“…Five remote-sensing-based methods for rapeseed mapping have been developed in recent decades: (1) machine learning methods (Griffiths et al, 2019;Preidl et al, 2020;She et al, 2015;Tao et al, 2020), (2) classification based on time series data (Ashourloo et al, 2019), (3) threshold segmentation based on phenology (Tian et al, 2019), (4) multi-range spectral feature fitting (Pan et al, 2013), and (5) mapping based on HSV (hue, saturation, and value) transformation and spectral features (Wang et al, 2018). Most of these methods, however, only produce rapeseed maps for a small area using very limited imageries taken on rapeseed peak flowering dates (Ashourloo et al, 2019;She et al, 2015). Rapeseed peak flowering dates vary by area and cultivar because of differences in natural conditions and cultivation habits, especially over large regions (d'Andrimont et al, 2020;Ashourloo et al, 2019;McNairn et al, 2018).…”