Physics of auroral phenomena : proceedings of the 37th Annual seminar, Apatity, 25 - 28 February, 2014 / [ed. board: A. G. Yahnin, N. V. Semenova]. - Апатиты : Изд-во Кольского научного центра РАН, 2014. - 125 с. : ил., табл.

CLASSIFICATION APPROACH TO THE ANALYSIS OF MAGNETIC DISTURBANCES IN AURORAL REGION AND THE MAGNETIC DISTURBANCE AT MIDDLE AND LOW LATITUDES 0 .M . Barkhatova1’2, N.A. Barkhatov2,1.A. Bazina2 1Nizhniy Novgorod State University o fArchitecture and Civil Engineering, Nizhniy Novgorod, Russia 2Nizhniy Novgorod State Pedagogical University, Nizhniy Novgorod, Russia Abstract. The degree of geomagnetic storms main phase intensity and duration on the presence or absence o f nonlinear connection between ring current components (symmetrical and asymmetrical parts) and the auroral electrojets is studied. For this purpose the classification analysis of 30 single-step geomagnetic storms main phases is made by Kohonen neural network. The intensity of considered current systems was estimated by the values of corresponding indices - SYM, ASY for symmetric and asymmetric parts of ring current, and the AU, AL for eastward and westward electrojets. On the base of classification under SYM, AU and SYM, AL parameters it was established that in the same class enter events corresponding to "classical" geomagnetic storm development with have small amplitude and duration. When it was classified by two indices, one of which is ASY index, in one class entered intense and long duration storms. 1. Introduction Recent studies devoted to magnetospheric and ionospheric current systems development during geomagnetic storm are demonstrated the connection between low-latitude disturbances caused by activation of ring current symmetric and asymmetric parts with disturbances in the auroral region [Grafe et. al., 1997; Kamide et. al., 1998; Huang and Foster, 2007]. For example, in paper [Sun andAkasofu, 2000 ] it was shown that during geomagnetic storms the ring current is saturated with ions 0+ which enters along field lines during magnetospheric substorms development. This means that ring current and auroral electrojets constitute a unified ionospheric-magnetospheric current system. In the papers [ Barkhatov et. al., 2008; Barkhatov et. al, 2007] the connection between indices which characterizing the development of the symmetric (SYM) and asymmetric (ASY) ring current parts with intensity indices of eastward (AU) and westward (AL) electrojets was searched on examples of specific geomagnetic storms. On the base o f linear correlations between specified indices at main phases intervals was established a connection between mid­ latitude and auroral current systems. But it should be noted that there is a view about an absence of direct connection between these currents [Grafe and Feldstein, 2000]. According to this assumption, current systems develops independently of each other and the occurrence of correlation between corresponding intensity indices is consequence of their common source. Thus the problem of current systems development and integration is still not folly investigated. The main purpose of this paper is the identification of common patterns in magnetospheric and ionospheric current systems dynamics that determine the presence or absence of nonlinear connection between ring currcnt elements and auroral electrojets for various intensity storms. For these aim the neural network approach is applied using Kohonen artificial neural network (ANN). At the same time the nonlinear connection between studied currents is established using Elman ANN algorithm with feedback and Fuzzy logic network. 2. Used data and research methods A study of nonlinear connection was conducted for the main phases of 30 single-step classical geomagnetic storms of various intensity (-300 < Dst < -50 nT) for pairs of indices SYM-AU, SYM-AL, ASY-AU and ASY-AL for the period from 2000 to 2003. The analyzed events selection held by Dst index. Since the nonlinear connection between mid-latitude and auroral geomagnetic activity indices have ambiguous character, their detection was performed using two neural networks types - ANN Elman algorithm with feedback and Fuzzy logic network. ANN training in each experiment was always performed anew at one of events and 29 remaining events consistently offered it as a test. In the learning process it has weights adjustment, i.e., ANN establishes the connection between SYM, ASY indices and AU, AL indices within the main phase of particular geomagnetic storm. The estimation of restoration quality of auroral electrojet intensity indices under ring current intensity indices was performed by calculation of linear correlation between the real and reconstructed sequences The restoration was considered satisfactory if the correlation coefficient is higher than 0.4. The presence of linear correlation shows that the connection between midlatitude and auroral current systems has the same character as in learning event. The absence of linear correlation indicates that a new event have fundamentally different nature “Physics o f Auroral Phenomena", Proc. XXXVII Annual Seminar, Apatity, pp. 6-9, 2014 © Kola Science Centre, Russian Academy of Science, 2014 Polar Geophysical Institute 6