Temporal Evolution of Chromospheric Oscillations in Flaring Regions - A Pilot Study
DD RAFT VERSION S EPTEMBER
25, 2018Typeset using L A TEX default style in AASTeX62
Temporal Evolution of Chromospheric Oscillations in Flaring Regions – A Pilot Study T ERESA M ONSUE , F RANK H ILL , AND K EIVAN
G. S
TASSUN Department of Physics and Astronomy, Vanderbilt University, Nashville, TN 37235, USA National Solar Observatory, 3665 Discovery Dr, Boulder, CO 80303 (Received 2016 February 6; Revised 2016 May 17; Accepted 2016 May 23)
Submitted to AJABSTRACTWe have analyzed H α intensity images obtained at a 1 minute cadence with the Global Oscillation NetworkGroup (GONG) system to investigate the properties of oscillations in the 0–8 mHz frequency band at the locationand time of strong M- and X-class flares. For each of three sub-regions within two flaring active regions, weextracted time series from multiple distinct positions, including the flare core and quieter surrounding areas.The time series were analyzed with a moving power map analysis to examine power as a function of frequencyand time. We find that, in the flare core of all three sub-regions, the low-frequency power ( ∼ Keywords:
Sun: oscillations – Sun: flares, Sun: helioseismology, sunspots – Sun: chromosphere – techniques:image processing INTRODUCTIONThe relationship between solar acoustic oscillations, active regions, and energetic flares remains an open question in solarphysics. A fundamental question is whether acoustic oscillations are enhanced, suppressed, or perhaps both, in active regions andby flares, and how these power enhancements and/or suppressions behave as functions of time before, during, and after energeticflaring events. Solar flares release great amounts of energy and so in principle are capable of exciting acoustic oscillations in themagnetically active sunspot regions, perhaps by exciting velocity oscillations in regions where a higher-class solar flare has takenplace (Kumar & Ravindra 2006). The first clear observations of helioseismic waves produced by a flare were by Kosovichev andZharkova (1998) using data from the Michelson Doppler Imager on board the Solar and Heliospheric Observatory space mission.It is expected that this production of waves will change the characteristics, such as the power, of the p-modes (Ambastha et al.2002, 2003b,a; Kumar & Ravindra 2006; Kumar et al. 2006, 2010, 2011).In addition, Braun et al. (1987) and others have observed that sunspots absorb acoustic power in the photosphere. Theseauthors found that outgoing waves were reduced in amplitude by 50% compared to incoming waves (Braun et al. 1987).Observations of acoustic modes in the chromosphere using the H α line were first carried out by Elliott (1969) and later byHarvey (1993). Observations of H α intensity oscillations in solar flares were performed by Jain et al. (1999). Chromosphericoscillations in the 3.4 mHz and 5.6 mHz p-mode frequencies were found, confirming Elliott’s study of their existence and alsoagreeing with Kneer & von Uexkuell (1983). They surveyed 18 locations around two flares with Fourier power spectra and clearlyfound prominent 5 and 3 minute modes in H α (Jain & Tripathy 1998). Furthermore, the observation of short-time variations of the Corresponding author: Teresa [email protected] a r X i v : . [ a s t r o - ph . S R ] M a r M ONSUE ET AL .Sun’s chromosphere observed in H α , contributes to our understanding of the atmospheric dynamics and could reveal progenitorsfor chromospheric heating mechanisms (Kneer & von Uexkuell 1985).We are interested in studying the acoustic frequency spectrum to investigate energy transfer in the chromosphere (Kneer &von Uexkuell 1983). Our method incorporates a short time-series analysis to study the time variations in the Sun’s chromosphereobserved in H α . Studying the chromosphere in the Fourier spectrum, and observing the temporal evolution gives us an advantagein that it provides a way to trace the temporal behavior of specific structures involved in the flare, which can be rather difficult tofollow over the course of several minutes or hours otherwise (Kneer & von Uexkuell 1985).A technique for studying the p-mode excitation due to solar flares in a three-dimensional Fourier analysis was devised byJackiewiez and Balasubramaniam (2013) in their “frequency-filtered amplitude movies (FFAMs)” . This method is simple andpowerful in that it preserves the initial 3-dimensional information in the inputted time series by incorporating a moving powermap method. Here we employ this novel approach to the Global Oscillation Network Group H α data set to create power-mapmovies (PMMs). From these PMMs we investigate the H-alpha oscillatory modes across the frequency band (0 < ν < α time-series data that we employ in our analysis. We specificallystudy three active regions, directly over sunspots, in which M- or X-class flares occurred on 2012 June 13 and 2012 July 12.Section 3 describes our data reduction and analysis methods. In Section 4 we then present the results of how the frequencydistribution evolves temporally and spatially by constructing a PMM of each region. We find that, in the core regions of allthree flares, the low-frequency power ( ∼ OBSERVATIONAL DATAThe data set used here comprises GONG H α intensity images centered at a wavelength of 6562.8 ˚A. The GONG H α networkbecame operational in 2010; it consists of a set of six detectors placed around the Earth for nearly continuous solar full-diskobservations. The images have a cadence of 1 minute and a format of 2048 × EMPORAL E VOLUTION OF C HROMOSPHERIC O SCILLATIONS IN F LARING R EGIONS T a b l e . T a b l e o f ob s e r v a ti on a li n f o r m a ti ono f t h e t h r ee d a t a s e t s . D a t a D a t e A c ti v e F l a r e A R E v e n t I m a g e s GONG C a d e n ce S e t R e g i on ( A R ) A v e r a g e L o ca ti on s C l a ss T i m e ( U T ) S t a ti on [ m i n ] A R J un e –2012 A R S E (- ” , - ” ) M : E l T e i d e A R J un e –2012 A R S E (- ” , - ” ) M : E l T e i d e A R J u l y–2012 A R S W ( ” , - ” ) X . : C e rr o T o l o l o1 M ONSUE ET AL . Figure 1.
Analysis was done on solar flare regions directly over sunspots where the magnetic field was concentrated. a) The event is the M1flare that occurred on 2012 June 13 at 13:19UT. b) Our region of interest is AR1, an ideal candidate with an active flaring region directly overa sunspot in AR11504, and the smaller active region AR2, a companion sunspot within that region. c) The main regions of interest have arectangular area of approximately between 55 (cid:48)(cid:48) × (cid:48)(cid:48) for AR1, and 65 (cid:48)(cid:48) × (cid:48)(cid:48) for AR2. EMPORAL E VOLUTION OF C HROMOSPHERIC O SCILLATIONS IN F LARING R EGIONS Figure 2.
Analysis was done on solar flare regions directly over sunspots where the magnetic field was highly concentrated. a) The event isthe X1.4 flare that occurred on 2012 July 12 at 16:53UT. b) Our region of interest, AR3, was a nearly spherical flaring region directly over asunspot in active region AR11520. The sunspot morphology of AR11520 is depicted in the GONG continuum image below. c) The main regionof interest, AR3, has a rectangular area of approximately 48 (cid:48)(cid:48) × (cid:48)(cid:48) . M ONSUE ET AL . DATA ANALYSIS AND REDUCTION METHODSWe employ a technique similar to the “FFAMs” (Jackiewicz & Balasubramaniam 2013) to construct a PMM. In summary, aPMM consists of a time series of acoustic power maps that show wave power as a function of frequency and space, but with aninitial starting time that is systematically offset. For a given starting time, the power maps are created by applying a fast Fouriertransform (FFT) in the temporal direction for each pixel in the region of interest. The starting time is then shifted by one minuteand the procedure is repeated.In detail, we start with a time series of GONG H α intensity images, I ( x, y, t ) in a data cube with a cadence of 1 minute. Wedefine a region of interest ( I ROI ) covering the area x ≤ x ≤ x and y ≤ y ≤ y within the large data cube. We extract thetime series at each ROI pixel over a temporal length N starting at time t i . Missing images are replaced by a value of zero. Theinput data for the PMM are constructed by incrementing t i . Explicitly, we have I ROI ( x (cid:48) , y (cid:48) , t (cid:48) , t i ) = I ( x ≤ x ≤ x , y ≤ y ≤ y , t i ≤ t ≤ t i + N ) (1)For our analysis here, t i is incremented by one minute, ≤ t i ≤ , and N = 60 , producing a set of 61 time series, each an hourlong, that start at one hour prior to flare maximum, and end at one hour after flare maximum.To create a single frame, P , in the PMM, we apply an FFT in the temporal direction to the time series for each spatial pixel in I ROI , and then take its modulus, producing P ( x (cid:48) , y (cid:48) , ν, t i ) = | F F T ( I ROI ( x (cid:48) , y (cid:48) , t (cid:48) , t i )) | (2)where ν is the temporal frequency. The power P is then averaged in frequency bins ν i , producing one frame in the PMM,Equation (3). P M M ( x (cid:48) , y (cid:48) , ν j , t i ) = ∆ νν − ν (cid:88) ν <ν j <ν P ( x (cid:48) , y (cid:48) , ν, t i ) (3)where ∆ ν is the frequency resolution of the power spectrum (277.8 µ Hz) and ν and ν are the lower and upper limits respectivelyof frequency bin j . We define a set of seven bins ν j of width 1 mHz starting at 1 mHz, with the highest band covering 7–8.33mHz.Within the areas AR1–AR3, we extract a number of smaller regions of 3 × (cid:48)(cid:48) in length, and thenaverage the PMM values within these small areas. Averaging the pixels over a smaller region improves the signal-to-noise ratioand allows us to isolate different physical conditions. We kept all subregions the same size to maintain consistency within theexperiment.To compare the PMM data with the intensity, we carry out a 60 minute running average of the intensity that is incremented byone minute to provide the same sampling as the PMM. Figure 3 shows a power-map frame for region AR3 for each ν j along withthe averaged intensity. The sampling window in Figure 3 is from 32 to 92 minutes in the total time series. EMPORAL E VOLUTION OF C HROMOSPHERIC O SCILLATIONS IN F LARING R EGIONS INTENSITY 1-2mHz [ a r c s ec ] Y - R O I s i ze X-ROI size [arcsec]0 10 20 30 40 L og P o w e r Figure 3.
The flare intensity and frequency-binned PMM frames of the X1.4 flare on 2012 July 12 in region AR3. The PMM frames are in thetime window of 32–92 minutes. Dark lanes are seen in each frame and are sharper in the 1–2 mHz and 2–3 mHz ranges. The power is measuredin arbitrary instrumental units. 4.
RESULTSIn this section, we present the results of our time-series power spectrum analysis on each of the three flaring active regionsdefined in Section 2. Within each of the three main analysis regions—AR1, AR2, and AR3—we define several subregions inand around the centers of the active regions in order to relate the temporal behavior of the oscillation power to different spatialpositions. As we discuss below, the overall results for all three analysis regions are qualitatively similar, so we describe the resultsfor one of the three regions in detail and then more briefly summarize the similar results for the other two regions. We find that, M
ONSUE ET AL .in the core regions of all three flares, the low-frequency power ( ∼ AR3
There are three main groups of subregions for area AR3. Figure 4 shows seven rectangular subregions, each approximately3.2 (cid:48)(cid:48) × (cid:48)(cid:48) , distributed around the solar flare and the dark lanes. The subregions are selected to probe the temporal behaviorat various levels of overall power and at various positions relative to the center of the active region. We present the results bygrouping the subregions accordingly. Obviously these are not the only subregions that could be selected; we emphasize that thisselection of subregions is made arbitrarily, and on the basis of visual impression, but in an attempt to sample the active regionat various representative locations relative to the peak of activity. The regions with the lowest power (log power 0.5 to 2.0), areAR3 1 and AR3 6 (Figure 7 (a) and (b)). The regions with mid-level power levels (log power 2.0 to 3.0) are AR3 2 and AR3 5(Figure 6 (a) and (b)). The regions located in the middle of the flare (AR3 3, AR3 4 and AR3 7) exhibited the greatest amountof power (Figure 5(a)–(c)), with logarithmic values from 3.0 to 4.5. EMPORAL E VOLUTION OF C HROMOSPHERIC O SCILLATIONS IN F LARING R EGIONS Y - R O I s i ze ( a r c s ec ) L og P o w e r AR3, 2-3mHz
AR3_1AR3_2AR3_3AR3_4 AR3_5AR3_6 AR3_7
Figure 4.
Region AR3 has a rectangular area of approximately 48.15 (cid:48)(cid:48) × (cid:48)(cid:48) . There are seven subregions sampled in AR3 for the X1 solarflare on 2012 July 12. The subregions are approximately 3.2 (cid:48)(cid:48) × (cid:48)(cid:48) in size. In the above figure representing AR3, the PMM frame depictedis in the time window of 34–94 minutes. Inner Flaring Regions –
Locations 3, 4, and 7 in AR3
These three subregions are placed along the dark lanes and in a bright region in the flare. Figure 5(a)–(c) shows the poweras a function of time and frequency in these subregions. We observe an increase in power across the entire frequency bandonce the flare begins. However, as the intensity increases in each subregion, we observe a suppression of power that beginsfirst at the higher frequencies, and then moves toward lower frequencies as time progresses. Maximum power suppression at allfrequencies occurs at the time of the maximum local intensity. The power then increases back toward pre-flare levels, with thelower frequencies recovering first. The result is the appearance of a “V”-shaped feature in the color plots of Figure 5. For eachsubregion, the lowest frequency band (1–2 mHz) shows the greatest power.The overall temporal extent of the “V”-shaped decrease in power is shortest for subregion AR3 4, which might be a conse-quence of the narrow spatial width of the dark lane in that region. The suppression in power in AR3 3 (Figure 5(a)) is wider intime, perhaps due to the larger width of the dark lane in that area. AR3 7 (Figure 5(c)), free of dark lanes, shows the longest time0 M
ONSUE ET AL .period of power suppression. Figure 5(d) shows the total average intensity variations for each of the three regions along with the
GOES
X-ray flux.
EMPORAL E VOLUTION OF C HROMOSPHERIC O SCILLATIONS IN F LARING R EGIONS AR3 Location 3,4 & 7 (Inner Flare Region) A R F r e qu e n c y ( m H z ) L og P o w e r (a) A R F r e qu e n c y ( m H z ) (b) A R F r e qu e n c y ( m H z ) (c) A v e r a g e I n t e n s it y (d) GONG Average Pixel Intensity - AR3 Location 3,4,7 5000600070008000900010000
AR3_3AR3_4AR3_7X-ray Flux
Figure 5.
Time-frequency power plots of the three subregions AR3 3 (a), AR3 4 (b) and AR3 7 (c) along with the corresponding averageintensity (d). The
GOES
X-ray flux is scaled as a reference for the overall solar flare event. The red dashed line indicates the time of flaremaximum in the time series at 39 minutes.
Outer Flaring Regions –
Locations 2 and 5 in AR3
ONSUE ET AL .Regions AR3 2 and AR3 5 both lie on the outer edge of the dark lanes, as shown in Figure 4. The overall power for these tworegions are in the middle range of log power, 2.0–3.0. The corresponding time-frequency plots are shown in Figure 6.These plots have a substantially different qualitative nature compared to those in Figure 5. Here, there is no sign of the “V”shape in these time-frequency images. For region AR3 2 (Figure 6(a)) there is an overall increase in power for the 4–5 mHzfrequency band, with logarithmic values in the range of 2.6 to 3.0 - also observed in region AR3 1 (Figure 7(a)). A decreasein power for frequencies above 5 mHz (Figure 6(a)) is present at the start of the time series and persists to around 35 minutes.There appears to be a trend of increasing power late in the time series in the frequency band of 1–2 mHz. For region AR3 5(Figure 6(b)) there are three areas of power suppression within the 1–2 mHz band. A period of fluctuating suppression is alsoobserved in the 5–6 mHz band.Figure 6 shows that these two subregions have substantially lower levels of average intensity variations than those in Fig-ure 5. Thus suggests that rapid changes in the average intensity are at least partially responsible for the power variations in theoscillations.
EMPORAL E VOLUTION OF C HROMOSPHERIC O SCILLATIONS IN F LARING R EGIONS AR3 Location 2&5 (Outer Flare Region) A R F r e qu e n c y ( m H z ) L og P o w e r (a) A R F r e qu e n c y ( m H z ) (b) A v e r a g e I n t e n s it y (c) GONG Average Pixel Intensity - AR3 Location 2&5 350040004500500055006000
AR3_2AR3_5X-ray Flux
Figure 6.
Analysis of the two outer flare regions, AR3 2 (a) and AR3 5 (b), along with the corresponding pixel intensity and
GOES
X-ray flux(c). The
GOES
X-ray flux is scaled as a reference for the overall solar flare event. The red dashed line indicates the solar flare event in the timeseries at 39 minutes.
Quiescent Regions –
Locations 1 and 6 in AR3
ONSUE ET AL .Regions AR3 1 and AR3 6 both lie on the outer edge of the solar flare and in very dark outer regions. The overall logarithmicpower for these two regions is in the lowest range of 0.6–2.0, Figure 7.These areas also do not show the “V”-shape suppression feature. For region AR3 1 (Figure 7(a)) there is an overall increasein the power in the frequency band of 1–6 mHz. Some power suppression is apparent in the time period of about 8–42 minutes.The overall power has a maximum in the 5–6 mHz band.Region AR3 6 (Figure 7(b)) shows an overall suppression of power in all frequency bands that starts at the time of maximumX-ray intensity. This region also shows a substantially higher power level at frequencies of 1–3 mHz than at 6–8 mHz.As in Figure 6(c), the curves of average intensity curves in Figure 7(c) do not show much relative variation. This is consistentwith the hypothesis that the “V” shape is related to the presence of strongly varying intensity.
EMPORAL E VOLUTION OF C HROMOSPHERIC O SCILLATIONS IN F LARING R EGIONS AR3 Location 1&6 (Quiescent Flare Region) A R F r e qu e n c y ( m H z ) L og P o w e r (a) A R F r e qu e n c y ( m H z ) (b) A v e r a g e I n t e n s it y (c) GONG Average Pixel Intensity - AR3 Location 1&6 380040004200440046004800
AR3_1AR3_6X-ray Flux
Figure 7.
Analysis of the two quiescent flare regions: AR3 1 (a) and AR3 6 (b), along with the corresponding pixel intensity and
GOES
X-rayflux (c). The
GOES
X-ray flux is scaled as a reference for the overall solar flare event. The red dashed line indicates the solar flare event in thetime series at 39 minutes.
AR1
ONSUE ET AL .Region AR1 has an area of approximately 55 (cid:48)(cid:48) × (cid:48)(cid:48) (Figure 8). As with AR3, we select seven subregions intended to probethe temporal evolution of the flaring event at various spatial locations relative to the center of the active region. Each of the sevensubregions samples a rectangular area with 3.2 (cid:48)(cid:48) per side. AR1 displays similar behavior to that of AR3 and so only the mostsignificant of results (Figure 9) will be discussed. We present the results for locations AR1 5, AR1 6 and AR1 7, the regionswhere the M1 flare on 2012 June 13 was the most intense. Y - R O I s i ze ( a r c s ec ) L og P o w e r AR1, 2-3mHz
AR1_1AR1_2 AR1_3AR1_4AR1_5AR1_6AR1_7
Figure 8.
AR1 has an area of approximately 55 (cid:48)(cid:48) × (cid:48)(cid:48) . There are seven subregions in AR1 for the M1 solar flare occurring on 2012 June 13.In this figure, the PMM frame depicted is in the time window of 32 to 92 minutes. EMPORAL E VOLUTION OF C HROMOSPHERIC O SCILLATIONS IN F LARING R EGIONS
ONSUE ET AL . AR1 Location 5,6 & 7 A R F r e qu e n c y ( m H z ) L og P o w e r (a) A R F r e qu e n c y ( m H z ) (c) A R F r e qu e n c y ( m H z ) (b) A v e r a g e I n t e n s it y (d) GONG Average Pixel Intensity - AR1 Location 5,6,7 3000400050006000
AR1_5AR1_6AR1_7X-ray Flux
Figure 9.
Time-frequency power plots for regions AR1 5, AR1 6, and AR1 7, showing the suppression of power in plots (a)–(c), correlatingwith the peak intensity (d). All three regions exhibit a peak in power around 3.75 at the lowest frequency bands of 1–2 mHz and then graduallydecreasing as the frequency bands increase. The
GOES
X-ray flux is scaled as a reference for the overall solar flare event in the time series at36 minutes.
AR2
EMPORAL E VOLUTION OF C HROMOSPHERIC O SCILLATIONS IN F LARING R EGIONS (cid:48)(cid:48) × (cid:48)(cid:48) in size (Figure 10). This was the smaller sunspot region involved in the M1 solar flareon 2012 June 13 (Figure 1). Once again, we defined several subregions in order to probe the temporal behavior of the activity atvarious spatial positions relative to the visual center of the activity. Six subregions of 3.2 (cid:48)(cid:48) in size were analyzed. AR2 displayedresults similar to those of both AR1 and AR3. Here we present an analysis of two of the subregions close to the center of theactive region (Figure 11, (a) and (b)), and a control quiet subregion for comparison (Figure 11(c)). Y - R O I s i ze ( a r c s ec ) L og P o w e r AR2, 2-3mHz
AR2_1AR2_2AR2_3 AR2_4AR2_5AR2_6
Figure 10.
AR2 has a rectangular area of approximately 66 (cid:48)(cid:48) × (cid:48)(cid:48) . There are six subregions in total in AR2 for the M1 solar flare on 2012June 13. In this figure, the PMM frame depicted is in the time window of 32–92 minutes. Region AR2 provides observations of the oscillatory power behavior when the average intensity is slowly varying. In Fig-ure 11(d) the intensity for AR2 3 is slowly increasing, while it is decreasing for AR2 4. The corresponding time-frequencyimages show power suppression slowly increasing for AR2 3 and decreasing for AR2 4, suggesting that the rate of change of0 M
ONSUE ET AL .the intensity is related to the rate of change of the oscillatory power. Again the lower frequency bands below 3 mHz show higherpower levels than the bands above 3 mHz. The relatively low signal-to-noise ratio in these panels can be increased by enlargingthe size of the selected subregions.Region AR2 5 (Figure 10) is a quiet region that provides an observation of the acoustic power characteristics outside flaringregions. This control region shows a constant oscillatory power with no systematic temporal changes within the frequency bands(Figure 11 (c)). While it does show higher power at lower frequencies, the enhancement is much lower than that seen in the flareregions. Quasi-periodic fluctuations are also seen with a periods of ∼ ∼ EMPORAL E VOLUTION OF C HROMOSPHERIC O SCILLATIONS IN F LARING R EGIONS AR2 Location 3,4 & 5 A R F r e qu e n c y ( m H z ) L og P o w e r (a) A R F r e qu e n c y ( m H z ) (b) A R F r e qu e n c y ( m H z ) (c) A v e r a g e I n t e n s it y (d) GONG Average Pixel Intensity - AR2 Location 3,4,5 20002500300035004000
AR2_3AR2_4AR2_5X-ray Flux
Figure 11.
Acoustic power observations for regions AR2 3, AR2 4, and AR2 5, showing the suppression of power in plots (a) and (b) and aquiet region for reference (c). The
GOES
X-ray flux is scaled as a reference for the overall solar flare event in the time series at 36 minutes.5.
DISCUSSION AND CONCLUSIONS2 M
ONSUE ET AL .The results in this paper demonstrate that H α observations of chromospheric oscillations in the p-mode band can provideinformation about the physical processes occurring in flaring regions. In particular, variations in the oscillatory power as afunction of frequency, spatial position, and time can be used to probe energy transport at different heights within a flare.Figures 5 and 9 show a suppression of power that first migrates in time from high to low frequencies in a flare, with a subsequentrestoration of the power starting at low frequencies and progressing back to high frequencies. This produces a “V”-shaped featurein the images of the power as a function of frequency and time. The shape can be understood as a consequence of the nature ofthe observations, the behavior of the H α spectral line during a flare, and the height dependence of the frequency of the maximumoscillation amplitude. Wang et al. (2000), with their high-cadence H α observations from the Big Bear Solar Observatory, foundhigh-frequency fluctuations that correlate with HXR elementary bursts. These hard X-ray emissions could be signatures of sitesof fine structures where individual magnetic reconnection processes are taking place (Wang et al. 2000).One of the earliest results of studies of solar oscillations showed that the frequency at which the waves reach their maximumamplitude increases with height in the solar atmosphere (Leighton et al. 1962). In the photosphere, the maximum amplitudeoccurs at periods around five minutes (frequency near 3.3 mHz), while in the chromosphere the maximum occurs at periods ofthree minutes (frequency near 5.5 mHz). It is also known that the wings of a spectral line are formed at lower heights in the solaratmosphere than the core of the line, thus observations of oscillations obtained in the wing of a spectral line will be dominatedby lower-frequency power than observations in the core of the line.The observations discussed here are obtained with a filter that is centered on the wavelength of the H α line core in the quietSun. During the course of the flare, the motion of the plasma will change the wavelength of the line due to the Doppler effect,so that the filter bandpass will admit a higher proportion of light from the wings of the spectral line rather than from the core.Since the wings of the line are formed at lower heights in the solar atmosphere, and since the peak amplitude of the p-modesoccurs at lower frequencies at lower heights, the net effect is to reduce power at high frequencies. This reduction moves tolower frequencies as the flare progresses and the spectral line is increasingly Doppler-shifted. The overall observed intensity alsoincreases as the brighter wings contribute a larger portion of the signal. As the flare energy decreases, the solar plasma motionsdie out and the spectral line core moves back toward the center of the filter bandpass, restoring the visibility of the high-frequencypower and decreasing the overall intensity. A plasma velocity of 5 km s -1 , easily created in a flare, would move the line core by0.1 ˚A , which is a substantial fraction of the 0.6 ˚A bandpass of the GONG filter.The dark lanes in wave power, also observed by Jackiewiez and Balasubramaniam (2013), could be where the magnetic fieldabsorbs or scatters the acoustic waves. In the photosphere, sunspots are known to be areas of suppressed acoustic mode power(Braun et al. 1987). The appearance of the dark lanes depends on frequency, as seen in Figure 3, which may provide informationon the structure of the magnetic field as well as aspects of energy transport during the flare. In addition, the time-frequency mapsof subregions located on dark lanes show diverse behavior (e.g. AR3 3 and AR3 2) further suggesting that there is a variationin the underlying magnetic field. Several deductions can be made about the dark appearance of the lanes, which could indicatethat energy is being removed from the observed wave frequencies and perhaps converted into the thermal energy of the flare, orscattered into other wave modes with frequencies higher than 8 mHz, or absorbed by the magnetic field in the flare, or dampenedby magnetic reconnection. We believe that the correct explanation is that the wave energy is being converted into thermal energy,due to the simultaneous increase in both GOES
X-ray flux and H α intensity. These possibilities can be investigated by applyingthe PMM technique to simultaneous magnetograms acquired by GONG. Furthermore, there is a trade-off between signal-to-noiseratio and spatial resolution. Larger subregions increase the signal-to-noise ratio but decrease the spatial resolution. In this paperwe chose to have higher spatial resolution in order to investigate oscillations within the narrow dark lanes.Generally, there is a tendency for an excess of power at low frequencies below 2 mHz compared to higher frequencies. Thisexcess can be as much a factor of 30 for the regions in Figure 5, but it is also present in the quiet region in Figure 11 at a muchlower level (a factor of about 4). Since these are ground-based intensity observations, it is quite possible that some of this excessis caused by fluctuations in the Earth’s atmospheric transparency. However, the marked increase in the flaring regions suggeststhat low-frequency power is enhanced during a flare. If this low-frequency power excess is a feature of strong flares, it may arisefrom an instability in the chromosphere and provide an early warning of the flare onset.This pilot project demonstrates that the application of PMMs to H α intensity observations opens up a number of new avenuesto explore the physical processes in flares. The temporal and spatial variations of acoustic wave power show intriguing featuresthat contain information about the energy transport and magnetic field variations as a function of height within flaring regions.There are several paths to follow that will further develop the method. The most informative step is the comparison of theresults with the GONG magnetograms. The correlation of changes in oscillatory power with the characteristics of the magneticfield should provide additional information on the underlying physical processes. In addition, the PMM method can also beapplied to the magnetograms since they are simultaneously observed at a cadence of once per minute. Additional steps will be EMPORAL E VOLUTION OF C HROMOSPHERIC O SCILLATIONS IN F LARING R EGIONS α network, operated by The National Solar Observatory (NSO) and The Association of Universities for Research inAstronomy (AURA), and which was originally commissioned by The Air Force Weather Agency (AFWA). T.M. acknowledgessupport from the Fisk-Vanderbilt Masters-to-PhD Bridge Program, including specifically funding support through NSF PAAREgrant AST-1358862 and a Harriett Jenkins Graduate Fellowship from NASA.REFERENCESnetwork, operated by The National Solar Observatory (NSO) and The Association of Universities for Research inAstronomy (AURA), and which was originally commissioned by The Air Force Weather Agency (AFWA). T.M. acknowledgessupport from the Fisk-Vanderbilt Masters-to-PhD Bridge Program, including specifically funding support through NSF PAAREgrant AST-1358862 and a Harriett Jenkins Graduate Fellowship from NASA.REFERENCES