, G., & Wirnitzer, B., 1983, Optics Letters Vol. 8, No. 7, 389
White, O. R., & Athay, R. G., 1979a, ApJs, 39, 317
White, O. R., & Athay, R. G., 1979b, ApJs, 39, 347
Williams, D. R., Phillips, K. J. H., Rudawy, P., Mathioudakis, M.,
et al., 2001, MNRAS, 326, 428–436
Williams, D. R., Mathioudakis, M., Gallagher, P. T., Phillips, K. J. H.,
et al., 2002, MNRAS, 336, 747–752
Zirin, H., & Stein, A., 1972, ApJ, 178, 85