Cosmic ray variation modeling according to neutron monitors data and detection of their intensity ground enhancement precursors

  1. Institute of Cosmophysical Researches and Radio Wave Propagation FEB RAS, Russia
  2. Kamchatka State Technical University, Russia
  3. Pushkov Institute of Terrestrial Magnetism, Ionosphere and Radio Waves Propagation RAS, Russia

   The paper proposes a way to model the data of cosmic ray time variation,
   that is based on the combination of wavelet transform and multilayer
   feedforward neural networks, allowing to describe the characteristic
   variation and to detect some peculiarities formed before strong increases
   in the ground level intensity. Based on wavelet transform, detection of
   characteristic components of cosmic ray variations is carried out and
   noise is suppressed. Selecting the best basic wavelet function and making
   an approximation, which provides the smallest error, the characteristic
   components are determined. The resulting characteristic components are
   modeled via neural networks. On the basis of the analysis of neural
   network error vector, precursors of strong increases in cosmic ray ground
   level intensity are identified. At the modeling stage, data of Moscow and
   Apatity neutron monitor stations for the period 2000-2005 were used. The
   modeling confirmed the efficiency of the proposed method and revealed the
   precursors of cosmic ray ground level enhancement.