DESIS and PRISMA: A study of a new generation of spaceborne hyperspectral sensors in the study of world crops

Advances in agricultural mapping and characterization are now possible through 1) new generation spaceborne hyperspectral sensors, 2) cloud computing platforms, and 3) advanced machine learning algorithms. The German Aerospace Center (DLR) Earth Sensing Imaging Spectrometer (DESIS) and the Italian Space Agency�s PRecursore IperSpettrale della Missione Applicative (PRISMA) provide unprecedented high-quality hyperspectral data. While their numbers of bands, spatial resolutions, and signal to noise ratios are comparable, important differences between the two may affect classification accuracy. DESIS is mounted on the International Space Station while PRISMA is polar-orbiting. DESIS has a higher spectral resolution (2.55 nm vs 10 nm), but a shorter spectral range (400-1000 nm vs 400-2505 nm). The challenges of using such high data volumes are ameliorated by cloud computing platforms such as Amazon Web Services, Microsoft's Azure, and Google Earth Engine (GEE), which allow users to access and process data stored in the cloud using the platform�s computing power. Lastly, recent advances in machine learning algorithms lead to higher classification accuracies even in such complex systems as agriculture. We demonstrate agricultural crop type classification using PRISMA and DESIS data in GEE using different machine learning algorithms. We compared classification accuracies using a DESIS image acquired on June 18, 2020 and a PRISMA image acquired on June 17, 2020 in the Central Valley, California. In GEE, we trained several machine learning models (including Support Vector Machine and Random Forest) using the USDA Cropland Data Layer as reference. Results of these analyses are presented here.