The mosaic was prepared from S10 data acquired throughout the year 2000 and freely available from the free ftp site of the VEGETATION programme at http://free.vgt.vito.be/
S10 data include 10-day composites of the spectral channels based on the Maximum NDVI value (i. e. the Top of canopy reflectance values of the day when NDVI is highest during the 10-day period are retained for the S10 synthesis).
The mosaic was prepared using a simple, quick and dirty approach. The S10 data can still retain from time to time cloud contaminated and hazy measurements as well as cloud shadows. In order to minimize this a principal component analysis was applied on the red, NIR and SWIR channels separately, i. e. for each channel the successive acquisitions throughout the year were used as input to the Principal Component analysis. Each time only the first component was retained, based on the assumption that throughout a long period of time atmospheric effects (are marginal and thus the most frequent situation is “clear sky” observation. In actual fact for the mosaic the 36 observations of the year 2000 were retained only for South America, Africa, Southern Asia and Australia, whereas for North America and Northern Eurasia only the 18 observations of the Northern hemisphere Summer were taken into consideration. The two areas were stitched together through a cutting line that minimizes visual effects of the stitching: i. e. through seas, ocean and desert areas as much as possible. It can be noticed in some areas, in particular of equatorial regions, that some haze can still be perceived: this occurs when atmospheric conditions are such that cloud cast sky is at least as frequent as cloud-free sky.
The first-components derived for each channels were then reduced to 8 bit and combined into a colour composite where SWIR is assigned to blue, IR to green and red to red in order to depict vegetated areas in green tones.
The product can facilitate the perception of how landscapes look like as seen from the satellite.
For more information on this topic, please contact Etienne Bartholomé