| Sumario: | Imagery provided by the NOAA Advanced Very High Resolution Radiometer (A VHRR) has proved to be very important for studying the dynamics of the Earth surface at global and regional scales. The success of the use of A VHRR imagery for global and continental studies spurred its application to finer scales. This implies the integration of the low-spatial, high-temporal resolution imagery, such as AVHRR and VEGETATION, and the high-spatial, low-temporal resolution imagery, such as SpotHRVIR or Landsat-TM. In SPATEM we use a segmentation-based classification for a down-scaling of the spatially-coarse multi-temporal imagery, with particular interest on Mediterranean forests and shrublands. In the pre-launch phase we have used AVHRRlkm multi-temporal imagery and a LISS-ID as simulations of, respectively, VEGETATION and HRVIR imagery. We have produced a classification of a Mediterranean forested area by means of segmentation, hierarchical classification and canonical analysis. We have used the classification to select pure AVHRR-lkm temporal signatures of NDVI for some of the classes and found that the ordering from higher to lower values of annual NDVI is coincident with the ordering of the same classes along a greeness axes produced by the canonical analysis of the LISS-ID image. We conclude that consistent differences in temporal NDVI are detectable with A VHRRlkm data for different types of Mediterranean forests. Such differences are likely to be more evident using SPOT-4 VEGETATION and HRVIR data, which improved radiometric and geometric characteristics will allow for an spectral mixture analysis.
|
|---|
