Analysis of ESA Campaign Datasets for the Generation of Sentinel-2 Simulated Images
Berthelot, Béatrice; saunier, sébastien; hillairet, emmanuel
Magellium, FRANCE

In the framework of the GMES initiative, the Sentinel-2 mission is central for a number of applications, in particular mission for the generation of biogeophysical maps. The purpose of the project is to contribute to the preparation of the L2b processor development before S2 launch and support geophysical algorithm development, validation and the simulation of future Sentinel-2 biophysical products. The activity focuses on a limited set of L2b products highly required by users:

  • Land cover classification;
  • LAI;
  • FAPAR;
  • Leaf chlorophyll content;
  • Leaf water content.
    To achieve this goal, ESA campaigns conducted in the past few years for which both ground measurements as well as airborne or spaceborne observations are available, are used. They include the six following campaigns.
  • Sparc (2003) and Sparc (2004);
  • Sen2Flex (2005);
  • Agrisar (2006);
  • Eagle (2006);
  • CEFLES2 (2007);
  • Sentinel-3 experimental campaign (2009).
    Magellium is involved in the campaign dataset analysis in order to assess their suitability for the validation of S2 products and for the generation of reference data used for the validation. This analysis addresses both spectral, geometry and radiometry data quality aspects, which are assessed as follow:
  • For geometry, co registration accuracy between overlapping images is assessed to estimate a relative accuracy. Geolocation accuracy is estimated for pair of images acquired by the same sensor at different dates or different altitudes or by different sensors, over various landscapes, agricultural, forest. The mean bias have been estimated for a large set of selected images acquired during the campaign and allows to assess the geometric quality each campaign.
  • For radiometry, different tests are undertaken. Airborne data are compared to ground measurement over a large sample of location for each campaign. Airborne data are also compared between each other for data acquired over common area (even if no ground measurements are present) in order to control the capabilities to produce mosaics. The quality of the processing is also discussed, i.e. the atmospheric correction.
    A synthesis and recommendations will be presented regarding Sentinel 2 mission requirements.