Scaling-up of land cover information by using multiple resolution satellite data

The authors conducted a feasibility study for scaling-up of land cover information by the combination of multiple resolution satellite data like Landsat-TM and NOAA-AVHRR. The scaling was performed by extrapolating the information about the proportions of land cover classes obtained from higher resolution data (TM) into the wide coverage area of lower resolution data (AVHRR). Therefore, a mixture analysis is necessary to be performed between the pixel-wise spectra of AVHRR data and the land cover proportions obtained by classified TM data. The category decomposition approach was employed in the mixture analysis because it was proved to be superior to the regressive approach by a preliminary study using a simulated multiple resolution data set from TM. Then the estimation accuracy of land cover proportions by the category decomposition method was examined using TM and AVHRR data obtained on the same day to assess the feasibility of accurate scaling-up of land cover information by the combination of TM and AVHRR data. (C) 2000 COSPAR. Published by Elsevier Science Ltd.