In recent studies several factors such as landslides, volcanoes, glaciers, tectonic factors and karst forms such as flooded pulleys resulting of karsts development cause the formation of lakes in mountainous areas. The mountain lakes of Iran, including Tar and Havir lakes, are mainly formed by landslides. In this research investigation of water basins changes for (Tar and Hoyer lakes) by remote sensing method in the period from 2013 to 2022 have been done. It has been tried to select appropriate Landsat 8 satellite images in terms of cloud cover and image quality from the time when the lakes are full of water. The training and testing data were selected with the same distribution in the whole image to make the classifications more accurate. After estimate kappa coefficient and overall accuracy for all kind of supervised and unsupervised methods, the results obtained from the two classification methods of maximum likelihood and support vector machine with linear kernel from Tar and Havir lakes, the conclusion was that the classification method of support vector machine with linear kernel have better distinguish land and water areas and it is the best method for classifying water basins located in mountainous areas. The amount of water in Tar and Havir lakes had been increased from 2013 to 2017but in 2018, the water area of ​​both lakes decreased significantly, and in 2019, the largest water area for the lakes was observed and calculated. In 2020, there is no big change in the water area of ​​the two lakes, and from 2020 to 2022, the amount of water in the two lakes is decreasing.