Abstract
Validation is mandatory to quantify the reliability of satellite biophysical products that are now routinely generated by a range of sensors. This paper presents the VALERI project dedicated to the validation of the products derived from medium resolution satellite sensors (www. avignon.inra.fr/valeri/). It describes the sites used, and the methodology developed to get the high spatial resolution map of the biophysical variables considered, i.e. LAI, fAPAR and fCover that can be estimated from ground level gap fraction measurements. Sites were selected to represent , with the other validation projects, the large variation of biomes and conditions observed over the Earth's surface. Each site is about 3×3 km² in size and should be flat and relatively homogeneous at the medium resolution scale. For each site, the methodology used to generate the high spatial resolution biophysical variable maps is described. It is mainly based on concurent use of local gound measurements and a high spatial resolution satellite image, generally SPOT-HRV. Local ground measurements should be representative of an elementary sampling unit (ESU) that has approximately the same size as a SPOT-HRV pixel. The ground measurements mainly consist of gap fraction measurements achieved with LAI-2000 or hemispherical photographs. The ESUs are selected over the whole 3×3 km² site in order to sample the range of vegetation types observed. A transfer function is subsequently established over the ESUs to relate the ground measurements of the biophysical variables considered to the correspodonding high spatial resolution satellite image data. Finally, co-kriging is applied to generate the high spatial resolution map of the biophysical variables over the 3×3 km² area. The methodology presented in this paper can serve as a basis for validating medium resolution satellite products. These methodological aspects are discussed and conclusions drawn on the limitations and prospects of beforementioned validation activity.