Automatic recognition of complex magnetic regions on the Sun in SDO magnetogram images and prediction of flares: Techniques and results for the revised flare prediction program Flarecast

Abstract

In the present paper, solar magnetograms provided by the Helioseismic and Magnetic Imager onboard Solar Dynamics Observatory spacecraft are used to identify active regions automatically by thresholding the line-of-sight component of the solar magnetic field. The flare potential of the regions is predicted by locating potential active regions with strong-gradient polarity inversion lines (SPILs) and estimating 18 physically relevant parameters of these regions. In particular, parameters of interest include the sum of north-south gradients, sum of east-west gradients, length of SPIL, and total integrated magnetic flux. For deterministic prediction of flares, analyses for thresholding of single parameters and different combinations, which include up to four parameters, are presented and compared. If the false alarm rate does not exceed 10% (20%), the probabilities for correct prediction of X-ray flares of class M and greater, M5 and greater, and X in the 24 h window are 71% (86%), 84% (96%), and 94% (100%), respectively. These probabilities are for the best four-parameter technique found. A technique for probabilistic forecasting was also developed. These deterministic and probabilistic techniques will be implemented in a revised version of the flare warning program, Flarecast, which will be operational in the Australian Space Forecast Centre.