Heavy rainfall infiltration is a key disaster-inducing factor that triggers the softening of surrounding rock and deformation of support structures in tuff gully tunnels. Based on the gully section of the left line of the Dabao Tunnel of the Leigongshan–Rongjiang Expressway in Guizhou Province, this study systematically reveals the synergistic disaster-inducing mechanism of “topography-seepage-softening” in tuff gully tunnels under heavy rainfall infiltration through laboratory tests and FLAC3D 3D numerical simulations. The main innovative conclusions are as follows: (1) The “phased” attenuation law of tuff mechanical parameters was quantified, and the critical water content for significant strength deterioration was determined to be 2.5%, with a saturated softening coefficient of 0.59. These results provide key data for early warning and evaluation of similar projects. (2) A “convergence-disorder” distribution pattern of pore water pressure controlled by gully topography was revealed. It was found that the rock mass directly below the aqueduct exhibits a disordered zone with downward-extending pore water pressure due to fluid convergence, with the maximum pore water pressure reaching 0.55 MPa. This clarifies the essence that tunnel stability is controlled by the coupling of topography and seepage field. (3) The key sensitive areas for tunnel stability—namely the gully bottom, arch haunches, and the area below the aqueduct—were accurately identified. The significant increase in displacement of these areas after rock stratum softening was quantified (e.g., the displacement at the crown of the secondary lining increased from 3 mm to 4 mm, and the influence range expanded to the arch haunches). This study clarifies the deformation characteristics and instability mechanism of tuff gully tunnels under heavy rainfall from two aspects: the “internal mechanism of rock mass softening” and the “external condition of topographic seepage control.” It can provide a theoretical basis and key technical pathway for disaster prevention and control as well as stability design of similar tunnels.

The cumulative displacement of landslides in alpine regions is jointly affected by rainfall, temperature variation, freeze–thaw cycles, and other factors, and usually exhibits nonlinear, non-stationary, and multi-scale fluctuation characteristics. To improve the accuracy of landslide displacement prediction under complex environmental conditions, this study takes the Taker Tubek Village landslide in Gongliu County, Xinjiang, China, as the study object. Cumulative displacement data from GNSS02 and GNSS03, together with daily rainfall and daily mean temperature, were used to construct a SOO-VMD-CNN-TimeXer hybrid prediction model. First, SOO was employed to adaptively optimize the VMD parameters, and the cumulative displacement series were decomposed into multiple IMF components. Then, CNN was used to extract local fluctuation features, while TimeXer was applied to model long-term temporal dependencies and the effects of exogenous variables. Finally, the predicted results of all components were reconstructed to obtain the cumulative displacement prediction. The results show that the proposed model achieved high prediction accuracy at both GNSS02 and GNSS03. The MSE, MAE, MAPE, and R² values were 0.0158, 0.0960, 0.0112, and 0.9464 for GNSS02, and 0.0483, 0.1590, 0.0203, and 0.9946 for GNSS03, respectively, outperforming LSTM, Informer, iTransformer, Crossformer, and other models. The results indicate that the SOO-VMD-CNN-TimeXer model can effectively characterize the cumulative displacement evolution of landslides in alpine regions and provide technical support for landslide deformation trend forecasting and disaster early warning.