Rockfall Hazard Assessment and Flexible Barrier Design for Steep Forested Rock Slopes Using Rockyfor3D and RAMMS::ROCKFALL: A Case Study in Qingtian, Zhejiang Province, China

Rockfall motion on high and steep forest-covered slopes is controlled by multiple factors, including source-area conditions, block characteristics, slope topography, and surface conditions. Differences among numerical models in representing block geometry, contact and impact processes, and forest-induced retardation may lead to inconsistent predictions of hazard extent and protection requirements. This study investigates a high and steep forest-covered rock slope in Qingtian, Zhejiang Province, China. A high-resolution digital elevation model (DEM) was constructed from LiDAR point-cloud data, and potential source areas, surface types, and forest parameters were determined through field investigations. Rectangular, ellipsoidal, and discoidal blocks of approximately equal volume were simulated under unprotected and flexible-barrier conditions using Rockyfor3D, a representative point-mass model, and RAMMS::ROCKFALL, a rigid-body dynamics model. Both models predicted the lowest downslope mobility for rectangular blocks and the highest for discoidal blocks. However, their dynamic responses to block shape differed markedly. Rockyfor3D predicted higher kinetic energies and bounce heights for ellipsoidal and discoidal blocks, whereas RAMMS::ROCKFALL predicted higher corresponding values for rectangular blocks. RAMMS::ROCKFALL also predicted a wider lateral spread for ellipsoidal blocks and more road crossings for discoidal blocks. The energy rating, height, and length of the barriers determined from the two models differed considerably, although both protection schemes substantially reduced the number of blocks crossing the road monitoring line. The trajectory envelopes predicted by both models did not cover the three historical rockfall deposition zones identified in the field, and the minimum distances to the nearest deposition zone were 61.1 m and 40 m, respectively. The incomplete identification of potential source areas, changes in historical terrain and vegetation conditions, the irregular geometry of natural rock blocks, and uncertainties in model parameters may all affect the reproduction of the actual spatial distribution of rockfalls. Overall, Rockyfor3D is more appropriate for rapid regional-scale probabilistic screening, whereas RAMMS::ROCKFALL is preferable for detailed assessments of critical slope sections requiring explicit consideration of rock-block geometry and rotational dynamics. In engineering applications, field evidence of rockfall should be incorporated, and multi-model cross-validation together with a conservative envelope approach should be used to delineate hazard zones and determine protection parameters.