Forest site types were interpreted with a maximum likelihood classification of Landsat imagery. On the basis of such interpretation, an operational system for site type classification was developed. A spectral site type model was also created. The study area was situated in northern Finland near the Arctic Circle. The numerical interpretation was tested by random sampling. Some classes were accurate enough for an operational system withoutfield checking. About two- thirds of the area needed more information. The new forest site type classification method, entitled KAUKO, is a multi-phase system. The first phase is the interpretation of Landsat MSS imagery, followed by utilization of base map data, uisual interpretation of color infrared photographs, and field checking. With KAUKO it was possible to make the forest site type classification three times as fast as with the traditional method. More than one quarter of the expenses of this classification were saved by using the KAUKO method in test classification.
|Journal||Photogrammetric Engineering and Remote Sensing|
|Publication status||Published - 1984|
|MoE publication type||A1 Journal article-refereed|