Detection of quartzite blocks in the River Rhine: Development of semi-automatic and automatic approaches for the detection of boulders

To detect quartzite blocks in the riverbed of the River Rhine near Düsseldorf, a highresolution multibeam echo sounder (MBES) survey, delivering bathymetry as well as backscatter data, was carried out. The first visual analysis of the retrieved dataset revealed more than 8,600 potential quartzite blocks. To enhance and automate the manual detection process and to obtain additional information about the boulders, two approaches, being a GIS method as well as an AI-approach using a convolutional neural network, are presented.