Process-Chain Analysis of Viscoplastic Landslide-Generated Impulse Waves from Initial Slide Conditions Through Water-Entry Dynamics to Wave Responses

A key scientific challenge in landslide-generated impulse-wave research is to determine how initial slide conditions are translated into wave responses through the intermediate water-entry process. Existing data-driven studies generally focus on direct prediction or variable-importance ranking and therefore provide limited insight into the pathways linking slide release, water entry, and wave generation. To address this issue, a three-stage process-chain framework was developed using 290 laboratory test cases, integrating initial slide conditions, water-entry dynamics, and wave responses. Cross-stage correlation analysis, standardized path modeling, mediation-effect decomposition, cascade prediction, bootstrap resampling, and diagnostic analyses were jointly applied. The results show that relative effective entry mass M is strongly correlated with relative wave amplitude A and relative wave height H, with Spearman coefficients of 0.96 and 0.91, respectively, while the corresponding coefficients for the entry Froude number $Fr$ are 0.71 and 0.80. Within the present dataset and fitted model, mediation analysis indicates that the pathways through M have the largest modeled indirect contributions: the indirect effects of initial slide mass through M are 0.652 for A and 0.600 for H, whereas the corresponding effects of the Bingham number are $-0.669$ and $-0.615$. The cascade framework further shows that the realized entry state can be systematically reconstructed from the initial conditions and used as an explicit intermediate layer for wave-response prediction. Prediction comparisons further show that the observed entry-state model is best interpreted as an explanatory or diagnostic benchmark because it uses measured entry-state variables, whereas the direct initial-condition and cascade models address prospective prediction from prescribed initial-condition information. The main advantage of the cascade framework therefore lies in its physical interpretability rather than a substantial gain in predictive accuracy. Within the present dataset and fitted model, the associations between the initial slide conditions and wave responses are expressed mainly through mass-related and inertia-related pathways involving the realized entry state.