|
Article information
2019 , Volume 24, ¹ 6, p.50-59
Kramareva L.S., Andreev A.I., Bloshchinskiy V.D., Kuchma M.O., Davidenko A.N., Pustatintsev I.N., Shamilova Y.A., Kholodov E.I., Korolev S.P.
The use of neural networks in hydrometeorology problems
The purpose of this paper is to study methods and technologies based on machine learning algorithms for solution of hydrometeorological problems associated with the satellite multispectral image classification. Methodology. The models presented in the paper are developed using various machine learning methods, such as Support Vector Machine (SVM) as well as Convolutional Neural Networks (CNN). Multispectral satellite images are used as input data from which training and test datasets are formed, consisting of more than 270 thousands units. For calculations, the resources of hybrid computing systems and specialized machine learning libraries were used. Results. The models have been developed for classifying the underlying surface. Methods for verifying snow masks and clouds are described. Findings. It is shown that the use of textures in training of CNN increases the classification accuracy compared to other methods, especially in situations where spectral characteristics are similar (cold ice clouds and snow surface). The testing process of the developed algorithms includes an assessment of accuracy using metrics (f-measure, false alarm detection etc.) calculated for test datasets, as well as comparison with state-of-art models and the results of manual segmentation by experienced meteorologists. The test results showed a fairly high level of reliability and accuracy of classification.
[full text] [link to elibrary.ru]
Keywords: remote sensing, machine learning, neural network, CNN, textures, satellite image classification
doi: 10.25743/ICT.2019.24.6.007
Author(s):
Kramareva Lyubow Sergeevna
Position: Director
Office: Far-Eastern Center of State Research Center for Space Hydrometeorology Planeta
Address: 680000, Russia, Khabarovsk, 18, Lenina st.
Phone Office: (4212) 21-42-21
E-mail: kramareva@dvrcpod.ru
SPIN-code: 6216-0029Andreev Alexander Ivanovich
Position: Junior Research Scientist
Office: Far-Eastern Center of State Research Center for Space Hydrometeorology Planeta
Address: 680000, Russia, Khabarovsk, 18, Lenina st.
Phone Office: (4212) 21-42-21
E-mail: alexander.andreev.mail@gmail.com
SPIN-code: 1693-4323Bloshchinskiy Vladislav Dmitrievich
Position: Programmer
Office: Far-Eastern Center of State Research Center for Space Hydrometeorology Planeta
Address: 680000, Russia, Khabarovsk, 18, Lenina st.
Phone Office: (4212) 21-42-21
E-mail: v.bloshchinsky@dvrcpod.ru
SPIN-code: 6120-5235Kuchma Mikhail Olegovich
Position: Junior Research Scientist
Office: Far-Eastern Center of State Research Center for Space Hydrometeorology Planeta
Address: 680000, Russia, Khabarovsk, 18, Lenina st.
Phone Office: (4212) 21-42-21
E-mail: m.kuchma@dvrcpod.ru
SPIN-code: 1736-9780Davidenko Aleksey Nikolaevich
Position: Deputy Director (Economy)
Office: Far-Eastern Center of State Research Center for Space Hydrometeorology Planeta
Address: 680000, Russia, Khabarovsk, 18, Lenina st.
Phone Office: (4212) 21-42-21
E-mail: a.davidenko@dvrcpod.ru
SPIN-code: 1505-3245Pustatintsev Igor Nikolaevich
Position: Head of Departament
Office: Far-Eastern Center of State Research Center for Space Hydrometeorology Planeta
Address: 680000, Russia, Khabarovsk, 18, Lenina st.
Phone Office: (4212) 21-42-21
E-mail: in.pustatintsev@dvrcpod.ru
Shamilova Yuliya Andreevna
Position: leading meteorologist
Office: Far-Eastern Center of State Research Center for Space Hydrometeorology Planeta
Address: 680000, Russia, Khabarovsk, 18, Lenina st.
Phone Office: (4212) 21-42-21
E-mail: ovp2@dvrcpod.ru
Kholodov Egor Igorevich
Position: Junior Research Scientist
Office: Far-Eastern Center of State Research Center for Space Hydrometeorology Planeta
Address: 680000, Russia, Khabarovsk, 18, Lenina st.
Phone Office: (4212) 21-42-21
E-mail: kholodovegor92@mail.ru
Korolev Sergey Pavlovich
Position: Research Scientist
Office: CC FEB RAS
Address: 680000, Russia, Khabarovsk, 18, Lenina st.
Phone Office: (4212) 703913
E-mail: serejk@febras.net
SPIN-code: 5884-4506 References:
[1] Khatami, R., Mountrakis, G., Stehman, V.S. A meta-analysis of remote sensing research on supervised pixel-based land-cover image classification processes: General guidelines for practitioners and future research. Remote Sensing of Environment. 2016; (177):89–100.
[2] Mal’kovskii, S.I., Sorokin, A.A., Korolev, S.P., Zatsarinnyi, A.A., Tsoi, G.I. Performance Evaluation of a Hybrid Computer Cluster Built on IBM POWER8 Microprocessors. Programming and Computer Software. 2019; 45(6):324–332.
[3] Astafurov, V.G., Skorokhodov, A.V. Multi-layer cloud classification from MODIS data using neural network technology and fuzzy logic approach. Current problems in remote sensing of the earth from space. 2015; 12(6):162–173. (In Russ.)
[4] Shi, X., Gao, Z., Lausen, L., Wang, H., Yeung, D.Y., Wong, W., WOO W. Deep learning for precipitation nowcasting: A benchmark and a new model. Advances in neural information processing systems. 2017:5617–5627.
[5] Campolo, M., Andreussi, P., Soldati, A. River flood forecasting with a neural network model. Water resources research. 1999; 35(4):1191–1197.
[6] Bankert, R.L., Mitrescu, C. Comparison of GOES cloud classification algorithms employing explicit and implicit physics. Journal of applied meteorology and climatology. 2009; 48(7):1411–1421.
[7] Raschka, S. Python machine learning. Packt Publishing Ltd; 2015: 417.
[8] Le Goff, M., Tourneret, J.-Y., Wendt, H., Spigai, M. Deep learning for cloud detection. Proc. of the Intern. Conf. ICPRS (8th International Conference of Pattern Recognition Systems). Madrid; 2017:1–6.
[9] Löfstedt, T., Brynolfsson, P., Asklund, T., Nyholm, T., Garpebring, A. Gray-level invariant Haralick texture features. PloS one. 2019; 14(2):e0212110. Available at: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0212110
[10] Ma, W.Y., Manjunath, B.S. A comparison of wavelet transform features for texture image annotation. Proc. of the Intern. Conf. on Image Processing. IEEE. 1995; (2):256–259.
[11] LeCun, Y., Bengio, Y. Convolutional networks for images, speech, and time series. The handbook of brain theory and neural networks. 1995; 3361(10):1995.
[12] Chen, W., Duan, J, Juan, M. Cloud Detection via Convolutional Neural Network in Visible Light Remote Sensing Images. Proc. of the Intern. Conf. on Artificial Intelligence: Techniques and Applications (AITA2017). 2017:38-43. Available at: http://www.dpi-proceedings.com/index.php/dtcse/article/viewFile/15987/15495
[13] Krizhevsky, A., Sutskever, I., Hinton, G.E. Imagenet classification with deep convolutional neural networks. Advances in neural information processing systems. 2012: 1097–1105.
[14] Simonyan, K., Zisserman, A. Very Deep Convolutional Networks For Large-Scale Image Recognition. Proc. of the ImageNet Large Scale Visual Recognition Challenge (ILSVRC). Publisher: Computing Research Repository (CoRR). Preprint / arXiv:1409.1556v6. 2015: 14. Available at: https://arxiv.org/pdf/1409.1556.pdf%20http://arxiv.org/abs/1409.1556.pdf
[15] Szegedy, C., Ioffe, S., Vanhoucke, V. Alemi, A.A. Inception-v4, inception-resnet and the impact of residual connections on learning. Proc. of the Thirty-First AAAI Conference on Artificial Intelligence. 2016: 4278–4284.
Bibliography link:
Kramareva L.S., Andreev A.I., Bloshchinskiy V.D., Kuchma M.O., Davidenko A.N., Pustatintsev I.N., Shamilova Y.A., Kholodov E.I., Korolev S.P. The use of neural networks in hydrometeorology problems // Computational technologies. 2019. V. 24. ¹ 6. P. 50-59
|
