Cloud coverage constitutes a huge problem for Earth observation satellites constellations. It leads to a significant proportion of unusable images that have to be rescheduled, which represents both a waste of time and money. Agile targeting systems combined with satellite planning optimization and weather forecasting allow to minimize the number of cloudy images. As demonstrated earlier by the authors, the computational efficiency of optical flow forecasting approaches allows to build plans with up-to-date forecasts with good spatial and temporal resolutions. This approach, developed and implemented in the field of view of a unique geostationary satellite, is in this work evaluated at worldwide scale by fusion of several geostationary satellites’ fields of view. Using a specific simulation framework, we evaluated the efficiency of this method against a more classical Numerical Weather Prediction model for 24 hours scenarios. Results showed that the optical flow method allows to reduce the rejection proportion of such scenario from thirty to forty percents.
|