Automatic solar flare tracking

Abstract

In October 2003 we began a four year project whose goal is to build a real-time space weather monitoring and forecasting system. A major component of the project is the use of image processing and pattern recognition techniques to detect and characterize three important solar activities in real–time: filament eruptions, flares, and emerging flux regions. In this paper we describe a method for automatic tracking of the apparent separation motion of two–ribbon flares using a support vector machine for flare detection and image segmentation for flare characterization.