Generalization of Landsat MSS interpretations of aquatic areas in southwestern Finland

Abstract

The digital remote sensing classification procedure applied to the lakes Pyhäjärvi (near Säkylä), Koskeljärvi and Köyliönjärvi in southwestern Finland is based on spectral reflectances. The aquatic surfaces were divided into five categories consisting of eleven classes connected with water depth, water turbidity, nature of the bottom and the aquatic vegetation. Generalization of this classification over other water areas of different types allowed the authors to evaluate the usefulness, value and accuracy of the classification. All water areas, although representing different types, were rationally categorized into five main types, although some pairs among the eleven classes were slightly confused, or became ambiguous. Hydrolittoral and aquatic Landsat MSS remote sensing seems to be useful for parametric mapping under circumstances in which field data collected from a small number of reference areas are to be extrapolated to apply to other areas within a single MSS frame. Some supplementary test field inspections and background information would be required, however, in order to guarantee that the primary reference areas were representative enough, or if not, to indicate the variety of aquatic areas classified together. With these restrictions, the machine-processed remote sensing would result in a practical and economical mapping procedure which should, as a next step, be developed for temporal studies in view of the longterm nature of the Landsat project. © 1986 D. Reidel Publishing Company.