Comparison of 3D and 2D rockfall models. Considering terrain model quality effect on respective model performances

Abstract

Rockfalls are blocks of rock that bounce and roll down slopes. They are commonly detached from rock-slopes with inclinations over 40-45°. They embody high energy and mobility and are a major cause of landslide related deaths. Rockfalls can be simulated using empirical relationships and modelling, of which 2D and 3D variants exist for the latter. 3D models allow for rock fragment- and surface shape to be used in impact calculations but require more detailed input and longer computation time.

This thesis compares a 3D rockfall model with a 2D equivalent in terms of output values. 2D software Rocfall 6.0 and 3D software RAMMS:Rockfall is used to back calculate 4 historical rockfalls as comparative basis. The comparison involves the study of individual model performances using digital elevation models from aerial LIDAR and terrestrial photogrammetry, resolutions between 10 cm and 10 m, and on 4 different terrains. A final 3D simulation is used to design a mitigative structure to illustrate rockfall modellings capability in rockfall engineering.

The research finds that 3D modelling generally calculates higher bounce heights and total kinetic energies than 2D modelling, particularly on gentle slopes like taluses. Where 2D modelling show rolling/sliding motion, 3D models often calculate bounces. Higher resolutions generally produce higher output kinetic energies and bounce heights for both 2D and 3D software even when run out length is similar. Simulations using digital elevation models with resolution much higher than 1 m output unrealistic settling locations even in steep slope sections. Steep areas in digital elevation models produced with aircraft based LIDAR are gentler than in those produced with terrestrial photogrammetry. The latter cause the calculation of notably higher bounce heights.