AbstractsGeography &GIS

Facing environmental issues in a global change context requires improving our knowledge about the land surface and its dynamics. Today, the accumulation of satellite time series at moderate resolution constitutes unique datasets to monitor Earth’s surface evolution on each squared km. The ambition of this thesis was to exploit the multispectral reflectance variability as observed by the SPOT-VEGETATION time series (1999-2011) to characterize land surface dynamics at multi-annual and seasonal scales. Because of their single year strategy, current global land cover maps are sensitive to temporary events and to year-to-year variation related to meteorological and phenology change. An innovative land cover mapping approach takes advantage of the long time series to derive a multi-year global land cover map. This was found efficient to reduce the sensitivity to the observation period found in all global annual maps. In addition, a change detection method capitalizes on the land cover trajectories over the whole time series to highlight forest, cropland, grassland, bare area, urban area and water losses and gains. It provides for the first time a complete portrait of the Earth’s land surface transformations systematically described by the year of change and the land cover transition. At seasonal scale, vegetation phenology of tropical rainforest is prone to a large debate about the ability of remote sensing to monitor any vegetation cycle due to atmospheric contamination and more recently viewing and illumination geometry variation. Capitalizing on the 13-y reflectance values, this thesis proves that the seasonal dynamics of appropriate vegetation index are driven by leaf phenological events, themselves responding to light availability in order to enhance photosynthetic gain, in both Amazon and Congo basins. (AGRO - Sciences agronomiques et ingénierie biologique)  – UCL, 2016 Advisors/Committee Members: UCL - SST/ELI/ELIE - Environmental Sciences, UCL - Ingénierie biologique, agronomique et environnementale, Gond, Valéry, Verbeeck, Hans, Mayaux, Philippe, Bogaert, Patrick, Defourny, Pierre, Delvaux, Bruno.