Author: Connor Macrae & Sergei Zharkov at the University of Hull.
Sunquakes, predicted in 1972 by Wolff, and first observed in 1998 , are an impulse-driven release of acoustic energy, observed as a local perturbation to photospheric surface oscillations. The strongest sunquakes can be observed as a burst of radially emanating ripples from a point source up to 30 minutes after the flare impulse.
Acoustic waves travel through the solar interior, surfacing in the vicinity of their upper turning points. Detection of sunquakes carried out by using helioseismic methods, such as time-distance diagram analysis and acoustic holography [2,3]. Time-distance analysis is a more observationally direct approach, however, it requires good prior knowledge of source location. Acoustic holography uses a model of the solar interior to estimate acoustic sources and sinks from observed Dopplergram series. Sunquakes are often indicated by compact and bright source kernels of ‘egression power’ which show the origin of the acoustic emission. However, it has recently been shown that these kernels are not always compact , so that by searching for compact sources we might be missing events.
Recent Seismic Surveys
Past surveys looking for seismic activity in coincidence with flares have been limited by data coverage and instrument resolution. For example, surveys in Solar Cycle 23 relied on SOHO/MDI observations . With higher resolution observations with SDO/HMI a greater number of detections have been made. A recent study  finds that of 41 M- and X-class flares exhibiting a disturbance in the photospheric Doppler signal (a Doppler transient), 18 are seismically active – 2 of which were X-class. Furthermore, each seismic detection had a corresponding white-light (WL) enhancement. In that study, detection of source kernels through acoustic holography was carried out via visual inspection, requiring egression power kernels to be sharp and clear.
We develop a new and improved method of detection based on acoustic holography statistical analysis, and survey X-class flares from the current cycle. Of the 44 X-class flares detected by GOES, we narrow our selection to events away from the limb, in order to avoid degradation of data during remapping.
Method of Detection
We carry out acoustic holography computation on running difference velocity Dopplergrams (SDO/HMI 45s cadence). The resultant egression power datacubes are processed to test for significant seismic sources. To rule out false detections due to stochastic noise we require co-spatial and co-temporal signals in multiple frequencies – for full details of statistical testing employed here, as well as a treatment of potential false signals, see . Both the statistical testing and checking for signal overlap is carried out through automated code. This greatly improves upon previous visual inspection methods, as it is faster and is able to detect very weak acoustic signals that still represent significant emission above background noise (especially in acoustically damped magnetised regions i.e. penumbra).
An example detection is seen in Figure 1 (here, for the 2013-11-10 X1.1 event), where we see the smoothed egression power signal for a detected source peak above a defined threshold. Figure 2 highlights the location of the detection in NOAA 11890 intensity continuum and LOS magnetogram (panels a and b respectively) snapshots – We clearly see detected signal overlap in the 6, 7 and 8mHz bands. The saturated 6mHz egression power map (panel c) shows our acoustic kernel location (see central contour). In fact, for this event we find an observable surface ripple – The animation below displays unprocessed running difference Dopplergrams where we can observe the flaring transient (lower box) followed by the photospheric ripple propagating upwards (upper box).
We identify 12 seismically active X-class flares from our sample of 18 – see Table 1. For each active event we have found egression power signals that exceed a defined threshold in the 6mHz band, and co-spatially in at least one other frequency band. Each seismically active event has Doppler and magnetogram transients during flare impulse, in addition to co-spatial, enhanced WL emission (Seismically inactive flare #7 displays enhanced WL emission).
The relationship between seismic and white light emission we find is in agreement with previous studies , and provides an indication of a causal relationship between localised transient acoustic emission and flare white-light heating . Similar to the already known 2014-03-29 event , we find a weak, non-compact acoustic source for the well studied 2011-09-06 X2.1 flare .
We find that 12 X-Class flares of our sample of 18 are seismically active. This is in excess of the occurrence rate seen in previous studies . This can be explained by both improvement in data quality and more reliable detection methods. Here, our statistical approach to identifying acoustic kernels yields weak and non-compact seismic signatures that can be confirmed with time-distance analysis. Advances in statistical methods used herein represent significant progress towards full automation of reliable detection. This will allow for expansion of our sample to include M- and C-class flares – providing a more reliable understanding of sunquake occurrence through the current solar cycle. Full details of our work will soon appear .
-  Kosovichev, A., Zharkova, V., (1998), Nature, 393:317
-  Buitrago-Casas, J.C., et al., (2015), Sol. Phys., 290:3151
-  Zharkov, S., Green, L., Matthews, S.A., Zharkova, V., (2013), Sol. Phys., 284:315
-  Judge, P.G., Kleint, L., Donea, A., Dakla, A.S., and Fletcher, L., (2014), ApJ., 796:85
-  Donea, A., (2011), Space Sci. Rev., 158:451
-  Macrae, C., et al., (2017), manuscript in prep.
-  Lindsey, C., and Donea, A., (2008), Sol. Phys., 251:627
-  Liu, C., et al., (2014), ApJ., 795:128