DEVELOPMENT OF VALIDATION DATA SETS FOR GLOBAL LAND COVER CLASSIFICATION USING ALOS/AVNIR-2 DATA

The GCOM Climate first generation satellite (GCOM-C1) project started in October 2009 and each research has proceeded to the aim of project. Our final aim in GCOM-C1 project is producing global land cover data sets that provide much useful information for the study of global environmental changes such as global biogeochemical cycles and climate change, and for the estimation of net primary production. In this study, we proposal a land cover class structure of global land cover, and apply to produce validation data sets for the land cover classification system with low resolution data using high resolution satellite data sets, ALOS/AVNIR-2 data sets. To determine the classification conditions in our classification system, we used the universal pattern decomposition method (UPDM) coefficients and the modified vegetation index based on the UPDM (MVIUPD). As validation data sets, we produced vegetation coverage degree rank data sets and dominant class data sets. Comparisons between the ground truth data sets and the results of our land cover classification system with ALOS/AVNIR-2 showed many areas where classes agreed. Using the classification results, validation data sets were produced.