Ground illumination varies with an angle of the sun to the ground surface. A varied illumination shound be corrected, therefore, in order to obtain an accurate classification of fallen snow surface at a mountanious district using Landsat MSS data. This requires us to obtain the reflectance from the illumination of the snow surface. First, an equation, of which variables derived from the ratio between the direct sun radiation and the diffuse sky radiation, as well as from the angle of the sun to the ground surface, should be introduced. The reflectance will then be estimated through the illumination of the snow surface, should be introduced. The reflectance will then be estimated through the illumination of the snow surface obtained from Landsat data, after finding the optimum coefficients for the equation. Yagizawa Dam Watershed was chosen for this experiment and the digital terrain model was constructed. An analytical study was conducted by having the Landsat MSS data corresponding with the digital terrain model. In deciding the coefficients for the equation, an interactive method was utilized with a color TV monitor. The ratio between the direct sun radiation and the diffuse sky radiation was gradually altered until the brightness on each side of the ridge on the imagery monitor becomes equal. A classification of the snow surface reflectance derived from the new data corrected using the coefficient showed that the classified patterns correspond with changes of contour lines. In conclusion, a geographical correction is necessary to obtain correct illumination from Landsat original MSS data. For more detailed classification of the snow quality, the relation between the classification pattern of snow surface reflectance and snow surface condition shoud be proved and it remains as a future subject.
