On May 12th, 2008, amagnitude ofRichter8.0 earthquake attackedWenchuan in SichuanProvince, China,which caused fatal death andmillions ofpeople homeless. The landslides caused by earthquake damed some riverways in themountainous areas, forming tensofquake lakes, amongwhich the Tangjiashan Quake Lakewas the largestone and the mostdangerous to threaten the downriver residents. Therefore, it is urgent to find solutions to solve the problem of the quake lake forgovernment. Fastand precisely calculation ofquake lakewater capacity is a key issue for solving the threat of it. Traditional capacity calculationmethods, taking thewater surface as a planewith a givenwater leve,l just consider the static capacity inwhich thewater level is required to bemonitored for a long tmi e. Complex hydrodynamicmodel and complicatedmodel inputs, which highly demand the computing resource and are compute-intensive, must be taken into considerationwhen calculating dynamic capacity. It cannot satisfy a fastprocess dealingwith the emergence.In thispaper, a pixel-based curvedwater surfacemethod isproposed to calculate the total capacity including dynamic capacitywith a natural curved water surface, and the technologies of Geographic Information System (GIS) and remote sensing are integrated in thismethod. The backwater area is extracted from a corrected geometrically SPOT-5 mi age, and then the riverway DEM is intersected to obtain the boundary ofwaterbody consisting ofa setofpoints. A 3D curvedwater surface, which depicts the large ratio of slope forwater surface in mountainous areas, can be draw out by interpolating these concrete points. Finally, capacity calculation is carried outbased on a calculationmodelafter interpolating the point boundary to a curvedwater surface. Themodel takes the space between the pixels onwater surface and the riverway DEM as an underside-squared prism. The volume composed of riverway DEM and water surface are divided intomany small prisms to calculate the capacity respectively. The totalwater capacity of the quake lake is obtained by summing up these volumes.Considering the calculation efficiency of thismethod, an mi proved, pixel-based river segmentation algorithm is pro-posed in this study. It is fully considered the rationality ofpixel-based curvedwater surfacemethod and the efficiency of the traditional static capacity calculationmethod. An expermi ent is carried out to mi plement these algorithms. At last, a comparison, among theArcGISmethod, the proposedmethods including the pixelbased curvedwater surfacemethod and the mi proved pixel-based river segmentationmethod, is conducted to prove that the mi proved method ismore acceptable for its calculation speed and accuracy. And itcan be used fordaily operation formonitoring the capacity ofa lake or reser-voir.