Журнал

About Editorial Board Information for authors and reviewers Publication ethics Contacts User Account: send / review a manuscript
Archive
Volume 21, 2024
Number 1
Volume 20, 2023
Number 1 Number 2 Number 3 Number 4 Number 5 Number 6
Volume 19, 2022
Number 1 Number 2 Number 3 Number 4 Number 5 Number 6
Volume 18, 2021
Number 1 Number 2 Number 3 Number 4 Number 5 Number 6
Volume 17, 2020
Number 1 Number 2 Number 3 Number 4 Number 5 Number 6 Number 7
Volume 16, 2019
Number 1 Number 2 Number 3 Number 4 Number 5 Number 6
Volume 15, 2018
Number 1 Number 2 Number 3 Number 4 Number 5 Number 6 Number 7
Volume 14, 2017
Number 1 Number 2 Number 3 Number 4 Number 5 Number 6 Number 7
Volume 13, 2016
Number 1 Number 2 Number 3 Number 4 Number 5 Number 6
Volume 12, 2015
Number 1 Number 2 Number 3 Number 4 Number 5 Number 6
Volume 11, 2014
Number 1 Number 2 Number 3 Number 4
Volume 10, 2013
Number 1 Number 2 Number 3 Number 4
Volume 9, 2012
Number 1 Number 2 Number 3 Number 4 Number 5
Volume 8, 2011
Number 1 Number 2 Number 3 Number 4
Volume 7, 2010
Number 1 Number 2 Number 3 Number 4
Issue 6, 2009
Volume 1 Volume 2
Issue 5, 2008
Volume 1 Volume 2
Issue 4, 2007
Volume 1 Volume 2
Issue 3, 2006
Volume 1 Volume 2
Issue 2, 2005
Volume 1 Volume 2
Issue 1, 2004
Volume 1
Search
Find:
русский
ISSN 2070-7401 (Print), ISSN 2411-0280 (Online)
Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa
CURRENT PROBLEMS IN REMOTE SENSING OF THE EARTH FROM SPACE

  

Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa, 2021, Vol. 18, No. 3, pp. 51-64

Detection of forest disturbances in Sentinel-2 images with convolutional neural networks

A.V. Tarasov 1 , A.N. Shikhov 1 , T.V. Shabalina 1 
1 Perm State University, Perm, Russia
Accepted: 17.06.2021
DOI: 10.21046/2070-7401-2021-18-3-51-64
Monitoring of forest disturbances is an important application field in forest management. Deep learning (convolutional neural network, CNN) is state-of-the-art approach to improve accuracy of forest changes detection. In this study, the algorithm for forest losses detection with Sentinel-2 images based on U-Net architecture was proposed. Training and evaluation were conducted on our own dataset compiled on Sentinel-2 images for several regions of the European Russia. More than 50 experiments with base U-Net architecture to find the best model were performed. It was found that Red, NIR, SWIR1, SWIR2 bands of Sentinel-2 images and their differences were the most important features for forest change detection. General model for all season and separate models for summer, winter and transitional season were developed. Against traditional methods based on map algebra substantial improvement (more than two times) of the accuracy of forest disturbances detection was achieved by the developed models. The developed CNN-based models identified selective loggings as single (unified) forest loss areas as opposed to traditional methods that detect only pixels with strong difference of spectral reflectance. The developed models can be applied in forest change detection in autumn-winter season due to lower shadow sensitivity. Main limitation of the models is time-consuming compilation of training dataset. However, training dataset size increase facilitates adaptation of the algorithm to new data.
Keywords: forest change detection, Sentinel-2 data, convolutional neural networks, deep learning, semantic segmentation, U-Net
Full text

References:

  1. Andreev A. I., Shamilova Yu. A., Kholodov E. I., Using Convolutional Neural Networks for Cloud Detection from Meteor-M No. 2 MSU-MR Data, Russian Meteorology and Hydrology, 2019, Vol. 44(7), pp. 459–466.
  2. Krylov A. M., Vladimirova N. A., Remote sensing data for forest monitoring, Geomatica, 2011, No. 3, pp. 53–58 (in Russian).
  3. Loupian E. A., Burtsev M. A., Proshin A. A., Kobets D. A., Evolution of remote monitoring information systems development concepts, Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa, 2018, Vol. 15, No. 3, pp. 53–66 (in Russian), DOI: 10.21046/2070-7401-2018-15-3-53-66.
  4. Allen T. R., Kupfer J. A., Spectral response and spatial pattern of Fraser fir mortality and regeneration, Great Smoky Mountains, USA, Plant Ecology, 2001, Vol. 156, pp. 59–74.
  5. Cocke A. E., Fulé P. Z., Crouse J. E., Comparison of burn severity assessments using Differenced Normalized Burn Ratio and ground data, Intern. J. Wildland Fire, 2005, Vol. 14(2), pp. 189–198.
  6. de Bem P. P., de Carvalho O. A. Jr., Guimarães R. F., Gomes R. A. T., Change detection of deforestation in the Brazilian Amazon using landsat data and convolutional neural networks, Remote Sensing, 2020, Vol. 12, DOI: 10.3390/rs12060901.
  7. Dice L. R., Measures of the amount of ecologic association between species, Ecology, 1945, Vol. 26, pp. 297–302.
  8. Isaienkov K., Yushchuk M., Khramtsov V., Seliverstov O., Deep Learning for Regular Change Detection in Ukrainian Forest Ecosystem with Sentinel-2, IEEE J. Selected Topics in Applied Earth Observations and Remote Sensing, 2021, Vol. 14, pp. 364–376.
  9. Hethcoat M. G., Edwards D. P., Carreiras J. M. B., Bryant R. G., França F. M., Quegan S., A machine learning approach to map tropical selective logging, Remote Sensing of Environment, 2019. Vol. 221, pp. 569–582, DOI: 10.1016/j.rse.2018.11.044.
  10. Khryashchev V., Ivanovisky L., Pavlov V., Rubtsov A., Ostrovskay A., Comparison of different convolutional neural network architectures for satellite image segmentation, Proc. 23 rd Conf. Fruct Association, Jyvaskyla, Finland, 2018, pp. 172–180.
  11. Larabi M., Liu Q., Wang Y., Convolutional neural network features based change detection in satellite images, Proc. 1 st Intern. Workshop Pattern Recognition, 2016, Art. No. 100110W.
  12. Li Z., Shen H., Cheng Q., Liu Y., You S., He Z., Deep learning based cloud detection for medium and high resolution remote sensing images of different sensors, ISPRS J. Photogrammetry and Remote Sensing, 2019, Vol. 150, pp. 197–212, DOI: 10.1016/j.isprsjprs.2019.02.017.
  13. Mou L., Bruzzone L., Zhu X. X., Learning spectral-spatial features via a recurrent convolutional neural network for change detection in multispectral imagery, IEEE Trans. Geoscience and Remote Sensing, 2019, Vol. 57(2), pp. 924–935.
  14. Nielsen A. A., Conradsen K., Simpson J. J., Multivariate alteration detection (MAD) and MAF postprocessing in multispectral, bitemporal image data: New approaches to change detection studies, Remote Sensing of Environment, 1998, Vol. 64(1), pp. 1–19.
  15. Ronneberger O., Fischer P., Brox T., U Net: Convolutional networks for biomedical image segmentation, arXiv preprint. arXiv: 1505.04597, 2015, 8 p., available at: https://arxiv.org/abs/1505.04597v12015, arXiv: 1505.04597.
  16. Scharvogel D., Brandmeier M., Weis M., A Deep Learning Approach for Calamity Assessment Using Sentinel-2 Data, Forests, 2020, Vol. 11, DOI: 10.3390/f11121239.
  17. Srivastava N., Hinton G., Krizhevsky A., Sutskever I., Salakhutdinov R., Dropout: A Simple Way to Prevent Neural Networks from Overfitting, J. Machine Learning Research, 2014, Vol. 15, Art. No. 56, pp. 1929–1958.
  18. Syrris V., Hasenohr P., Delipetrev B., Kotsev A., Kempeneers P., Soille P., Evaluation of the potential of convolutional neural networks and random forests for multi-class segmentation of Sentinel-2 imagery, Remote Sensing, 2019, Vol. 11(8), Art. No. 907, DOI: 10.3390/rs11080976.
  19. Wang F., Xu Y. J., Comparison of remote sensing change detection techniques for assessing hurricane damage to forests, Environmental Monitoring and Assessment, 2010, Vol. 162, pp. 311–326.
  20. Wieland M., Li Y., Martinis S., Multi-sensor cloud and cloud shadow segmentation with a convolutional neural network, Remote Sensing, 2019, Vol. 230, Art. No. 111203.
  21. Wu L., Li Z., Liu X., Zhu L., Tang Y., Zhang B., Xu B, Liu M., Meng Y., Liu B., Multi-type forest change detection using BFAST and monthly Landsat time series for monitoring spatiotemporal dynamics of forests in subtropical wetland, Remote Sensing, 2020, Vol. 12(2), Art. No. 341, DOI: 10.3390/rs12020341.
  22. Xie F., Gao Q., Jin C., Zhao F., Hyperspectral image classification based on superpixel pooling convolutional neural network with transfer learning, Remote Sensing, 2021, Vol. 13(5), pp. 1–17.