Virtual EO Constellation for Global Environmental Intelligence

The overall concept and objective of the EU project is to develop AI-based technology to enable coherent fusion of multiple EO systems to yield a single source of consolidated, robust and dependable and secure EO data, which is to serve as the baseline for machine learning models to help farmers optimize their crop production in Europe, mainly through reducing input cost and increasing yield-potential through variable optimization (fertilization/seeding/crop protection). FET Proactive call focuses on emerging themes and to establish a critical mass of European researchers in a number of promising exploratory research topics. The solution is to create machine learning models which can help farmers optimize the fertilizer, crop protection and seed quantities they apply within each field at the optimal time. This project will enable this through providing the technology necessary to classify the different zones of each field to optimize the inputs applied in each area, not only in terms of quantity but also timing. This would not only help the farmers reduce their input cost significantly but it would also optimize yield potential and reduce their Co2 footprint. One key component of our model and product is the ability to monitor the crops during the growth season including the ability to analyse fertilization-uptake, yield potential and performance, where Sentinel 1 and 2 and other EO imagery is a critical component. The current technological restrictions with existing EO data consisting of the different systems constitute a disjoint assortment of sources that cannot be readily combined and effectively jointly exploited. A notable example of the current limitations is illustrated by the domain of precision agriculture. The vast majority of agricultural fields are too small to be analysed from a public satellite such as Sentinel-2, while the high-resolution commercial images are much too expensive to make a commercially viable business case.