Potential of Sentinel-1 time series data for the estimation of season length in winter wheat phenology
Time series analysis has high potential for the monitoring of agricultural management intensity and crop yield. The C-band synthetic aperture radar (SAR) data from Sentinel-1 provide a unique source to create area-wide dense time series of indicators that are sensitive to crop parameters. Here, time series of backscattering coefficient ratio from Sentinel-1 were established in individual winter wheat fields over three consecutive years. Phenology metrics were computed in order to indicate the length of the season, where the plant is substantially growing in height and building biomass. The average estimated lengths of season of winter wheat were 112 days in 2017, 77 days in 2018 and 91 days in 2019. The observed lengths of the season in the reference were 114 days in 2017, 73 days in 2018 and 88 days in 2019. The results for the individual winter wheat fields show that the length of the season was estimated with an RMSD (root mean squared deviation) of less than 2 weeks for all three years. The results confirmed that the VH/VV ratio has high potential for monitoring phenological features.