Figure S1 (a) A benchmark model considering self-weight stress (gravity) for a cubic solid (400 × 400 × 1,000 m) with fixed boundary and bottom. (b) A cross-section at 500 m depth for reference.

Figure S2 (a) FEM model, the black box represents the section at a certain depth, which is the DDA model. (b) DDA model corresponding to cross section in Figure S1b, used for the benchmark. The model is constrained by the boundary blocks with fixed points, which does not allow deformation in the x-y plane.

Figure S3 (a) Comparison of DDA method in simulating in-situ stress with FEM and theoretical solution; (b) The relative error between two methods and theoretical solutions.

Figure S4 The flowchart of the DDA model for the rock avalanche. (a) Conversion between dynamic iteration and static iteration to realize simulation of fault creep and rock avalanche initiation and movement; (b) DDA method processing flow, including parameter input, contact judgement, iterative calculation and convergence judgement. The black dotted line indicates open-close iteration; (c) in the process of rock avalanche movement, the strength weakening based on normal stress is considered, so that the process of rock avalanche movement be described accurately.

Figure S5 Stress evolution of monitored blocks during self-stabilization stage. (a) Horizontal stress σ_x (negative for compaction) evolution of blocks at different depths of the footwall; (b) horizontal stress σ_x evolution of blocks at the same horizontal levels of footwall. (c) Horizontal stress σ_x evolution of blocks at different parts of the slope.

Figure S6 (a) Edge-to-edge contact. Edge-to-edge contact can be equivalent to two vertex-to-edge contacts. Black dotted box refers to where we embedded the spring and dashpot; (b) Creep constitutive law is incorporated into the DDA method. Note that t₁ and t₂ represent the transition points between the primary and secondary stages and between the secondary and tertiary stages, respectively. t_R represents the final moment of rock creep (i.e., when t = t_R, D = 1, see Text S3), and t_s is the moment at which the rock enters the tertiary phase (i.e., when t = t_s, D = 0).

Figure S7 Evolution of strain of the monitored blocks within the rock slope as a function of time. The index of rock block is indicated in Figure 8b.

Figure S8 (a-f) The snapshots of strain and sliding process of the simulated rock avalanche caused by mass rock creep at the time of t = 0, 0.45, 0.9, 2.45, 4 and 4.2 ky, respectively. The time t = 0 marks the onset of the mass rock creep deformation process after the self-stabilization stress state, and t > 0 corresponds to the later time.

Figure S9 Evolution of the Mohr circle for the slope toe before the complete detachment of slope toe from the hanging wall. (a) Evolution of principal stress σ₁ and σ₃ at the bottom of slope toe in the case of c = 1.5 MPa and friction angle φ = 22.5°; (b) evolution of principal stress σ₁ and σ₃ at the bottom of slope toe in the case of c = 2.1 MPa, friction angle φ = 19.5°.

Table S1 Rock avalanches that occurred in the vicinity of active fault zones worldwide.

Table S2 Rock mass mechanical and joint physical parameters used in the benchmark model (FEM and DDA model).

Table S3 The control parameters for DDA in the benchmark model.

Table S4 The key parameters for MRC used in the DDA.