Compilation and evaluation of a small-scale land resource map

A small-scale land resource map was compiled by combining generalised large-scale maps with auxiliary data and interpreted satellite images. The large-scale maps covered areas below the treeline and were generalised using an expansion-contraction algorithm (ECA). Areas above the treeline were added by manual interpretation of a Landsat TM image (Band 4,5,3). Certain features were also copied from small-scale topographic maps. The thematic accuracy was evaluated individually for different segments of the map, according to the production technique. The general accuracy was satisfactory in areas where the ECA technique had been applied. Areas mapped using auxiliary data were rendered inaccurately both below and above the treeline, but the result was still found acceptable. The results from satellite interpretation above the treeline were promising, but need to be improved. The image interpretation can either be refined by using alternative sensors, by improved training of the interpreters, or by adjusting the definition of the land resource classes in order to attain a better match between the class definitions and the result of the classification.