An Automated Detection Method for Solar Radio Type III Bursts Using Phase Clustering and Hierarchical Thresholding

Solar radio type III bursts are among the most common types of radio bursts in solar activities. They are characterized by features such as rapid frequency drift and serve as a precursor tracer for solar activity events like solar flares and coronal mass ejections. Existing identification methods face problems such as weak capability in detecting continuous bursts and insufficient learning with small samples, leading to high false detection rates and low recall rates. To address these shortcomings, this paper proposes an adaptive contour detection algorithm based on the phase grouping method. This method integrates gray gradient phase information with multi-level threshold filtering, fits the spectral contour of solar radio type III bursts through morphological operations, and realizes the automatic identification of the spectrum of solar radio type III bursts. Applied to the observation dataset from the Chengjiang Meter-wave Solar Radio Telescope of Yunnan Observatory, it has been verified that this method achieves a 93.5% recall rate in identifying solar radio type III bursts.