Paper
6 July 2022 Near-real-time flood mapping of the Amur River basin from sentinel-2 MSI data using deep learning
M. O. Kuchma, Yu. A. Shamilova, Yu. A. Amelchenko, A. I. Andreev, E. I. Kholodov
Author Affiliations +
Proceedings Volume 12296, International Conference on Remote Sensing of the Earth: Geoinformatics, Cartography, Ecology, and Agriculture (RSE 2022); 1229602 (2022) https://doi.org/10.1117/12.2642787
Event: International Conference on Remote Sensing of the Earth: Geoinformatics, Cartography, Ecology, and Agriculture, 2022, Dushanbe, Republic of Tajikistan
Abstract
In this paper, the authors propose an algorithm for automatic near-real-time flood mapping of the Amur River basin from Sentinel-2 MSI data using a U-net convolutional neural network adapted to the task. As a training set, we used Sentinel-2 Level-2A data and vector maps of river floods, created manually by specialists from the Far-Eastern Center of State Research Center for Space Hydrometeorology “Planeta”. According to the training results, Precision was 94.91%, Recall - 90.76%, F1-measure - 92.79%. High accuracy estimates and fast operation speed make it possible to use the developed algorithm for automatic near-real-time flood mapping of the Amur River basin in complex monitoring problems.
© (2022) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
M. O. Kuchma, Yu. A. Shamilova, Yu. A. Amelchenko, A. I. Andreev, and E. I. Kholodov "Near-real-time flood mapping of the Amur River basin from sentinel-2 MSI data using deep learning", Proc. SPIE 12296, International Conference on Remote Sensing of the Earth: Geoinformatics, Cartography, Ecology, and Agriculture (RSE 2022), 1229602 (6 July 2022); https://doi.org/10.1117/12.2642787
Advertisement
Advertisement
RIGHTS & PERMISSIONS
Get copyright permission  Get copyright permission on Copyright Marketplace
KEYWORDS
Floods

Multispectral imaging

Algorithm development

Associative arrays

Scene classification

Image processing

Image segmentation

Back to Top