Towards digital-twin-enabled tsunami hazard assessment: Landslide-Tsurrogate v1.0

Landslide-Tsurrogate v1.0 is an open-source Python and MATLAB framework designed to efficiently estimate tsunami hazards generated by submarine landslides. Rather than relying on thousands of computationally expensive deterministic simulations in real time, the tool constructs surrogate models that can rapidly reproduce tsunami responses at a fraction of the computational cost once an event occurs. The approach is based on generalized polynomial chaos expansion, which enables an efficient exploration of uncertainties in landslide parameters and their impact on tsunami generation.

The framework allows users to perform sensitivity analyses, identify the most influential parameters, and quantify the variability of tsunami outcomes in a probabilistic manner. To facilitate accessibility and transparency, Landslide-Tsurrogate v1.0 is distributed with a Jupyter Notebook User Manual and interactive MATLAB and Jupyter Notebook interfaces, enabling straightforward model configuration, surrogate construction, and result visualization.

The performance of the model is demonstrated through a real-world application to five submarine landslide-prone zones offshore Mayotte (France). In this case study, surrogate convergence is achieved with only 135 deterministic simulations per zone, and probabilistic tsunami hazard estimates are produced in less than 2 seconds on a standard laptop. These results highlight the strong computational efficiency of the approach.

Beyond this application, the framework is readily transferable to other coastal regions exposed to submarine landslide hazards. By combining physical modeling, statistical methods, and user-oriented design, Landslide-Tsurrogate v1.0 provides a fast, transparent, and practical tool for probabilistic tsunami hazard assessment.