A. S. Stepanov1, K. N. Dubrovin2, A. L. Verkhoturov3, N. A. Selezneva1, A. A. Sunyaikin1
1Far Eastern Research Institute of Agriculture, Khabarovsk Federal Research Center, Far Eastern branch, Russian Academy of Sciences, ul. Klubnaya, 13, s. Vostochnoe, Khabarovskii r-n, Khabarovskii krai, 680521, Russian Federation
2Computing Center of the Far Eastern Branch of the Russian Academy of Sciences, Khabarovsk Federal Research Center, Far Eastern branch, Russian Academy of Sciences, ul. Kem Yu Chena, 65, Khabarovsk, 680000, Russian Federation
3Mining Institute of the Far Eastern Branch of the Russian Academy of Sciences, Khabarovsk Federal Research Center, Far Eastern branch, Russian Academy of Sciences, ul. Turgeneva, 51, Khabarovsk, 680000, Russian Federation
Abstract. The control over the observance of crop rotation and identification of crops using Earth remote sensing data is one of the most important tasks of practical agriculture. For the experimental fields of the Far Eastern Research Institute of Agriculture, located in the Khabarovsk region, we obtained a series of optical and radar images from the Landsat-8 and Sentinel-1 satellites for the period from May to September 2017. Using satellite data for 12 calendar dates in 2017, we calculated the values of optical NDVI (Normalized Difference Vegetation Index) and DpRVI (Dual polarimetric Radar Vegetation Index) for individual fields with soybean, oat, and timothy grass. These indices characterized the growth of photosynthetically active plant biomass. The study aimed to assess the possibility of using the NDVI and DpRVI indices for identifying crops in the fields of the Khabarovsk region. To this end, we plotted seasonal variations of the vegetation indices for these crops, approximated by various nonlinear functions. Subsequently, using cluster analysis, we assessed the quality of the classification of the experimental fields. When using the double logistic function, the DpRVI approximation error was 10.4%; when using the 3rd degree polynomial, the NDVI approximation error was 14%. The cluster analysis resulted in allocating three classes, while all soybean fields were correctly classified with DpRVI only. The plotted curves of the seasonal variation of DpRVI values for soybean, oat, and timothy grass, as well as wheat with post-sowing of timothy grass, have a characteristic appearance for each class. The proposed approach could be used to control the crop rotation of soybean, which is the main crop in the south of the Far East. It is also promising for the general classification of arable land.
Keywords: crops; identification; remote sensing of the Earth; vegetation indices; approximation; crop rotation.
Author Details: A. S. Stepanov, D. Sc. (Pharm.), leading research fellow (e-mail: Этот адрес электронной почты защищён от спам-ботов. У вас должен быть включен JavaScript для просмотра.); K. N. Dubrovin, junior research fellow; A.L. Verkhoturov, senior research fellow; N. A. Selezneva, research fellow; A. A. Sunyaikin, senior research fellow.
For citation: Stepanov AS, Dubrovin KN, Verkhoturov AL, et al. [Prospects of using optical and radar images for monitoring crop rotations in the Khabarovsk Territory]. Dostizheniya nauki i tekhniki APK. 2021;35(12):23-8. Russian. doi: 10.53859/02352451_2021_35_12_23.