A random forest regression model is used to predict percent mud/sand/gravel content for the North Sea and UK shelf. Predictions are made using outputs from hydrodynamic models as well as optical remote sensing data from satellite platforms and bathymetric variables, which are mainly derived from acoustic remote sensing. GeoTIFFs with a 500 m resolution of mud/sand/gravel fractions are given as well as substrate classes (EUNIS and Folk).