Physics of auroral phenomena : proceedings of the 34th Annual seminar, Apatity, 01 - 04 March, 2011 / [ed.: A. G. Yahnin, A. A. Mochalov]. - Апатиты : Издательство Кольского научного центра РАН, 2011. - 231 с. : ил.

“Physics ofAuroral Phenomena ", Proc. XXXIVAnnual Seminar, Apatity, pp. 106-109 2011 © Kola Science Centre, Russian Academy of Science, 2011 © Polar Geophysical Institute C LA SS IF ICAT ION OF SPACE W EA TH ER CO M P L EX E S T A K IN G INTO ACCOUNT CH ARAC T ER IST IC S OF PERTURB ING SO L A R S T R E AM AND ITS GEOM AGNET IC D ISTURBANCE P A R AM E T E R S N.A. Barkhatov1, A.E. Levitin2, E.A. Revunova1 1Nizhniy Novgorod State Pedagogical University, Nizhniy Novgorod, Russia 2Pushkov Institute o f Terrestrial Magnetism, Ionosphere and Radio Wave Propagation, Troitsk, Russia Abstract. Work is devoted to development of neural network classification o f space weather complexes including information about type of solar source, characteristics of perturbing solar streams (solar wind parameters and interplanetary magnetic field components) and parameters of its geomagnetic manifestations in the form of Dst- index dynamics. As result of the executed numerical experiments the basic complexes of space weather, establishing connection between types of solar streams and geomagnetic disturbances of various intensities, are allocated. The allocated complexes will allow carrying out the analysis and specification of influence on Earth magnetosphere of plasma streams from various solar sources, and as forecasting of geomagnetic conditions. 1. Introduction Classification of space weather complexes has direct relation to an establishment of causal link between solar activity and geomagnetic disturbances. The basic difficulty in its creation is necessity to include the information about characteristics of perturbing solar streams and parameters of their subsequent geomagnetic manifestations. Now as result of carrying out of numerous researches [Tsurutani et a l, 1995; Ivanov, 1996; Yermolaev, et a l, 2009] the basic types of solar plasma streams and their prominent features are established. To them carry: heliospheric current sheath (HCS), stream from coronal streamers, filament streams, high-speed streams from coronal holes (HSS) and corotation interaction region before them (CIR), and also coronal mass ejections (CME). At movement from Sun CMEs often take the form of closed structures with original behavior of plasma and magnetic field in them - magnetic clouds (MC). Thanks to specific spiral distribution of magnetic field in MC they are one of the most geoeffective plasma structures in solar wind. Because velocities CME/MC usually above velocity of quiet solar wind before them Shock and Sheath - region with sharp change of all interplanetary medium parameters is often formed. Each type of solar stream is characterized by specific dynamic of solar wind parameters (PSW) and interplanetary magnetic field (IMF) components that allows carrying out their identification in interplanetary space. Despite exhaustive data on solar wind streams the uniform standard classification including as type of solar source, characteristic of plasma stream, and parameters of its geomagnetic manifestations till now is not created. The majority of researches of plasma streams are carried out without attraction of data on their geomagnetic manifestations [Ivanov, 1996; Vennerstroem, 2001; Echer and Gonzalez, 2004]. Attempt to execute neural network classification of space weather complexes taking into account specified above factors has been undertaken in work [Barkhatov, et al, 2006]. The neural network approach for decision of classification problems is perspective. Neural network approach makes it possible division of input images into the set number o f classes on the basis of nonlinear correlation processing of experimental data and to state quantitative estimation o f results. In the present work as classification tool the neural network on which input characteristics of perturbing solar stream - solar wind density, solar wind velocity, average value of interplanetary magnetic field vector and its components was used. However here the geomagnetic storms divided on intensity of Dst-index on weak, moderated and strong were considered only. The basic condition of successful work of classification neural network is necessity to submit on input of equal duration event. In work [Barkhatov, et al, 2006] this problem was solved by artificial prolongation of short magnetic storms till the sizes of the longest. Besides, in work the big values number in an input signal that complicates work of neural network was used. All it should be reflected in quality of received results. In the present work contains the development of the neural network method for classification o f space weather complexes including the information on characteristics of perturbing solar stream and its geomagnetic manifestations not limited type and duration of geomagnetic disturbance. Here the way of representation of input data, allowed to reduce values number in them, is changed. Also the set of input parameters at the expense of addition southern Bs components IMF, temperatures, dynamic pressure and electric field of solar wind is expanded. In basis of carried out classification of space weather complexes is put on the analysis of communication phases of sudden commencement and main phases of geomagnetic storms with dynamics PSW and IMF perturbing stream As it is known, what exactly the most intensive physical processes in magnetosphere, connected with arrival the perturbing solar stream to the Earth [Gonzalez et a l, 1994]. 2. The analysis of geomagnetic disturbances, establishment of their solar sources and corresponding types of plasma streams Classification of space weather complexes was carried out on hour data o f satellite system OMNI about PSW and 106