Identifying Acoustic Wave Sources on the Sun. I. Two-dimensional Waves in a Simulated Photosphere

The solar acoustic oscillations are likely stochastically excited by convective dynamics in the solar photosphere, though few direct observations of individual source events have been made and their detailed characteristics are still unknown. Wave source identification requires measurements that can reliably discriminate the local wave signal from the background convective motions and resonant modal power. This is quite challenging as these noise contributions have amplitudes several orders of magnitude greater than the sources and the propagating wave fields they induce. In this paper, we employ a high-temporal-frequency filter to identify sites of acoustic emission in a radiative magnetohydrodynamic simulation. The properties of the filter were determined from a convolutional neural network trained to identify the two-dimensional acoustic Green's function response of the atmosphere, but once defined, it can be directly applied to an image time series to extract the signal of local wave excitation, bypassing the need for the original neural network. Using the filter developed, we have uncovered previously unknown properties of the acoustic emission process. In the simulation, acoustic events are found to be clustered at mesogranular scales, with peak emission quite deep, about 500 km below the photosphere, and sites of very strong emission can result from the interaction of two supersonic downflows that merge at that depth. We suggest that the method developed, when applied to high-resolution high-cadence observations, such as those forthcoming with the Daniel K. Inouye Solar Telescope, will have important applications in chromospheric wave studies and may lead to new investigations in high-resolution local helioseismology.

Automated summary

This paper introduces a new method to identify and study the emission locations and characteristics of solar acoustic oscillations. By combining a high-temporal-frequency filter with a convolutional neural network, it reveals the clustering and depth properties of acoustic events at mesogranular scales. This method can be applied in future high-resolution high-cadence observations, potentially leading to significant breakthroughs in chromospheric wave studies and high-resolution local helioseismology.