[P-EM10] Dynamics of Magnetosphere and Ionosphere

[PEM10-07] A penalized background subtraction model for scaling of low signal-to-noise ratio Ionogram video images

Yuu Hiroshige¹, *Akiko Fujimoto¹, Shuji Abe², Akihiro Ikeda³, Akimasa Yoshikawa² (1.Kyushu Institute of Technology, 2.Kyushu University, 3.National Institute of Technology, Kagoshima College)

Keywords:Ionospheric Observation, Ionogram scaling, Computer Vision, Motion Detection

The upper atmosphere, known as the ionosphere, can affect shortwave communications and cause satellite positioning errors. Measuring the altitude distribution of electron density in the ionosphere, using High-Frequency radio wave reflections often causes the low signal-to-noise ratio of ionospheric echoes due to radio frequency interference. We propose a model for converting low-signal-to-noise-ratio ionospheric echo video images (Ionogram) into noise-reduced images using image processing techniques, for tracing the ionospheric echoes from Ionogram. The proposed method consists of three processing parts: STEP1. noise removal optimized for individual Ionogram images, STEP2. extraction of ionospheric echoes by penalized background subtraction technique, and STEP3. fine-tuning of ionospheric echo signals using a minimum spanning tree algorithm. For unstable signal-to-noise-ratio Ionograms, the model automatically determines the boundary threshold between signal and noise using ridge regression for STEP1 and non-fixed penalized parameters for STEP2. The proposed model successfully reproduces fine Ionograms.