Space weather effects in Odesa magnetic anomaly
Résumé
Analysis of large time series of geomagnetic data obtained at the Odesa and Lviv geomagnetic observatories in Ukraine reveals the difficulty of determining geomagnetic (including sudden) storms. Calculation of traditionally used K-indexes has some disadvantages: they provide no link to the physical features of magnetic storms’ development, do not account for their duration, and evaluate geomagnetic activity only in three-hour intervals. Here we considered the use of neural network clustering and wavelet analysis for detecting geomagnetic storms in large data series over one or several months. We used a self-learning neural network (Kohonen Map), attracting data of the INTERMAGNET catalog (Lviv magnetic station) for June and July 2023. Automatic neural network clustering was found to effectively single out from a large array of data exactly the events with bursts of geomagnetic variations (on components X, Y, Z of geomagnetic field), which most often correspond to phases of magnetic storms’ development. Additionally, the wavelet method for constructing global wavelet power spectra with frequency averaging (spectral power distribution over time) was considered. High-amplitude bursts in global wavelet spectra also usually correspond to bursts in geomagnetic variations during storms. This is a relatively simple and good visual method for quickly assessing the occurrence and duration of magnetic storms. Disadvantages leading to incorrect determination of the occurrence of some magnetic storms were also considered. The described methodology has become useful for the work on compiling a ``catalog of geomagnetic storms'', which is carried out using magnetometric monitoring data in the area of the Odesa regional magnetic anomaly (Ukraine).
Lire l’article

PDF de 18 pages, tel qu’imprimé dans la revue, pages 299 à 316. En libre accès.
Pas encore de consultation enregistrée.