Predictability is about understanding the key triggers of landslides and how they depend on lithology, vegetation, topography, climate, hydrologic conditions, tectonics, and landsliding and land use history. It will be an innovation engine for the center, refining existing models and conceptualizing new theories for cross-spatial scale applications, from a single-slope to regional-scale analyses and modeling.
The goal of this research thrust is to augment regional-scale predictability by combining scientific underpinnings and large-scale datasets. The proposed research will:
- Advance single-slope analyses and process-based models to uncover and systematically address the sources of uncertainty with the greatest impact on predictability,
- Leverage the latest remote sensing capabilities and evolving national and worldwide databases using ML approaches,
- Maximize the synergy between process-based models and data-driven predictive models using new advances in physics-informed ML and differentiable modeling.