Scholarlink

Abstract <p>This study proposes an original approach to the automatic classification of the upper ionosphere state through machine identification of images captured by sky cameras, also known as all-sky imagers. Based on 10 years of sky observations within the auroral oval (Kola Peninsula, Russia), represented by 163 899 images with a 10-minute sampling interval, an intelligent information system was developed using convolutional neural networks. This system identifies whether an input image belongs to one of seven predefined classes and subsequently interprets the result. The analysis of performance metrics for the system, built on the ResNet50 neural network architecture, demonstrated a classification accuracy of 96<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\%\)</EquationSource> <!--BPhysMGU2570291Vorobev-m1--> </InlineEquation>, a level practically unattainable through manual processing of datasets of this scale. This approach holds the highest practical significance in Russia’s polar regions, where reliable and accurate geomagnetic data coverage is sparse (Taymyr Peninsula, Gydan Peninsula, northern areas of Yakutia, etc.). In these regions, auroras serve as the only widely accessible indicator of space weather conditions and the state of the upper ionosphere.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Application of Convolutional Neural Networks for Upper Ionosphere Remote Sensing Using All-Sky Camera Data

  • 期刊论文

  • 发表日期：

    2025年12月1日

查看全文
  • A. V. Vorobev,
  • G. R. Vorobeva

摘要

Abstract

This study proposes an original approach to the automatic classification of the upper ionosphere state through machine identification of images captured by sky cameras, also known as all-sky imagers. Based on 10 years of sky observations within the auroral oval (Kola Peninsula, Russia), represented by 163 899 images with a 10-minute sampling interval, an intelligent information system was developed using convolutional neural networks. This system identifies whether an input image belongs to one of seven predefined classes and subsequently interprets the result. The analysis of performance metrics for the system, built on the ResNet50 neural network architecture, demonstrated a classification accuracy of 96 \(\%\) , a level practically unattainable through manual processing of datasets of this scale. This approach holds the highest practical significance in Russia’s polar regions, where reliable and accurate geomagnetic data coverage is sparse (Taymyr Peninsula, Gydan Peninsula, northern areas of Yakutia, etc.). In these regions, auroras serve as the only widely accessible indicator of space weather conditions and the state of the upper ionosphere.