The influence of active region information on the prediction of solar flares: an empirical model using data mining

M. Nuñez (1) , R. Fidalgo (1) , M. Baena (1) , R. Morales (1)

1 Department of Computer Sciences

Predicting the occurrence of solar flares is a challenge of great importance for many space weather scientists and users. We introduce a data mining approach, called Behavior Pattern Learning (BPL), for automatically discovering correlations between solar flares and active region data, in order to predict the former. The goal of BPL is to predict the interval of time to the next solar flare and provide a confidence value for the associated prediction. The discovered correlations are described in terms of easy-to-read rules. The results indicate that active region dynamics is essential for predicting solar flares.

Domaines: Océan, Atmosphère; Sciences de la Terre

M. Nuñez, R. Fidalgo, M. Baena, R. Morales. The influence of active region information on the prediction of solar flares: an empirical model using data mining. Annales Geophysicae, 2005, 23 (9), pp.3129-3138. ⟨hal-00318009⟩