Current Proceedings on Technology
Yazarlar: Inna Petrova, Ruslan Latipov
Konular:-
Anahtar Kelimeler:Cluster analysis,Hierarchical method,Total electron content,Ionosphere
Özet: Total Electron Content (TEC) maps provide great capabilities for ionosphere researches. However, difficulty for the data analysis is associated with significant variability of spatial TEC distributions. The purpose of this study is the analysis of capability for classification of global TEC distributions in different seasons, geophysical and geomagnetic conditions. Such classification allows determining a typical kind of spatial distributions for various conditions. Comparison of the experimental distribution with a typical patterns permits to select anomalies in the spatial TEC distribution. For construction of classification we use a hierarchical method of cluster analysis and methods based on neural networks (Kohonen self-organizing maps). Hierarchical method is more effective but requires great memory capacity for data saving. We have offered modification of this method allowing to reduce a demanded memory size.