Our “Probabilistic Tsunami Forecast” workflow adapts existing workflows to be more readily and dynamically deployed on federated HPC resources. The goal is to provide Tsunami Warning Centers with a robust, likelihood-based, prediction as to the tsunami impact following a large earthquake.
Immediately after a large offshore earthquake, there is great uncertainty as to exactly where, and how large, the earthquake was and which impact an associated tsunami will have. Decision makers in Civil Protection need the best possible basis for issuing warnings and/or evacuation orders, and for best directing post-disaster relief. In Eflows4HPC we exploit HPC to calculate very rapidly the outcomes of many different post-earthquake scenarios.
Though HPC is necessary for large scale urgent tsunami computations, the threshold for implementing advanced workflows on HPC facilities is currently high and it needs to be lowered to make HPC accessible to more users and wider applications.
Our “Probabilistic Tsunami Forecast” workflow adapts existing workflows to be more readily and dynamically deployed on federated HPC resources. The goal is to provide Tsunami Warning Centers with a robust, likelihood-based, prediction as to the tsunami impact following a large earthquake. This is a case of Urgent High Performance Computing.
Goals
Our researchers are currently working on speeding up and improving ease of deployment. We want to provide an efficient interpretation of streams of data. To do so our current work focuses on:
- Developing the ability to respond to real-time incoming streams of data – seismic data, tsunami data, and other sources.
- Exploring ways to be able to update dynamically our ensemble of scenarios and/or the associated likelihoods according to the most up-to-date data.
- Applying Machine Learning to predict the outcomes of the numerical tsunami simulations – mainly for improving the efficiency.
The “Probabilistic Tsunami Forecast” workflow fits within our wider aim of enabling the accessibility and reusability of applications to reduce the time to solution. The goal of Eflows4HPC is to create a European workflow platform which enables the design of complex applications that integrate HPC processes, high performance data analytics, and artificial intelligence.