Abstract
One of the goals of machine learning is to eliminate tedious and arduous repetitive work. The manual and semi-automatic classification of millions of hours of solar wind data from multiple missions can be replaced by automatic algorithms that can discover, in mountains of multi-dimensional data, the real differences in the solar wind properties. In this paper we present how unsupervised clustering techniques can be used to segregate different types of solar wind. We propose the use of advanced data reduction methods to pre-process the data, and we introduce the use of Self-Organizing Maps to visualize and interpret 14 years of ACE data. Finally, we show how these techniques can potentially be used to uncover hidden information, and how they compare with previous manual and automatic categorizations.
Highlights
The effects of solar activity on the magnetic environment of the Earth have been observed since the publication of Edward Sabine’s work in 1852 (Sabine, 1852)
We present in this figure the outcome of our model, combining a non-linear AE transformation of the Advanced Composition Explorer (ACE) data set with the unsupervised classification of the encoded data using the Dynamic SelfOrganizing Map (DSOM) method
In this paper we show how the categorization of solar wind can be informed by classic unsupervised clustering methods and SelfOrganizing Maps (SOM)
Summary
The effects of solar activity on the magnetic environment of the Earth have been observed since the publication of Edward Sabine’s work in 1852 (Sabine, 1852). It was natural to classify the solar wind by defining a boundary between fast and slow winds (Neugebauer and Snyder, 1966; Schwenn, 1983; Schwenn and Marsch, 1990; Habbal et al, 1997) The former has been associated with mean speed values of 750 km/s (or in some publications with values larger than 600 km/s), while the later shows a limit at 500 km/s, where the compositional ratio (Fe/O) shows a break (Feldman et al, 2005; Stakhiv et al, 2015). These particles have multiple energies and show a variety of kinetic properties, including non-Maxwellian velocity distributions (Pierrard and Lazar, 2010; Matteini et al, 2012)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
