In this paper, a portable three-dimensional (3D) scanning system for the accurate characterization of large raw material (e.g., cereal grain, coal, etc.) stockpiles is presented. The system comprises an array of high resolution millimeter-wave radars and a cm-level accuracy positioning system to accurately characterize large stockpiles by means of a high-resolution 3D map, making it suitable for automation purposes. A control unit manages the data received by the sensors, which are sent to a computer system for processing. As a proof of concept, the entire sensor system is evaluated in a real environment for electromagnetically scan a scaled stockpile of coal, used in the industry for handling raw materials. In addition, a highly efficient processing adaptive algorithm that may reconstruct the scanned structure in real-time has been introduced, enabling continuous dynamic updating of the information. Results are compared with those from a photogrammetry-like technique, revealing an excellent agreement.