This paper propose the visualization of large landscape from the fusion of simplified versions of digital elevation models and satellite photos. Such approximations are extracted from independent hierarchical quadtrees, that organize successive subdivisions of the two-dimensional space imposed by criteria based on local curvature and color standard deviation (at terrain and image, in that order). The approximations are selected by a heuristic based on multicriteria analysis that guarantee the accurate visual quality for the available computational resources. In addition, the paper include an example application on actual large elevation models and satellite photos.