Forecasting solar flares with a new topological parameter and a supervised machine-learning method

Luca Giovannelli and Cristina Campi

Hosted by Tor Vergata University of Rome, Italy on November 4, 2021


Solar flares originate from active regions (ARs) hosting complex and strong bipolar magnetic fluxes. Forecasting the probability of an AR to flare and defining reliable precursors of intense flares, i.e., X- or M-class flares, are extremely challenging tasks in the space weather field. In this talk, we focus on two metrics as flare precursors, the unsigned flux R*, tested on MDI/SOHO data and calibrated for higher spatial resolution SDO/HMI maps, and a novel topological parameter D representing the complexity of a solar active region. The parameter D is based on the automatic recognition of magnetic polarity inversion lines (PILs) in identified SDO/HMI ARs and is able to evaluate their magnetic topological complexity. We use both a heuristic approach and a supervised machine-learning method to validate the effectiveness of these metrics to predict the occurrence of X- or M-class flares in a given solar AR during the following 24 hr period. Our feature ranking analysis shows that both parameters play a significant role in prediction performances. Moreover, the analysis demonstrates that the new topological parameter D is the only one, among 173 overall predictors, that is systematically ranked within the top 10 positions.


  • Luca Giovannelli (Dipartimento di Fisica, Universit√† degli Studi di Roma Tor Vergata)
  • Cristina Campi (Dipartimento di Matematica, Universit√† di Genova)


Cicogna et al., Flare-forecasting Algorithms Based on High-gradient Polarity Inversion Lines in Active Regions, ApJ, 915, id.38, 2021