EXTRACTING CROP RADIOMETRIC RESPONSES FROM SIMULATED LOW AND HIGH-SPATIAL-RESOLUTION SATELLITE DATA USING A LINEAR MIXING MODEL

The images from optical sensors with a broad path width (e.g. NOAA-AVHRR) are used for monitoring vegetation on a regional scale. The European agricultural land uses, which are generally heterogeneous, can be coarsely distinguished by these radiometers. Such sensors, however, do not allow the discrimination of seasonal radiometric changes of a given crop. Some future Earth observation platforms will carry two types of instruments on board. The first instrument will have moderate spatial resolution but a broad path width to allow almost daily observations of the emerged areas. The second will have high spatial resolution and a narrow path width to give the opportunity of making land use thematic maps from the few images recorded per year. The combination of these two types of data allows the medium resolution signal to be unmixed in order to restore the radiometric evolution of a particular crop or of a group of crops. From the application of a linear mixing model to coarse spatial resolution data, this article presents an unmixing method based on the techniques of multiple regression. This approach has been applied to a simulated coarse resolution dataset to calculate the spectral response of the mixture components. The most promising results of this first study encourage us to assess the method with real images (e.g. NOAA-AVHRR). Additionally, the results can be seen as an argument in favour of the complementary use of these two types of optical instrument.