Abstract: In remotely sensed monitoring of sand mining in river channel,the area with high concentration of suspended solids in water body is usually classified as suspected area according to empirical threshold. The segmentation threshold is difficult to be automatically and objectively determined due to the accuracy limitation of atmospheric correction and suspended solids concentration inversion mode. A method of automatically recognizing abnormal turbid water body based on dominant wavelength of water color is proposed. The dominant wavelengths of water color information of remote sensing images of red,green and blue bands are extracted using Sentinel-2 and Landsat8 satellite image data based on CIE-XYZ color system. Furthermore,the dominant wavelengths of water color parameters indicating the concentration of suspended solids are automatically classified into high,medium,and low levels by using the natural breaks method,and the high value area is extracted as the suspected area of sand mining. The method was applied to reflect the temporal and spatial distribution of turbid water bodies,and had successfully identified the sand mining in the Jiuzhou River and Huazhou Bay waters of the Hedi Reservoir in Guangdong in 2016-2017 and in the Xunwushui River of the Fengshuba Reservoir in Guangdong from November 2019 to April 2020. Experiments have proved that the method is of strong resistance to atmospheric interference,high automation,stability and reliability,and is suitable for multi-source remote sensing images. The method also offers timely and effective tool for the temporal and spatial distribution of sand mining suspected areas for traditional manual inspections and supervision,and improves the capacity of normalized sand mining supervision.

Key words: sand mining supervision, remotely sensed monitoring, dominant wavelength, water color parameter, natural breaks