The Use of Classified Landsat-5 Thematic Mapper Imagery in the Characterization of Landscape Composition: a Case Study in Northern England

Abstract Earth-orbiting remote-sensing satellites are becoming increasingly used as sources of land cover data in land use planning and resource inventory. The present study compares land cover data derived from a maximum-likelihood classification of spectral reflectance data from the Landsat-5 Thematic Mapper (TM) sensor with that obtained by field survey of a sample of 169 1-km2 grid squares in northern England. Selection of grid squares was stratified within classes of an environmental landscape classification. Landscapes included in the study ranged from coastal and urban, through pastoral and arable, to upland forestry, bogs and moorland. The land cover composition of Landsat and field cover maps was compared at the 1-km grid-square resolution. Results were interpreted with the aid of an a priori comparison of the definitions of the cover types recognized in the two survey methods. The study demonstrates that classified TM data can provide accurate estimates of the land cover composition of 1-km squares. Differences between the cover maps derived from Landsat and ground survey often reflected the subjectivity of cover type definitions, although mis-classification of pixels falling on complex boundaries and linear features was noted. The use of remotely-sensed data in land use research is discussed in light of the results and it is concluded that the Landsat cover classification tested in the present study has great potential in the fields of land use planning and environmental monitoring.