Machine learning surrogates for approximating tsunami wave height time series nearshore and maximum inundation depth onshore for the Japan Tohoku region. Related article available as preprint - Advancing nearshore and onshore tsunami hazard approximation with machine learning surrogates available at https://doi.org/10.5194/nhess-2024-72
_input/_output/_tsunami/
_inputs/(input source parameters for DOE and historic events)dtopo_his/(dtopo files for historic events and plotting)dtopo_sift/(for type B)dtopo_slab/(for type A)
MAP/(onshore surrogate for maximum inundation depth prediction)TS/(nearshore surrogate for time series prediction)
_plots/_results/_stats/
Following are the YAML files with information on the Python packages and requirements to run:
- GeoClaw 2DNLSE simulation:
/geoclaw/geoclaw.yml - Machine learning:
/surrogates/pytorch.yml - PyGMT plotting:
/paper/pygmt.yml
Each directory contains a more detailed README.md.
Some large input files for the geoclaw simulation and the post-processed inputs for machine learning need to be downloaded from https://doi.org/10.5281/zenodo.10817116
-
https://github.com/rjleveque/MLSJdF2021 - A project using VAE for tsunami forecasting problem, developed in Python/Pytorch. Liu, C.M., Rim, D., Baraldi, R. et al. Comparison of Machine Learning Approaches for Tsunami Forecasting from Sparse Observations. Pure Appl. Geophys. 178, 5129–5153 (2021).DOI: 10.1007/s00024-021-02841-9
-
Tsunami Inundation Emulator - A project for tsunami inundation depth prediction using machine learning, developed in Julia/Flex. Erlend Briseid Storrøsten, Naveen Ragu Ramalingam, Stefano Lorito, Manuela Volpe, Carlos Sánchez-Linares, Finn Løvholt, Steven J Gibbons, Machine Learning Emulation of High Resolution Inundation Maps, Geophysical Journal International, 2024;ggae151. DOI: https://doi.org/10.1093/gji/ggae151