Recently, methods based on satellite image processing have often been used to monitor arable land. The automated classification of arable lands using remote sensing data of the Earth makes it possible to identify the declared crop rotations with the real state of affairs, to identify unused lands without labor-intensive trips. At the same time, the accuracy of the models for the classification of agricultural crops in the southern part of the Far East is lower than for traditional agricultural regions of Russia, which is due, firstly, to the peculiarities of the vegetation of agricultural crops in the region, and, secondly, to the insufficiently high quality of information from the Unified Federal Information System of agricultural crops. The paper presents a comparison of the results of the classification of the arable land of the Khabarovsk krai with a total area of more than 4,000 hectares using the contours of the Unified Federal Information System and the corresponding sample of fields with specified boundaries. A series of optical images with a resolution of 10 m (Sentinel-2A/B) was considered in the period from April to October 2021, time series of Normalized Difference Vegetation Index (NDVI) values were calculated for each of the pixels. Classification was carried out by the quadratic discriminant method, 6 classes were determined: soy, buckwheat, fallow, steam, perennial grasses, oats. It was found that the overall accuracy when using Unified Federal Information System of agricultural crops was 83.1 %, and with refined field contours – 94.1 %. The values of f1 metrics for soybeans and deposits increased from 0.88 to 0.95, buckwheat – from 0.70 to 0.93, steam – from 0.80 to 0.85, perennial grasses – from 0.32 to 0.70. To clarify the contours of the fields, it may be recommended to conduct a preliminary classification based on the Unified Federal Information System of agricultural crops, in the future, a sample with specified boundaries will be used for classification in order to identify individual fields, check crop rotations and solve other problems of digital agriculture.

Alexey S. Stepanov, Konstantin N. Dubrovin, Andrey L. Verkhoturov, Tatiana A. Aseeva

1. Ivanov A. L., Kozubenko I. S., Savin I. Yu., Kiryushin V. I. Tsifrovoe zemledelie [Digital farming]. Vestnik rossiiskoi sel'skokhozyaistvennoi nauki. – Bulletin of the Russian Agricultural Science, 2018; 5: 4–9 (in Russ.).
2. Torikov V. E., Pogonyshev V. A., Pogonysheva D. A., Dornykh G. E. Sostoyanie tsifrovoi transformatsii sel'skogo khozyaistva [State of digital transformation of agriculture]. Vestnik Kurskoy gosudarstvennoy sel'skokhozyaystvennoy akademii. – Bulletin of the Kursk State Agricultural Academy, 2020; 9: 6–13 (in Russ.).
3. Sergeeva O. S. Primenenie geoinformatsionnykh tekhnologii dlya povysheniya effektivnosti zemel'nogo nadzora [The application of geo-information technologies to increase the effectiveness of land supervision]. Geograficheskiy vestnik. – Geographical bulletin, 2019; 4:154–162 (in Russ.).
4. Arzamastseva N. V., Prokhorova N. V., Khamidova L. L. Problema dostovernosti i polnoty informatsii o sostoyanii i ispol'zovanii zemel' sel'skokhozyaistvennogo naznacheniya [Problem of the accuracy and completeness of information on the status and use of agricultural lands]. Izvestiya Timiryazevskoy sel'skokhozyaystvennoy akademii. – Izvestiya of Timiryazev Agricultural Academy, 2021; 3: 119–128 (in Russ.).
5. Plotnikov D. E., Kolbudaev P. A., Bartalev S. A., Lupyan E. A. Avtomaticheskoe raspoznavanie ispol'zuemykh pakhotnykh zemel' na osnove sezonnykh vremennykh serii vosstanovlennykh izobrazhenii Landsat [Automated annual cropland mapping from reconstructed time series of Landsat data]. Sovremennye problem distantsionnogo zondirovaniya Zemli iz kosmosa. – Modern problems of remote sensing of the Earth from space, 2018; 15; 2: 112–127 (in Russ.).
6. Miklashevich T. S., Bartalev S. A., Plotnikov D. E. Interpolyatsionnyi algoritm vosstanovleniya dlinnykh vremennykh ryadov dannykh sputnikovykh nablyudenii rastitel'nogo pokrova [Interpolation algorithm for the recovery of long satellite data time series of vegetation cover observation]. Sovremennye problem distantsionnogo zondirovaniya Zemli iz kosmosa. – Modern problems of remote sensing of the Earth from space, 2019; 16; 6: 143–154 (in Russ.).
7. Skvortsov E. A., Nabokov V. I., Nekrasov K. V., Skvortsova E. G., Krotov M. I. Primenenie tekhnologii iskusstvennogo intellekta v sel'skom khozyaistve [Application of technologies of artificial intelligence in agriculture]. Agrarnyi vestnik Urala. – Agrarian Bulletin of the Urals, 2019; 8: 91–97 (in Russ.).
8. Griffiths P., Nendel C., Hostert P. Intra-annual reflectance composites from Sentinel-2 and Landsat for national-scale crop and land cover mapping. Remote Sensing of Environment, 2019; 220: 135–151.
9. Liu J., Shang J., Qian B., Huffman T. Crop yield estimation using time-series MODIS data and the effects of cropland masks in Ontario, Canada. Remote Sensing, 2019; 11: 2419.
10. Zhang H., Kang J., Xu X. Accessing the temporal and spectral features in crop type mapping using multi-temporal Sentinel-2 imagery: A case study of Yi’an County, Heilongjiang province, China. Computers and Electronics in Agriculture, 2020; 176: 105618.
11. Hao P., Tang H., Cheng Z. Early-season crop type mapping using 30-m reference time series. Journal of Integrative Agriculture, 2020; 19: 1897–1911.
12. Tufail R., Ahmad A., Javed M., Ahmad S. R. A machine learning approach for accurate crop type mapping using combined SAR and optical time series data. Advances in Space Research, 2021; 69: 331–346.
13. Yan S., Yao X., Zhu D., Liu D. Large-scale crop mapping from multi-source optical satellite imageries using machine learning with discrete grids. Internшational Journal of Applied Earth Observation and Geoinformation, 2021; 103: 102485.
14. Loupian E. A., Savin I. Yu., Bartalev S. A., Tolpin V. A., Balashov I. V., Plotnikov D. E. Sputnikovyi servis monitoringa sostoyaniya rastitel'nosti ("VEGA") [Satellite service of monitoring of the vegetation state ("VEGA")]. Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa. – Modern problems of remote sensing of the Earth from space, 2011; 8; 1: 190–198 (in Russ.).
15. Loupian E. A., Bartalev S. A., Tolpin V. A. Ispol'zovanie sputnikovogo servisa VEGA v regional'nykh sistemakh distantsionnogo monitoringa [The use of VEGA satellite service in regional remote monitoring systems]. Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa. – Modern problems of remote sensing of the Earth from space, 2014; 11; 3: 215–232 (in Russ.).
16. Tolpin V., Loupian E., Bartalev S. Possibilities of agricultural vegetation condition analysis with the "VEGA" satellite service. Atmospheric and Oceanic Optics, 2014; 27: 581–586.