First simulations with SeisSol on Marconi100 and Piz Daint
09 July 2020

Within the ChEESE Center of Excellence for Exascale in Solid Earth (ChEESE),  the earthquake simulation software SeisSol, one of the 10 ChEESE flagship codes, is being ported to GPU-based supercomputers. First simulations have now been  executed on Marconi100 and Piz Daint, two of the largest European supercomputers.

SeisSol simulates highly complicated earthquake processes using detailed models of fault zone physics and of the geological subsurface and topography.  Landmark simulations have been executed for the 2004 Sumatra earthquake and for the 2018 Palu earthquake - both events triggered devastating tsunamis.  Within a ChEESE pilot demonstrator on "physics-based tsunami-earthquake interaction", ChEESE researchers are developing novel simulation approaches using SeisSol  that will help to explain in more detail how earthquakes generate tsunamis, and thus to better assess cascading hazards.


Palu SeisSol image


SeisSol modelled earthquake and linked tsunami across Palu Bay from Ulrich et al., 2019: Left: Seismic waves being generated while the earthquake propagates southward in a ‘superfast’ manner. Warm colours denote higher movements across the geological faults and higher ground shaking (snapshot after 15 seconds of earthquake simulation time). Right: The movements of the earthquake beneath the bathtub shaped Palu Bay generate a ‘surprise’ tsunami (snapshot of the water waves after 20s of simulation time of the tsunami scenario).


"Fully coupled multi-physics modelling has the potential to identify the key factors and physical links controlling earthquake and tsunami dynamics. By integrating a wide range of observations, from local to megathrust scales,  such models bridge scientific disciplines and space-time scales," states Alice-Agnes Gabriel of LMU Munich, who leads the ChEESE activities on  "physics-based tsunami-earthquake interaction".

"We expect that such simulations will require several hundred billion degrees of freedom and will need to exploit the entire computing power of the upcoming  European exascale supercomputers,"  says ChEESE participant Michael Bader from Technical University of Munich (TUM)

While SeisSol has set several performance records on petascale supercomputers,  it currently relies on regular CPU-based computer architectures. However,  most projected exascale machines will draw their computing power from GPUs  (general-purpose graphics processors). 

To achieve best-possible performance on CPU-based supercomputers, SeisSol relies  on a code generator, YATeTo, developed at TUM. More than 80% of SeisSol's  core routines are generated in this way. 

"GPU architectures are essentially different from CPUs. The most important step was therefore to find a data decomposition for the ADER-DG method that can result in good GPU performance. Moreover, integrating a different kind of computation into SeisSol via YATeTo is a big challenge," said Ravil Dorozhinskii, doctoral candidate at TUM, who is responsible for the SeisSol GPU porting.

Together with ChEESE partners at CINECA, the "alpha"  version of the SeisSol GPU port is now tested on the Piz Daint supercomputer (at the Swiss National supercomputing centre) and on Marconi100, the recently installed at CINECA, the Italian National Supercomputing centre. The GPU version of SeisSol achieved almost 8.5 single precision TFLOPS on a single node of CINECA's Marconi100 cluster.

The earthquake simulation software SeisSol is available as open software software on The GPU version of SeisSol is expected to be released in autumn 2020.

ChEESE contributes to Spain´s unique insurance scheme against natural catastrophes
06 July 2020

Jorge Macías is a researcher from the University of Malaga and a member of the Differential Equations, Numerical Analysis and Applications (EDANYA) group. EDANYA is an internationally renowned group in tsunami modelling and the developers of the Tsunami-HySEA code, a software used in predicting the behaviour and effects of tsunamis. In this interview, Macias talks about why it is important to study these natural phenomena and how ChEESE´s work can have a great impact on the insurance landscape in Spain. 

What are tsunamis and why is it important to study them?

Tsunamis are natural phenomena that can be triggered in bodies of water such as lakes, dams, seas or oceans. They are somehow similar to ordinary sea waves, such as those we surf on, at least from a theoretical perspective. However, tsunamis are very different from a practical point of view such as the time it takes from one wave to be followed by the next one, the distance between waves or the height of the waves. In turn it all translates into the amount of energy each type of wave has. Any ordinary Atlantic Ocean wave is followed by the next one after a few seconds usually a few hundred metres apart. Tsunami waves, meanwhile, take minutes or even hours from one to the other and tend to be hundreds of kilometres apart. Ordinary waves can also be as high as 20 metres or more, particularly in the midst of a storm in the middle of the ocean, whereas tsunamis are much smaller and imperceptible in deep waters. While ordinary waves usually do not wet more than a few meters from the coast line, tsunamis can travel very far inland, hence the peril they pose to lives and property. Basically, a tsunami is a rare high-energy wave with potentially devastating effects. While ordinary waves are triggered by ordinary phenomena, such as winds or tides, tsunamis are mostly linked to extraordinary issues, such as big earthquakes, big volcanic eruptions or landslides, or even rarer, due to meteorite impacts on water bodies. 

Studying tsunamis is important because we can prepare for what we know might happen, but we cannot face things we do not know about. Extreme tsunamis can cause tens of thousands in casualties (as in the 2004 Indian Ocean tsunami) and huge economic loses, but with a proper plan, we can avoid most of it by mitigating their effects. We need to be prepared, and we need to know what to prepare for.
How often do tsunamis occur in Spain and what are their socioeconomic impacts?

Analysing past data has shown that the Mediterranean and the Atlantic coasts of Spain have suffered the impact of tsunamis in historical times. The biggest of these events, being the largest known natural disaster ever suffered in Europe, occurred in 1755, almost wiping out Lisbon and causing severe damage also in Spain. However, the earthquake that triggered that tsunami was quite a rare and big event. Significant tsunamis can happen any moment, anywhere near shore. Whenever there was an earthquake (or tsunami) in the past, there would be more in the future. The chances a tsunami could hit Spain are relatively low, something like 14% every year within the Mediterranean area (this means one very minor tsunami every 7-8 years) and about a 5% for the Atlantic area (one every 20 years). However, a significant tsunami (any tsunami capable of major damage) would much be less frequent and would likely happen every hundreds of years apart. As a rule of thumb, the bigger the tsunami, the lower the chances. We are studying tsunamis today in order to provide more reliable information about tsunami probabilities linked to socioeconomic consequences, so we can focus our efforts towards preventing them.

You have previously given a talk at the Consorcio de Compensación de Seguros (Insurance Compensation Consortium) annual meeting regarding the effects of tsunamis. Why are they particularly interested in this?

Spain has a unique insurance scheme regarding natural catastrophes (nat-cat) as compared to other countries, thanks to the Consorcio de Compensación de Seguros (CCS), a private-public partnership that has provided many success cases. The idea is the following: whenever you buy any kind of insurance in Spain, you are also contributing to strengthening the nat-cat insurance with such a tiny surplus you would not even notice it. More importantly, you are also buying yourself nat-cat insurance, which would ultimately be provided by the CCS and not by your actual insurance company. Thus, the penetration rate of such scheme is very high, meaning the vast majority of people in Spain are in fact insured against natural catastrophes, including tsunamis. But, how much damage could a tsunami cost to this system? That is the million dollar question we are trying to solve together with an outstanding team brought together by CCS comprised of scientists and engineers, including geologists, mathematicians, physics, and many others. 

What is ChEESE doing to mitigate the effects of tsunamis?

The main objective of ChEESE is to establish a new Center of Excellence (CoE) in the domain of Solid Earth (SE) targeting the preparation of 10 Community flagship European codes for the upcoming pre-Exascale (2020) and Exascale (2022) supercomputers. This means making available the best computers at hand, the largest computational resources, together with the best European codes in Solid Earth and, in our particular case, for tsunami modelling. Highly efficient and robust codes on top of HPC resources gives as result much Faster-Than-Real-Time (FTRT) tsunami simulations. This means that we are able to compute the evolution (simulate) of the tsunami wave and its impact on coast faster (much faster) than the time it takes for the actual tsunami wave to arrive to the real coast. This was impossible some years ago and it is one of the aims of the ChEESE project. Providing tsunami warning centres with computational tools to predict the impact of a tsunami some time in advanced it reaches the coast. This is one among several applications, but an extremely important one as it aims at saving lives.

How do you think the results of ChEESE will positively affect the insurance landscape in Spain?

ChEESE will provide the necessary tools to provide flooding scenarios after tsunamis. First of all, to perform such an extensive study, covering all the Spanish coast (more than 10,000 km) requires the use of HPC resources. Simulating a large amount of potential and historical scenarios is a major computational task. Targeting is such a vast coast is a major challenge not performed before. Linking such flood scenarios with cadastral information and other social and economic databases is the next step since this will provide a deeper understanding not only of the natural phenomena, but how it will likely affect insured assets. This can help in many ways, beyond the insurance landscape. It can help citizens to understand the benefits of being insured, and the far greater benefits of avoiding exposure. Public awareness is key towards better planning, including developments and self-protection or self-mitigation tools, such as insurance. Moreover, the higher the public awareness, the more likely losses will be reduced, provided nobody wants to be hit by a tsunami.