For this dataset, we used 50k soybean samples collected from thematic maps produced by the Global Land AnalysisDiscovery