Japan Geoscience Union Meeting 2019

Presentation information

[J] Oral

P (Space and Planetary Sciences ) » P-CG Complex & General

[P-CG25] Planetary Magnetosphere, Ionosphere, and Atmosphere

Tue. May 28, 2019 1:45 PM - 3:15 PM A05 (TOKYO BAY MAKUHARI HALL)

convener:Kanako Seki(Graduate School of Science, University of Tokyo), Takeshi Imamura(Graduate School of Frontier Sciences, The University of Tokyo), Hiroyuki Maezawa(Department of Physical Science Osaka Prefecture University), Naoki Terada(Graduate School of Science, Tohoku University), Chairperson:Takeshi Imamura(Graduate School of Frontier Sciences, University of Tokyo), Naoki Terada(School of Science, Tohoku University)

2:30 PM - 2:45 PM

[PCG25-14] Automated segmentation of Martian dust storms in MGS/MOC images using a deep learning technique

*Kazunori Ogohara1, Ryusei Gichu2 (1.School of Engineering, University of Shiga Prefecture, 2.Graduate School of Engineering, University of Shiga Prefecture)

Keywords:Mars, dust storm, deep learning

The shape, texture, and size of dust storms and the climatology of these characteristics are important clues to understanding Martian dust storms. The relations between such characteristics and the phases of the variety of atmospheric waves as well as the frequency of textured/curvilinear dust storms and the spatial and seasonal variations of the cumulative area are also worth investigating (Wang 2007; Wang et al. 2011; Guzewich et al. 2015, 2017). However, it is time-consuming to detect all dust storms visually because of the vast number of images taken by just two instruments, the MGS/MOC and MRO/MARCI. It is also difficult to objectively define which features characterize dust storms. Even approximate categorization of dust events into textured dust storms, untextured yet discrete dust storms, and haze depends on the subjective experience of the human observer. For these reasons, visual detection and categorization of the three types of dust storm textures introduced by Kulowski et al. (2017) would not be exactly reproduced by another human viewing and characterizing the images. Furthermore, we perceive a need to update the criteria for detecting dust events, especially obscured ones such as untextured yet discrete dust storms and haze, via visual detection. Therefore, it would be useful to be able to automatically detect dust events, measure their shape, pattern, and size, and improve the objectivity and reproducibility of dust storm detection.
We have developed a novel algorithm for automated segmentation of Martian dust storms using a encoder-decoder type of Convolutional Neural Network. We can separate dust storm areas from the surface and cloud areas in subsets of the global swath images taken by MGS/MOC without any information from other instruments (e.g. MGS/TES). Assuming that segmentation based on authors' experience is the ground truth, 84% of the true dust storm areas in the Arabia Terra is recognized as dust strom. On the other hand, just 69% of the true dust storm areas in the Hellas Basin is correctly recognized as dust storm.