Mapping annual 10 m rapeseed extent using multisource data in the Yangtze River Economic Belt of China (2017-2021) on Google Earth Engine

Knowledge on the distribution of rapeseed enables accurate yield forecasting and supports agricultural planning to assist global security in edible oil supply. Large-scale remotely sensed maps of rapeseed nevertheless are nonexistent in China because of landscape fragmentation, cloudy weather conditions, and phenological variations. Existing rapeseed mapping methods based on spectral, phenological, and machine learning algorithms are challenging to accurately and seamlessly extract the planting distribution of rapeseed over a large spatial region. In view of this, the annual extent of rapeseed in the Yangtze River Economic Belt (YREB) of China from 2017 to 2021 was mapped. Multisource data including Sentinel-1/2, Google Earth very-high-resolution (GE-VHR) images, and Global Land Data Assimilation System (GLDAS) data were used to develop a systematic methodology. First, the ground reference dataset was collected from fieldwork and all available GE-VHR images to construct a rapeseed point collection (RPC). Second, RPC, Sentinel-1, and GLDAS data were used to estimate the annual rapeseed peak-flowering date maps (RPDMs) based on a random forest regression model. Third, a hybrid mapping strategy was implemented on Google Earth Engine leveraging the dense Sentinel-1/2 data to generate the annual seamless 10 m rapeseed extent maps (REMs). Accuracy assessment illustrates the reliability of the RPDM with an R2 of 0.86 (RMSE = 5.32 days). Comparisons with other rapeseed mapping methods and comprehensive validation attest to the high precision of the REMs, with an average F1 score of 0.93. Moreover, the rapeseed cultivation area decreased over the five study years. Rapeseed fields in the YREB are highly fragmented, averaging 18 per ha, but trend analysis suggests a potential for crop intensification. This study provides an essential reference and information service for rapeseed production and agricultural resource survey in China.