Tool grinding is an essential process for the production of cemented carbide tools. In that context, the investigation of specific effects like the resulting surface profile and the fluid dynamic processes is of great interest, but requires microscopic modeling of the grinding wheel including its individual grains and bonding material. This paper introduces an approach for a parametric grinding wheel model, which provides a topography on microscopic scale depending on the grinding wheel specification and dressing conditions for subsequent use in material removal simulations. Scalable abrasive grains and variable distributions embedded in a shiftable bond layer are applied. Optical laser scans are used to derive surface parameters for an adaption and evaluation of the model. The prediction quality in terms of surface roughness is evaluated in surface grinding reference experiments.