Using Physically Based Methods (PBMs) and ALICE® Methodology
ALICE® (Assessment of Landslides Induced by Climatic Events) is designed around a physically coherent chain of processes:
Rainfall → Infiltration → Groundwater Rise → Pore Pressure Increase → Effective Stress Reduction → Shear Strength Mobilization → Failure
The Morgenstern & Price (1965) formulation ensures full force and moment equilibrium, essential when scaling stability analysis from single slopes to thousands of basin-wide profiles.
Visual representation of ALICE® methodology components and results
Discretization of terrain using Local Drain Direction (LDD) networks. The basin is divided into physically meaningful slope profiles that align with natural drainage paths and potential failure directions.
Legend detailing lithological units, structural features, and soil classification used in the geological-geotechnical model construction.
Groundwater level surfaces for 5-, 10-, 25-, and 50-year rainfall return periods, explicitly linking precipitation events with probabilistic groundwater elevations.
Map conventions and symbols for groundwater level surface analysis, including elevation contours, hydraulic gradients, and boundary conditions.
Combined hazard surface including all modeled failure mechanisms expressed as annual probability of failure. Complex landslides dominate basin-scale hazard despite limited inventory representation.
Probability of failure classes with color coding and statistical thresholds from very low to very high risk zones.
Detailed probability ranges and failure mechanism classifications: shallow translational, deep-seated rotational, and complex failure modes.
Critical findings from physically based hazard zoning implementation.
Groundwater level rise is identified as the primary triggering variable, driving non-linear reductions in effective stress and abrupt increases in failure probability. This represents a fundamental shift from empirical correlations to physically meaningful relationships.
The generalization of soil thickness strongly affects failure geometry. Shallow soils dominate steep slopes while deeper colluvial deposits accumulate in concave areas, requiring spatially distributed thickness models for accurate hazard assessment.
Complex landslides dominate basin-scale hazard despite limited inventory representation. This highlights the importance of physics-based methods in identifying mechanisms not captured in historical records due to their infrequent occurrence or difficult detection.
Hazard escalation with rainfall is threshold-driven and non-linear. The relationship between precipitation intensity/duration and slope instability shows critical thresholds beyond which failure probability increases dramatically.
Physically based zoning reduces epistemic uncertainty by replacing correlation with mechanics. The dominant uncertainty becomes related to subsurface conceptualization rather than numerical implementation, enabling more defensible hazard assessments.
Model outputs are evaluated against independent landslide inventories using ROC curves, AUC, Kappa Index, and Correct Classification Rate. Results demonstrate that physically based methods:
Better identify physically unstable domains compared to empirical approaches, providing more reliable hazard assessments for critical infrastructure planning.
Less sensitive to inventory incompleteness, making them more robust for regions with limited historical landslide data or changing environmental conditions.
Adapting ALICE® methodology for Colombian geological and regulatory context