The successful launch of FORMOSAT-2 on 21 May 2004 has proved the concept that the temporal resolution of a remote sensing system can be much improved by deploying the high-spatial-resolution sensor in a daily revisit orbit. The sun-synchronous and geo-synchronous orbit makes FORMOSAT-2 an ideal satellite for site surveillance. The unique orbit and the arrangement of the CCD lines on-boarded FORMOSAT-2, however, has also raised new challenges in image processing. This paper describes the procedures that are employed by the Disaster Prevention Research Center at National Cheng-Kung University to process FORMOSAT-2 imagery. The major focus is to develop a fast and automatic approach to process a large amount of FORMOSAT-2 images for the purpose of site surveillance, including level-2 product generation, pan-sharpening and quasi-orthorectification. One example of sequential threeday images of Hsin-Chu taken by FORMOSAT-2 with results of automatic change detection is given and discussed. Experience obtained from this work would benefit the system design and image processing of the future satellite missions with similar specifications, such as the Pléiades HR scheduled to be launched in 2008.