The accurate extraction of mountain river information is highly significant in water resource investigation and ecological environment protection. However, there are some problems in the existing methods of river information extraction, which are mainly the interference from shadows and buildings. Such interference leads to erroneous and redundant extraction of river information, which leads to inaccuracy or incompleteness. In this study, a precise extraction method of mountain river information is established using Landsat8 image and digital elevation model. The main steps of the river extraction method are as follows: (1) we propose the optimized spectral threshold water index to extract river information; (2) based on digital elevation model data, we simulate the mountain shadows of the study area to remove interference from them; (3) we establish the buffer zone of the river network using digital elevation model data to solve the problem of redundant extraction of river information; (4) we separately calculate and then standardize land surface temperature, albedo, and normalized different building index. The effects of buildings near the river are removed. Results show a relative accuracy of 97.52%. The new method decreases the interference of mountain shadows and buildings.

The accurate extraction of mountain river information is highly significant in water resource investigation and ecological environment protection. However, there are some problems in the existing methods of river information extraction, which are mainly the interference from shadows and buildings. Such interference leads to erroneous and redundant extraction of river information, which leads to inaccuracy or incompleteness. In this study, a precise extraction method of mountain river information is established using Landsat8 image and digital elevation model. The main steps of the river extraction method are as follows: (1) we propose the optimized spectral threshold water index to extract river information; (2) based on digital elevation model data, we simulate the mountain shadows of the study area to remove interference from them; (3) we establish the buffer zone of the river network using digital elevation model data to solve the problem of redundant extraction of river information; (4) we separately calculate and then standardize land surface temperature, albedo, and normalized different building index. The effects of buildings near the river are removed. Results show a relative accuracy of 97.52%. The new method decreases the interference of mountain shadows and buildings.