Multi-year analysis of FUV auroral images

An adaptive image segmentation technique is used with a detector-specific statistical dayglow model to automatically identify auroral morphologies in images from the Polar Ultraviolet Imager (UVI). This algorithm ignores the outlier pixels and segments the rest pixels of the image based on the histogram distribution. The segmented information is stored in a database known as the UVI Online Search Tool that includes image-derived search criteria such as auroral boundaries, integrated power and distribution, and auroral morphology. The database spans the entire UVI mission lifetime (1996 to present) with a 10 minute resolution. This database of image information represents a unique collection of auroral activity that can be used to perform extended (multi-year) studies of auroral variations. This paper will discuss the challenges of automatically extracting features from space-based auroral images and then present samples of synoptic studies enabled by this new approach. Studies include examination for temporal correlation of UVI-derived auroral power deposition with solar wind epsilon as a gauge of the efficacy of magnetopause power coupling, investigation of the morphology of LBH band brightness mean and variance for Bz North conditions, and examination of correlation between IMF orientation and local time asymmetries seen in substorm morphologies.