DUE Preparatory Activities for Sentinel 2 Exploitation
Arino, Olivier; Koetz, Benjamin; Paganini, Marc; Pinnock, Simon; Seifert, Frank Martin
ESA, ITALY

Time series analysis is becoming a new R&D area as the access to time series of HR optical data is increasing such as with the recent opening of the long term data archive of Landsat and the new Landsat-8 and Sentinel 2 missions. The ESA Sentinel 2 preparatory symposium in April 2012 has shown how Sentinel-2 will bring significant breakthrough in the exploitation of spaceborne optical data, by transferring land cover, agriculture and forestry EO applications, at 10 to 30m spatial resolution, from mapping to monitoring, from snapshot to time series data analysis, from image-based to pixed-based processing. The 5-days temporal revisiting of the Sentinel 2 satellites, when both units will be operated together, will usher in new era for time series analysis at high spatial resolutions. The participants also recognized, within the symposium recommendations, that time series analysis methods (being pixel- or object-based) need to be developed to exploit fully the uniquely dense and long Sentinel 2 data series. This implies that new developments must be initiated to exploit at best the vast amount of information available in dense time series of satellite data. Modelling the spectral behaviour over time in order to exploit the seasonal variations in phenology and hydrology is one of the many R&D areas to be studied.

Recently CNES has decided to use the SPOT-4 deorbiting to simulate Sentinel 2 time series. SPOT-4 at 20m spatial resolution will be produced on predefined sites, for a period of 4 months (February-May 2013), with a temporal revisiting of 5 days. Five ESA projects have been solicited to study the adequacy of these time series of HR optical data respectively in the fields of wetlands, coastal zones, forest, food security and agriculture mapping and monitoring. The SPOT-4 time series will be complemented by other multi temporal acquisitions from the RapidEye and Landsat-8 missions, which will form a unique collection of HR optical time series on a limited number of case studies. Each project will develop HR time series analysis methods on two test sites, for each of the 5 domains of Sentinel 2 applications. The case studies are integrated in five large Data User Element (DUE) and GMES Service Element (GSE) projects developing EO applications of relevance for the Sentinel-2 mission. Preliminary results of the case studies will be shown at the 2013 Living Planet symposium.