Contamination Assessment of Soybean Crops in the South of the Russian Far East Based on Sentinel-2 Satellite Data

The study was conducted to develop approaches to identifying weed-infested areas of soybean crops using the data of Earth remote sensing (ERS) from space. A total of 288 Sentinel-2 satellite images were collected from May 1 to October 31 for ten weed-infested areas in the Khankaisky raion of Primorsky krai (2–10 hectares in area) and weed-free soybean fields in Khabarovsk and Primorsky krais (240 fields in 2022, 307 fields in 2023). After filtering using a cloud mask, NDVI (Normalized Difference Vegetation Index) time series were constructed for each day of the period, approximated using nonlinear functions. The average maximum NDVI values of weedy areas (weed count in the range of 100–1200 pcs/m2) were significantly lower (by 0.05–0.09) than the corresponding values of weed-free soybean fields (weed count less than 30 pcs/m2), while the date of the maximum of the approximated NDVI seasonal variation series did not depend on the degree of weed infestation of crops. The seasonal variation of NDVI for weedy areas had a more extended peak in time compared to weed-free fields; therefore, it was proposed to use the width of the peak of the NDVI seasonal variation curve at half (d1/2) and three-quarters (d3/4) of the peak height as a weediness criterion. Average d1/2 values for weedy areas in 2022 ranged from 122 to 127 days, while they ranged from 111 to 121 days in 2023, significantly (p < 0.05) higher than those for weed-free fields (93 to 101 days in 2022 and 93 to 98 days in 2023). Similarly, significant differences were found for d3/4: for weedy areas, they ranged from 74 to 77 days and 74 to 79 days, while they ranged from 60 to 68 days and 57 to 62 days for weed-free areas, respectively. The proposed method can be used to develop a strategy for implementing protective measures for the following season at the field level.