Applying New Methods to Flare Prediction Using Photospheric Vector Magnetic Field Data.

摘要

The feasibility of using photospheric vector magnetic field data for solar flare prediction is tested using three distinct approaches. First, we examine on a case-by-case basis, the magnitude and variations of over eighty parameters (almost two hundred when both mean and slope are considered), derived from the vector magnetic field data which describe the magnetic state of the active-region photosphere. We find that when pre-event variations and levels are examined using both flare-producing and flare-quiet regions, any obvious or uniquely flare-related signatures are rare. Second, a statistical approach is used for all parameters and all seven active regions (ten flares and fourteen flare-free samples). We demonstrate that (I) combinations of parameters can be found which perfectly distinguish between flare-imminent and flare-quiet samples, but (2) that these combinations are not unique, primarily due to the small sample; still, (3) we are able to identify parameterizations of the distribution of electric current density, magnetic-field twist, and current helicity as well as neutral-line shear angle which routinely appear in well- performing discriminant functions. Third, we demonstrate an application of the Magnetic Charge Topology model to photospheric data, with which we can quantify the complexity of the coronal fields as related to flare events.