Analysis of the dynamics of cosmic ray variations and detection of Forbush effects based on the machine learning method

Remote sensing of the Earth (ERS) using space, aviation and ground-based technology equipped with various types of recording equipment is an important basis for studying the Earth's surface. Neutron monitors, which are one of the ERS tools, are detectors that record the intensity of cosmic ray variations. Data on cosmic ray variations are currently actively used in fundamental research in the field of solar-terrestrial physics and in applied problems of forecasting various natural phenomena and events [1,2]. A relevant area of ​​research, in particular, is the operational monitoring of wave processes associated with the arrival of high-speed solar wind flows to the Earth, associated with coronal mass ejections preceding the onset of geomagnetic storms that can lead to radiation hazards for astronauts, crews and passengers of high-altitude aircraft, as well as the loss of satellites and failure of space and ground equipment [2-4]. Therefore, methods aimed at the timely detection of anomalous manifestations in cosmic ray variations are of particular practical importance. The paper studies a new method for analyzing cosmic ray variations and detecting Forbush effects, based on a combination of machine learning methods with elements of statistical decision theory [5]. The method is based on cognitive rules for choosing a decision on the state of data, constructed by the authors, based on displaying data in wavelet space and allowing one to automatically obtain near-optimal estimates of the characteristics of the natural process under study. The numerical algorithms for implementing the method proposed in the paper include adaptation tools and allow one to interactively detect and evaluate anomalous changes in data on cosmic ray variations, indicating the occurrence of Forbush effects. The study demonstrated the effectiveness of the developed method and algorithms for detecting anomalous changes in the rate of cosmic ray arrival to Earth, observed several hours before the onset of moderate and extreme geomagnetic storms (levels G3-G5) in the period 2024-2025.
The authors express their gratitude to the institutes that support the neutron monitor stations [https://www.nmdb.eu, http://spaceweather.izmiran.ru) and data on the state of the interplanetary medium (http://ipg.geospace.ru, https://omniweb.gsfc.nasa.gov/ow.html) and the magnetosphere [https://wdc.kugi.kyoto-u.ac.jp], which were used in the work.
The work was supported by IKIR FEB RAS State Task (subject registration No. 124012300245-2).
References:
1. Getmanov V.G., Gvishiani A.D. et al. Early diagnostics of geomagnetic storms based on observations of space monitoring systems // Solar-Terrestrial Physics. 2019. Vol. 5. No. 1. P. 59-67.
2. Kuznetsov V.D. Space weather and risks of space activities // Space engineering and technology, 2014, No. 3 (6). Pp. 3–13.
3. Demyanov V.V., Yasyukevich Yu.V. Space weather: risk factors for global navigation satellite systems // Solar-Terrestrial Physics. 2021. Vol. 7. No. 2. Pp. 30-52.
4. Belov A.V., Villoresi D. et al. Influence of the space environment on the functioning of artificial earth satellites // Geomagnetism and Aeronomy. 2004. Vol. 44. No. 4. Pp. 502-510.

5. Mandrikova O., Mandrikova B. Hybrid model of natural time series with neural network component and adaptive nonlinear scheme: application for anomaly detection. Mathematics. 2024. 12. 1079. https://doi.org/10.3390/math12071079.