(1) the accuracั of detecting the collapsed buildings in terms of pixels based on the two scenarios of the difference DEMs are rather consistent and the overall accuracy obtained is better than 94%. (2) However, in both difference DEMs, the user’s and producer’s accuracies of detecting the collapsed buildings are lower than those of detecting the buildings that have not collapsed. The reason for this is that there are far more pixels of buildings that have not collapsed than pixels of collapsed buildings.