The damaged areas of the 2003 Bam, Iran, earthquake were detected using 15-meter-resolution satellite imagery acquired by Terra-ASTER. First, fluctuation of digital numbers was modeled as a normal random variable based on 17 pre-event images on a pixel-by-pixel basis. Then, the deviation value of each digital number in the post-event image was evaluated and converted into the confidence level, which indicates the possibility of an abnormal groundsurface change. The detected damaged areas were verified with a highresolution satellite image and it was observed that the areas with earthquake influence were mostly identified. However, the pixels with significant change were induced not only from heavily damaged buildings but also dusty roads, possibly due to demolition work. It was suggested that prior knowledge like a high-resolution pre-event image would assist the interpretation of the detected result.
