Dominant Factors and Prediction Model of Toppling Collapse for Steep Cliffs in Strong Earthquake Zones

Abstract Earthquake-induced collapses of steep cliffs with hazardous rock masses represent a prevalent geological hazard in mountainous regions. Current stability assessment methods for grouped rock failures under seismic loading exhibit significant limitatiion. This study systematically investigates failure modes, critical collapse thresholds, and risk factors through an integrated approach combining scaled shaking table tests and 3DEC discrete element modeling, focusing on hazardous rock formations along the China-Pakistan Highway.ons, particularly in addressing complex failure mechanisms like toppling collapse. Key findings reveal: (1) The displacement angle (≈ 15°) serves as a robust collapse indicator, outperforming conventional metrics; (2) Among four primary influencing factors (rock fragmentation, joint inclination, height, and strength), weathering degree emerges as the dominant control; (3) Our developed BP neural network model identifies toppling and sliding as predominant failure modes, with joint inclination being the most critical controlling parameter; (4) Collapse initiation demonstrates nonlinear dependence on rock geometry (size, aspect ratio, and layering configuration), where increased joint inclination accelerates failure while reduced rock size promotes rapid collapse progression. The derived critical vibration velocity establishes a reliable criterion for toppling collapse prediction. Factor importance analysis ranks controlling parameters as: rock size > joint inclination > shape > layering configuration. These findings provide a theoretical framework for predicting and preventing seismic-induced rock collapses in earthquake-prone regions.