MAP-GUIDED CLASSIFICATION OF REGIONAL LAND-COVER WITH MULTITEMPORAL AVHRR DATA

Cartographers often need to use information in existing land-cover map s when compiling regional or global maps, but there are no standardize d techniques for using such data effectively. An iterative, ''map-guid ed'' classification approach was developed to compile a spatially and thematically consistent, seamless land-cover map of the entire Intermo untain Semi-Desert ecoregion from a set of semi-independent subregiona l maps derived by various methods. A multi-temporal dataset derived fr om AVHRR delta was classified using the subregional maps as training d ata. The resulting regional map attempted to meet the guidelines of th e proposed National Vegetation Classification System for classificatio n at the alliance level. The approach generally improved the spatial p roperties of the regional mapping, while maintaining the thematic deta il of the source maps. The methods described may be useful in many sit uations where mapped information exists but is incomplete, has been co mpiled by different methods, or is based on inconsistent classificatio n systems.