Estimating persistent and random components in spatio-temporal evolutionary patterns of active regions using information theoretic approach

Abstract:

The latitude-belt in which active regions appear migrates equatorward as solar cycle progresses. However, often these active regions tend to appear at the same longitude where an active region previously occurred. On one hand, they don’t appear at one specific longitude always, on the other hand they also don’t appear randomly all over the solar surface. There exists a systematic persistence as well as some randomness in spatio-temporal evolutionary patterns of active regions. By implementing the ‘kmean’ clustering algorithm of machine learning, we evaluate the centroids of each cluster of points representing a given polarity of bipolar active regions in synoptic magnetograms, and apply information theory to the latitude and longitude coordinates of these active regions to derive the information flow at specific latitude-longitude locations with time. We discuss the persistence of locations and information flow from the past to the future. We will determine the spatial and temporal scale of the information flow and persistence of the active regions.