Abstract:
This paper presents a comprehensive approach to solving the problem of automatic recognition of groups of whistling atmospherics (whistlers) in the time-frequency spectra (spectrograms) of VLF radio signals. Such radio signals are generated by atmospheric electrical discharges passing through the magnetospheric waveguide and serve as natural markers of the Earth's magnetosphere state. Object of study: spectrograms of VLF radio signals containing whistling atmospherics. Subject of study: algorithms for the automatic recognition and identification of groups of whistling atmospherics on time-frequency spectrograms. The proposed method involves a multi-stage algorithm for processing the source signal. The first stage involves signal filtering, which consists of two parts: modified median filtering and selection of significant samples. Next, a transition to a new coordinate system is performed, which allows for the transformation and “straightening”of the curvilinear patterns of whistlers. This transformation significantly simplifies subsequent analysis. The next stage is the recognition of a single whistler or multiple whistlers in the signal fragment under consideration (this stage was addressed by the authors in previous work). The final stage is the search for groups of whistlers whose straightened patterns intersect at a single point on the time axis. For testing the final stage, signal fragments of two types were generated: an ideal group consisting of two straight lines (whistlers) that converge at a single point; two groups of straight lines (whistlers) approximating real-world whistler propagation conditions (by adding Gaussian noise and reducing the intensity of the second signal in the group). The described algorithm automates the process of identifying whistler groups, which enhances the objectivity and speed of analysis compared to visual methods.
Fulltext:
PDF file (2381 kB)