A method for the automated detection of solar radio bursts in dynamic spectra

Abstract

The variability of the solar corona, including flares and coronal mass ejections, affects the space environment of the Earth (heating and ionization of the atmosphere, magnetic field disturbances, and bombardment by high-energy particles). Electromagnetic emissions are the first signatures of a solar eruptive event which by modifying the electron density in the ionosphere may affect airborne technology and radio communications systems. In this paper, we present a new method to detect automatically radio bursts using data from the Nançay Decametre Array (NDA) in the band 10 MHz–80 MHz. This method starts with eliminating unwanted signals (Radio-Frequency Interference, RFI and Calibration signals) by analyzing the dynamic spectrum of the signal recorded in time. Then, a gradient median filter is applied to smooth and to reduce the variability of the signal. After denoising the signal, an automated solar radio burst detection system is applied. This system is based on a sequential procedure with adaptive constant-false-alarm rate (CFAR like detector) aimed to extract the spectra of major solar bursts. To this end, a semi-automatic software package is also developed to create a data base of all possible events (type II, III, IV or other) that could be detected and used for our performance assessment.