Multi-date Satellite Images for Reducing Classification Confusion between Deciduous Forest and Evergreen Forest

This paper used the Landsat images to do the maximum likelihood classification. Firstly, we used one early Sept. image to do the maximum likelihood classification. There is significant confusion between deciduous forest and evergreen forest. Then we used early Sept. and late Oct. images together to do the maximum likelihood classification. Due to the apparent different spectral signature in the two-date images, we find after using this new method, the confusion between deciduous forest and evergreen forest is reduced significantly. In order to better assess the accuracy, we used the revised area estimates, and provided the 95 percent confident intervals.