Science et changements planétaires / Sécheresse


Land cover characterization using MODIS-TERRA remotely-sensed data in Tunisia Volume 23, issue 2, Avril-Mai-Juin 2012


See all figures

Université Laval 2325, rue de l’Université Faculté de foresterie, de géographie et de géomatique G1V 0A6 Québec Canada, Cirad UR 105 « Biens et services des écosystèmes forestiers tropicaux » Campus de Baillarguet TA-C105/D214 34398 Montpellier France

Although many uncertainties remain about the speed, the magnitude and the geographical distribution of climate change, there is now a consensus within the scientific community as to its reality. The occurrence of drought and degradation of vegetation cover and wetland areas on all continents underline the importance of this phenomenon. North African countries are among the most vulnerable regions due to their special geographical situations limited by the Sahara in the south and the sea in the north. In the context of global change, spatial-temporal monitoring of the surface state in Tunisia will be a good source for policy makers and scientists. Weekly MODIS-Terra Satellite images purged of atmospheric effects, cloud and its shadow, and with good temporal and radiometric resolutions are good tools for time-tracking land cover. Hence, unsupervised and supervised classification methods are used to classify them, identify changes and adjust the time series reflecting the evolution of the area occupied by land cover from 2000 to 2009. Spectral analysis and digital filtering were used to show that the temporal evolution of these classes is essentially annual and closely tied to rainfall. However, we distinguish the existence of large-scale variability that is not too strong in the order of 8-9 years within the 10-year time series obtained.