Japan Geoscience Union Meeting 2024

Presentation information

[E] Poster

P (Space and Planetary Sciences ) » P-EM Solar-Terrestrial Sciences, Space Electromagnetism & Space Environment

[P-EM15] Study of coupling processes in solar-terrestrial system

Wed. May 29, 2024 5:15 PM - 6:45 PM Poster Hall (Exhibition Hall 6, Makuhari Messe)

convener:Mamoru Yamamoto(Research Institute for Sustainable Humanosphere, Kyoto University), Yasunobu Ogawa(National Institute of Polar Research), Satonori Nozawa(Institute for Space-Earth Environmental Research, Nagoya University), Akimasa Yoshikawa(Department of Earth and Planetary Sciences, Kyushu University)

5:15 PM - 6:45 PM

[PEM15-P07] Development of automatic scaling technique for FMCW ionograms

*Michi Nishioka1, Septi Perwitasari1, Takuya Tsugawa1 (1.National Institute of Information and Communications Technology)

Keywords:automatic scaling, ionogram, semantic segmentation

The National Institute of Information and Communications Technology (NICT) has been conducting ionosphere observations using ionosondes across Japan, Southeast Asia, and Antarctica. Automatic scaling of ionograms has been implemented for Japanese observation, marking a significant evolution from the original geometric parameter extraction method to utilizing an instance segmentation model for detecting ionospheric echo rectangles, and further to employing a semantic segmentation model capable of identifying ionospheric echo traces directly. However, observations in Southeast Asia and Antarctica utilize FMCW ionosondes for power and space efficiency, presenting challenges in automatic scaling due to low signal-to-noise ratios and O/X mode unseparated ionograms.
This study aims to adapt the successful semantic segmentation model, used for scaling domestic ionograms, to FMCW ionograms. Over 1,000 ionograms from Southeast Asia and 300 from Antarctica were manually traced to create a teacher dataset. Following model development, parameters such as foF2 and h’F were accurately scaled by over 70% compared to manual readings. This scaling technique enables the automatic scaling of historical data, which will be made available on NICT’s website starting March 2024.