Features of the use of neural network technology for recognizing graphic images of thermokarst objects

№4 (2024)

Zhebsain V.V., Gololobov A.Yu., Basharin N.U., Poselsky A.F.

УДК 528.854.2
https://doi.org/10.47148/1609-364X-2024-4-39-47

AbstractAbout the AuthorsReferences
The article presents the results of using neural networks to solve the problem of identifying thermokarst formations observed in regions of permafrost, in particular Central Yakutia, using graphical data from UAVs and Google and Yandex mapping services. Two different approaches were used to solve the problem under consideration, based on the use of multilayer and convolutional neural networks. The dependences of the efficiency and errors of neural networks on the speed and number of learning epochs are studied. The results of the approaches under consideration were compared and the possibility of using neural network technologies to identify thermokarst polygonal formations characteristic of Central Yakutia was assessed.

Vasiliy V. Zhebsain
PhD of Physical and Mathematical Sciences
Head of the Department
North-Eastern Federal University
48, Kulakovsky str., Yakutsk, 677000, Russia
e-mail: zhebs@mail.ru
ORCID ID: 0000-0002-2976-8721
SCOPUS ID: 57198444070
ResearcherID: AAP-4230-2020
SPIN: 7080-6266

Artem Yu. Gololobov
PhD of Physical and Mathematical Sciences
Senior Researcher
Yu.G. Shafer Institute of Cosmophysical Research and Aeronomy
of Siberian Branch of the Russian Academy of Sciences
31, Lenin Ave, Yakutsk, 677000, Russia
e-mail: gololobov@ikfia.ysn.ru
ORCID ID: 0000-0002-3869-4728
SCOPUS ID: 57196248457
ResearcherID: S-7094-2018
SPIN: 2298-7721

Nikolay I. Basharin
Junior Researcher
Permafrost Institute of Siberian Branch of the Russian Academy
of Sciences
36, Merzlotnaya str., Yakutsk, 677000, Russia
e-mail: nikolay_b89@mail.ru
ORCID ID:000-0002-8501-9186
SCOPUS ID: AAC-8710-2019
ResearcherID:57204391163
SPIN:1617-5177

Ayal F. Poselsky
Graduate Student
North-Eastern Federal University in Yakutsk
48, Kulakovsky str. Yakutsk, 677000, Russia
e-mail: al.poselsky@gmail.com
ORCID ID: 0009-0003-2602-8044
ResearcherID: KWT-7859-2024
SPIN: 3432-2450

  1. Sal’va A.M. Tracking of areas of thermokarst occurrences from satellite images (on the example of the main water pipeline route in central Yakutia). Arctic and Antarctica. 2020;(2):126–137. DOI: 10.7256/2453-8922.2020.2.32860.
  2. Konstantinov P., Zhelezniak M., Basharin N., Misailov I., Andreeva V. Establishing of Permafrost Thermal Monitoring Sites in East Siberia. Land. 2020;9(12):1–10. DOI: 10.3390/land9120476.
  3. Basharin N.I., Egorova L.S., Vasilev N.F., Fedorov N.A., Fedorov A.N. Using unmanned aerial vehicles for thermokarst activity assessment. Vestnik of North-Eastern Federal University. Earth sciences. 2020;(3):36–44. DOI: 10.25587/SVFU.2020.19.3.005.
  4. Rashid T. Make Your Own Neural Network. Scotts Valley: CreateSpace, 2016. 222 p.
  5. Bagaev I.I. Concept analysis neural network and conventional neural network, conventional neural network training using the TensorFlow module. Software of systems in the industrial and social fields. 2020;8(1):15–22. DOI: 10.18503/2306-2053-2020-8-1-15-22.

Key words: neural network; thermokarst; modelling; satellite data; aerospace sensing.

Section: Modeling geo objects and geo-processes