Aurora Phenomenon Localization, Classification, and Temporal Evolution Tracking

Our work aims to produce an improved on-line search tool to the large archive of aurora images produced by the NASA Polar mission’s UVI sensor and the IMAGE mission’s FUV sensor. There are approximately 9 million images in these collections, and it is challenging for researchers to find all images of interest for their investigations. Previously, Co-I Germany developed the On-line Search Tool (OST), an interface to the database of UVI images. The tool allows retrieval of images based on time of collection and some simple auroral features. While OST was a great advance, it is limited in that it can retrieve based only on a few simple features. It is also limited by various complications in the images, including high levels of noise, low contrast, and stars in some images. Finally, presence of day glow in many images can confuse the methods used by the tool’s image miner. There is no existing tool that allows retrieval of FUV images based on simple auroral features. The project aims to improve the robustness of the existing OST to image complications and to extend the OST capabilities to allow FUV retrieval. In particular, one aim is to develop more robust methods to locate the auroral oval in the imagery, primarily by exploiting the known shape of the auroral oval. Given the auroral oval, characteristics of the aurora in each image can be determined, allowing retrieval of imagery based on aurora characteristics. A second aim is to support retrieval of images based on temporal features, for example, retrieval of images during substorm onset.