Time Series of High Resolution Satellite Data as a Means to Produce and Validate Biogeophysical Parameters

The retrieval of biogeophysical parameters time series from space has up to now concerned mostly low or medium resolution satellite data, the highest resolution being on the order of 250 m with MODIS or MERIS observations. However, the forthcoming of new satellite data at high resolution and high revisit frequency such as the Sentinel-2 satellite from the European Space Agency (ESA) may induce new efforts to process time series of biogeophysical variables at higher spatial resolution. Sentinel-2 will produce data at 10-20 m resolution in the visible and near infrared and data at 60 m resolution in the blue and middle infrared with a wide field of view allowing a repetitiveness of 10 days with 1 satellite and 5 days with two satellites at the Equator. The expected launch dates of the two Sentinel-2 satellites are 2014 and end of 2015. Orthorectified top of atmosphere reflectances will be delivered globally free of charge within 24 hour delay after acquisition by ESA. The Sentinel-2 data will offer great opportunities to generate time series of biogeophysical variables exhaustively over the world with high spatial resolution, in particular Essential Climate Variables over land such as the Leaf Area Index (LAI) or the Fraction of Absorbed Photosynthetically Active Radiation (FAPAR). Algorithms are currently being developed to transform such data into products ready to be used by the community. To validate the corresponding products it is important to build appropriate satellite data sets of high spatial resolution and high temporal frequencies well before launch along with the associated ground measurements. CNES, CESBIO and INRA have been active these last ten years in building such data sets. Over the last ten years, the Kalideos program of CNES has provided time series of SPOT data over 3 sites in France and 1 site in Romania to the scientific community. CESBIO has gathered over 800 Landsat images and 400 Formosat-2 images (multispectral imagery at 8 m resolution, 1 day repetitiveness) over the South of France for the years 2002, 2003, 2006 to 2011. These data have served as a platform for the test and validation of an atmospheric correction scheme based on both the multispectral and multitemporal dimensions of the signal. Their use for the validation of biophysical variable retrieval has also started. To anticipate the massive coming of these high spatial resolution time series of biophysical products, a validation strategy should be designed and implemented. It should capitalize on the experience gained with the validation of medium spatial resolution products, in order to reach stage 4 of the validation, i.e. "Quantify uncertainties in the product and its associated structure with systematic updates when new product versions are released and as the time-series expand". For this a collection of satellite data and ground data on a variety of sites and ecoclimatic situations should be progressively built with an easy and transparent access for the community. The data sets described above could be part of such a collection.