Abstract:
Solar radio bursts are unambiguous and sensitive tracers of non-thermal electrons in the corona. These non-thermal electrons owe their origin to episodes of solar activity and lead to a wide variety of radio emissions, making observations of solar radio bursts a very useful probe of solar activity. These emissions are usually bright enough to outshine the quiescent solar emission. The well known classes of solar radio bursts differ dramatically in their appearance in the time-frequency plane and carry useful information about solar activity, especially when combined with observations in higher frequency bands (EUV and X-ray). Spectrographs have been the most commonly used tool for observing solar radio bursts and have played a pivotal role in building our current understanding of these phenomena.
With notably few exceptions, these radio spectrographs have been operating in what is traditionally referred to as low radio frequency bands (< 500 MHz), and the nature of these burst emissions in higher parts of the radio band are yet to be studied in similar detail. This project focuses on studying the archival data from a solar radio observatory operated by the National Institute of Information and Communications Technology (NICT), Japan, namely the YAMAGAWA radio spectrograph, which has been operating since 2016 to the present times covering the band from 70-9000 MHz. The instrument cover bands well beyond the traditional low frequency bands, observe routinely from sunrise to sunset with comparatively few data gaps, provide good quality data including L and R polarizations and span a period of 10 years.
Algorithms for the detection and the classification of bursts was implemented on the data provided by the instrument. The data required extensive pre-processing before the implementation of detection algorithms. The study also explore feature detection and extraction algorithms using contours and other algorithms for a robust detection of solar signal. Further more two machine learning models were implemented on the data to detect bursts - Random Forest Classifier and YOLO. The insights gained from this study make future investigations more practical and impactful survey of solar radio bursts in the YAMAGAWA data; the algorithms heuristics used in this study are transferable to other telescopes.
