The mapping of nearshore bathymetry based on spaceborne radiometers is commonly used for QC ocean colour products in littoral waters. However, the accuracy of these estimates is relatively poor with respect to those derived from Lidar systems due in part to the large uncertainties of bottom depth retrievals caused by changes on bottom reflectivity. Here, we present a method based on mini unmanned aerial vehicles (UAS) images for discriminating bottom-reflected and water radiance components by taking advantage of shadows created by different structures sitting on the bottom boundary. Aerial surveys were done with a drone Draganfly X4P during October 1 2013 in optically shallow waters of the Saint Lawrence Estuary, and during low tide. Colour images with a spatial resolution of 3 mm were obtained with an Olympus EPM-1 camera at 10 m height. Preliminary results showed an increase of the relative difference between bright and dark pixels (dP) toward the red wavelengths of the camera's receiver. This is suggesting that dP values can be potentially used as a quantitative proxy of bottom reflectivity after removing artefacts related to Fresnel reflection and bottom adjacency effects.