Computer Vision for GNSS-based Detection of the Auroral Oval Boundary

The auroral oval is a region of footprints of the Earth’s magnetic field lines associated with high energy plasma. The high energetic charged particles penetrate into the atmosphere in this region producing irregularities. The oval origin irregularities affect the radio wave propagation degrading radio system operation. For scientific and technical tasks it is important to know the position of the oval boundaries. A technique for estimating the equatorward boundary of the auroral oval is suggested. The technique involves computer vision for data from global navigation satellite systems. First, mathematical morphology to expand data and decrease data gaps is used. Second, K-means and Otsu techniques to cluster image data is employed. Third, a boundary between two cluasters and fit a second-order curve (ellipse) to obtain a continuous boundary is found. Otsu technique results in smaller error than K-means. Obtained technique could be useful to analyze auroral effects on navigation and telecommunication systems, as well as to monitor possible operational threats.