干旱区生态环境知识资源中心

The knowledge of land cover is a prerequisite for efficient agricultural water management and sustainable development of semi-arid areas. Remote sensing is the only feasible mean of providing this information with a consistent space and time basis. In this context, the objective of this communication is to present a first analysis of high resolution and high repetitivity FORMOSAT-2/RSI images. The work is performed from a time series of FORMOSAT-2 images acquired over the semi-arid Tensift-Marrakech plain during the 2005/2006 agricultural season. The main characteristics and the processing of these data are firstly presented. NDVI space time variations are analysed for the dominant land classes over an irrigated area where cereal and non irrigated lands are dominant Finally, the first results obtained using a decision tree with criteria applied on NDVI time series averaged by fields are discussed.