Comparative analysis of the quality of prediction for fluences of relativistic electrons of the outer radiation belt of the Earth at different phases of the solar activity cycleтезисы доклада Электронная публикация Тезисы

Дата последнего поиска статьи во внешних источниках: 9 марта 2017 г.

Работа с тезисами доклада


[1] Myagkova I. N., Dolenko S. A. Comparative analysis of the quality of prediction for fluences of relativistic electrons of the outer radiation belt of the earth at different phases of the solar activity cycle // 11th International Conference “PROBLEMS OF GEOCOSMOS”. — St. Petersburg, 2016. — P. 79–79. The forecast of radiation environment at geosynchronous orbit (GEO) near the boundary of the Outer Earth’s Radiation Belt (OERB) is very important due to the large number of satellites populating this region. Relativistic electrons of the outer ERB are even called "killer electrons" since they can damage the electronic components of spacecraft. The flux of the electrons with the energy >2 MeV at GEO orbits is very unstable within time intervals about days and even hours. It shows a strong temporal dependence on epoch of solar activity (SA) relative to the onset of geomagnetic disturbances (AE-min and AE-max models). Any generally accepted theory that explains all the features of the experimentally measured behavior of outer ERB is not created, so the statistical methods are usually less dependent on the selected physical models, it is useful to have an effective forecasting model based on statistical methods. Artificial neural networks (ANN) model that would work as good or better than any other statistical model. The intensity of the OERB depends on the values of the components of interplanetary magnetic field, on solar wind, and on geomagnetic storms. It is known that all of these values are strongly dependent on SA cycle phase. We have already created an ANN prediction model for hourly average values of the electrons of OERB at GEO with prediction horizon from 1 to 12 h (R2 =0.98 and 0.86) using data without dividing it by the level of SA. The hypothesis that separate networks trained for different phases of SA would be more successful than the single network using the whole data for the overall cycle has been checked in presented study. This study was supported by the Russian Science Foundation, grant № 16- 17-00098.

Публикация в формате сохранить в файл сохранить в файл сохранить в файл сохранить в файл сохранить в файл сохранить в файл скрыть