Abstract Accurate and spatially detailed information on soybean cultivation is essential for production estimation and land-use management, particularly as cultivated areas expand in emerging producing regions such as South Africa. This research presents SoySA10, a national 10-m soybean mapping dataset for South Africa covering 2018–2025. The dataset was generated from Sentinel-2 time series using a generalised classification model – combining random forest with dynamic time warping – with spectral-phenological features. SoySA10 includes time-series soybean distribution maps, field type maps, and a soy-cultivation change map. Accuracy assessment showed an overall accuracy of 0.89, with F1-scores of 0.66 for soybean and 0.94 for non-soybean classes. Comparisons between mapped and reported soybean areas exhibited high consistency at provincial (R 2 = 0.97 and nRMSE = 7%) and district levels (R 2 > 0.80 and nRMSE < 20%). Strong spatial agreement was observed between SoySA10, the Spatial Production Allocation Model (SPAM) 2020, and soybean field boundary polygons at both aggregate and detailed scales. The dataset supports analyses of South Africa’s soy-cultivation dynamics and related downstream applications.
Huang et al. (Tue,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: