Abstract
The satellite snow index, consisted of visible, infrared and middle infrared channels, was applied to the 10-day composite SPOT-VEGETATION data. The noises on the images, such as cloud, haze and system noise, were reduced by applying the LMF processing. The daily snow depth data of the AMEDAS dataset were collected as the ground information and the snow cover data of every 10 days was created from them. It shows that the area where snow covers less than 30 days is not detected by the 10-day composite data. Although the correlation coefficient between SPOT and the ground data was 0.90, the satellite data had the trend to overestimate the duration of snow cover, and, on the contrary, they could not detect snow in some places where snow coved for even a few months.
