Lake bathymetry can provide abundant information for basin planning and governance, watershed erosion and siltation management, water resource utilization, as well as environmental protection. In particular, lake bathymetry is essential for lake development and conservation, and is also closely related to sediment deposition and removal. A number of sounding techniques, such as single-beam sonar sounding, multi-beam sonar sounding, and unmanned ship sounding, have been applied to extract underwater topography of lakes. These techniques, albeit with high measurement accuracy, are time-consuming and costly. Therefore, it is of great need to develop a simple and cost-effective method. To reach this goal, recent advances of remote sensing technologies provide an alternative means of accurately measuring lake bathymetry information. Taking Dongting Lake as the study area, this paper estimated the lake bathymetry through extracting the boundaries of lake areas using multi-temporal Landsat and MODIS image series. In particular, the water level corresponding to each reference point of the lake boundary is retrieved based on trend surface analysis and kriging interpolation technology. Then the water levels of these points are regarded as the elevation points to retrieve the landform of the lakebed of the Dongting Lake. The reliability and accuracy of the developed methods were assessed through the comparsion with the actual measurements. Results indicate that the kriging interpolation technology performs well, with the average of cross-validated error target less than 0.2 meter, and the retrieval error of lake boundary with more reference points less than 1 meter. This paper suggests that an improved method for accurately and rapidly extracting lake bathymetry information can be achieved through examining the changes of water levels. This method might be of great significance for further study on lake evolution, lake development planning, and water ecological protection.