Are There Natural Categories of Solar Wind?
Friday, 19 December 2014
What seem to be the most obvious categories of solar wind, such as fast and slow, often turn out to be difficult to pin down on closer examination. For example, while slow winds tend to be dense and nonAlfvenic, there are significant exceptions, with some slow winds being not only very Alfvenic but also exhibiting many fast wind traits. Here we use "unsupervised" data mining to look for "natural" solar wind types. We use a set of variables to represent the state of the system and apply what are now standard algorithms to look for natural clustering of these variables. We have done this process for the solar wind density, speed, a carbon charge state ratio (6+ to 5+), the cross-helicity, and the "residual energy." When using the first three of these, we find two groups that tend to be slow and fast, but with the boundary between the groups that is a combination of speed and density. When all five variables are used, the best characterization of the states is as three basic groups in the cross-helicity vs residual energy space, i.e., in terms of "turbulence" measures rather than simple parameters. The three-variable case is largely but not completely reproduced in its subspace. We will suggest what the results could mean for the understanding of issues such as solar wind acceleration.