Landcover information is a critical parameter for environmental assessment but difficult to collect and update using conventional approach. This paper demonstrates a systematic approach of using remote sensing images to distinguish different types of landcover automatically or semiautomatically is a more economical and effective approach for environmental assessment. focuses were placed on utilizing multiple space-borne images to produce and update landcover maps in different scales of entire Taiwan. For moderate resolution requirement, landcover types were classified into several categories of interest, such as water, forest, grass, soil, and buildings etc. using SPOT-5 and FORMOSAT-2 images. Classification results are aggregated into cells in specified resolutions storing percentages of individual types. For regions of special interest, advanced analysis algorithms are applied to hyperspectral and high resolution images to produce detail vegetation maps. The landcover monitoring results serve as the valuable input parameters of environmental assessment modeling.