Accurate calibration of contact parameters in the discrete element method (DEM) remains one of the critical challenges for predictive modelling of granular flows and wear processes. In this work, we propose and experimentally validate a novel calibration strategy based on the morphological analysis of granular heaps formed under controlled low-mass deposition. Operating deliberately in the sensitivity-threshold regime, where reproducibility is preserved while local interactions remain highly distinguishable, we demonstrate that the geometry of the heap provides diagnostic information beyond the conventional single-angle-of-repose test. A systematic series of experiments with sand deposited on substrates of varying roughness revealed a distinct dual-slope profile, reflecting the separation between particle–substrate and particle–particle interaction domains. Internal and external slope angles, along with height distribution and top-shape descriptors, were employed as multi-criteria calibration metrics. These features proved highly sensitive to substrate roughness and feed conditions, allowing for the independent identification of friction and rolling resistance parameters in DEM simulations. Numerical modelling combined with surrogate-based regression confirmed the feasibility of recovering key particle–surface parameters from the experimental heap geometry. The proposed framework offers a resource-efficient and reproducible approach to DEM calibration, reducing the computational load while maintaining sensitivity to local contact mechanics. The methodology is directly applicable in tribological modelling of granular-induced wear, particle-surface interactions, and powder-based processing systems.