The Copernicus Sentinel-2 mission operated by the European Space Agency (ESA) provides comprehensive and continuous multi-spectral observations of all the Earth’s land surface since mid-2015. Clouds and cloud shadows significantly decrease the usability of optical satellite data, especially in agricultural applications; therefore, an accurate and reliable cloud mask is mandatory for effective EO optical data exploitation. During the last few years, image segmentation techniques have developed rapidly with the exploitation of neural network capabilities. With this perspective, the KappaMask processor using U-Net architecture was developed with the ability to generate a classification mask over northern latitudes into the following classes: clear, cloud shadow, semi-transparent cloud (thin clouds), cloud and invalid. For training, a Sentinel-2 dataset covering the Northern European terrestrial area was labelled. KappaMask provides a 10 m classification mask for Sentinel-2 Level-2A (L2A) and Level-1C (L1C) products. The total dice coefficient on the test dataset, which was not seen by the model at any stage, was 80% for KappaMask L2A and 76% for KappaMask L1C for clear, cloud shadow, semi-transparent and cloud classes. A comparison with rule-based cloud mask methods was then performed on the same test dataset, where Sen2Cor reached 59% dice coefficient for clear, cloud shadow, semi-transparent and cloud classes, Fmask reached 61% for clear, cloud shadow and cloud classes and Maja reached 51% for clear and cloud classes. The closest machine learning open-source cloud classification mask, S2cloudless, had a 63% dice coefficient providing only cloud and clear classes, while KappaMask L2A, with a more complex classification schema, outperformed S2cloudless by 17%.

Short temporal baseline regular Synthetic Aperture Radar (SAR) interferometry is a tool well suited for wide area monitoring of agricultural activities, urgently needed in European Union Common Agricultural Policy (CAP) enforcement. In this study, we demonstrate and describe in detail, how mowing and ploughing events can be identified from Sentinel-1 6-day interferometric coherence time series. The study is based on a large dataset of 386 dual polarimetric Sentinel-1 VV/VH SAR and 351 Sentinel-2 optical images, and nearly 2000 documented mowing and ploughing events on more than 1000 parcels (average 10.6 ha, smallest 0.6 ha, largest 108.5 ha). Statistical analysis revealed that mowing and ploughing cause coherence to increase when compared to values before an event. In the case of mowing, the coherence increased from 0.18 to 0.35, while Sentinel-2 NDVI (indicating the amount of green chlorophyll containing biomass) at the same time decreased from 0.75 to 0.5. For mowing, there was virtually no difference between the polarisations. After ploughing, VV-coherence grew up to 0.65 and VH-coherence to 0.45, while NDVI was around 0.2 at the same time. Before ploughing, both coherence and NDVI values were very variable, determined by the agricultural management practices of the parcel. Results presented here can be used for planning further studies and developing mowing and ploughing detection algorithms based on Sentinel-1 data. Besides CAP enforcement, the results are also useful for food security and land use change detection applications.