Standard land-cover classification scheme for remote-sensing applications in South Africa

Abstract

For large areas, satellite remote-sensing techniques have now become the single most effective method for land-cover and land-use data acquisition. However, the majority of land-cover (and land-use) classification schemes used have been developed around specific user objectives (namely, agriculture and conservation), and are often influenced by geographical location and actual data capabilities. Thus very few are directly comparable. At present, no universally accepted standard classification system exists for remote-sensing studies in South (or southern) Africa. There is thus an urgent need for a predetermined framework for land-type classifications that will provide standardized, baseline specifications to ensure consistency and conformity in map data produced by various organizations from satellite imagery. This paper presents a standard hierarchical framework for the classification of remotely sensed data, designed to suit the South African environment. The framework is based on known land cover classes that can be derived from high-resolution, remotely sensed data such as SPOT or LANDSAT Thematic Mapper (TM) imagery, and links, as far as possible, to existing classification systems or codes that have been used within various organizations.