Surface water monitoring is the foundation of water resources management. However, large-scale surface water monitoring based on long-sequence Landsat images is easily affected by cloud cover. In this study, taking the Tonle Sap Lake in the lower Mekong as an example, a series of water-land boundaries with height attributes based on water level data were extracted from multi-temporal Landsat images without cloud cover using the optimal water index and the lakeshore elevation model (LEM) of Tonle Sap Lake was constructed. On this basis, with the help of the limited land-water boundary provided by the Landsat image over a longer period, long-term monitoring of the changes in the water area of the entire lake area was realized. The results show that the Landsat image data in conjunction with the LEM can effectively remove the effect of cloud cover in monitoring the change of long time series water bodies. Reconstruction of historical data on changes in the waters of Tonle Sap Lake with the proposed method shows that in the past 30 years, there has been no significant change in the water areas of Tonle Sap Lake in either the dry or rainy seasons. Water area changes of the lake have a delayed response of 1-3 months to changes in the water level of the Mekong River, which is more affected by the mainstream recharge. The water area of the lake is unaffected in the dry season; however, it is weakly affected in the rainy season by the Xiaowan and Nuozhadu hydropower stations in China.