The Pawsey Supercomputing Centre

Content

Better disaster prediction for South East Asia

Southeast Asia is a tectonically active region that sees many natural disasters such as tsunamis and earthquakes occur as a result of these specific conditions. A team of researchers led by Dr Erdinc Saygin from the Australian National University is using the power of the Pawsey Supercomputing Centre to employ advanced imaging techniques that will significantly increase understanding of the tectonic profile of the region. This work aims to develop processes that may provide early warnings in a broad range of threat areas, potentially saving immeasurable lives and resources.

Challenge

In 2014, the United Nations Statistical Yearbook for Asia and the Pacific ranked the area as the “world’s most disaster prone region”, with deaths from natural disasters in 2004-13 increasing over 300% over the previous decade. Many of these deaths occur as a result of disasters such as earthquakes, tsunamis and volcanic eruptions that are produced as a result of tectonic activity.

Full waveform inversion is a technique developed over the past decade that enables imaging of the Earth and its tectonic activity with an unprecedented resolution by utilising all of the information contained in seismic waveforms.

However, it is extremely computationally intensive and requires use of a high performance computing facility like the Pawsey Supercomputing Centre.

Map showing distribution of seismic imaging stations used to collect data. Image: Dr Erdinc Saygin

Map showing distribution of seismic imaging stations used to collect data. Image: Dr Erdinc Saygin

Dr Saygin is working to estimate the seismic velocity structure of the Southeast Asian crust and uppermost mantle using large amounts of full seismic waveform data collected in Australia, Indonesia and neighbouring countries.

The Australian tectonic plate is subducting underneath the Eurasian plate and this movement is causing large earthquakes and volcanism in the region.

Full waveform inversion will provide much higher resolution than traditional methods but requires considerably more computational resources and data storage, making it impractical for desktop or small cluster environments.

“Simulation of propagation of seismic waves in 3D is a computationally demanding problem,” says Dr Saygin.

“It involves comparison of simulated and observed waveforms, and then updating the underlying model to increase the similarity between these waveforms.”

Solution

Using the Pawsey Supercomputing Centre, Dr Saygin was able to draw upon the world-class computational resources required to conduct his research in an exponentially faster time frame.

“Without the facilities at Pawsey, this project could not be materialised,” says Dr Saygin.

“To give an idea, I would have needed 192 individual computers working at the same time to perform the same work, and each step producing over 3TB of data. This would have been impractical.”

The work conducted at the Pawsey Supercomputing Centre can now be employed in developing a more comprehensive understanding of local tectonic conditions.

Outcome

Shear wave velocity distribution of South East Asia at 70km. Image: Dr Erdinc Saygin

Shear wave velocity distribution of South East Asia at 70km. Image: Dr Erdinc Saygin

The research products of this project will increase understanding about the 3D structure of the South East Asian crust.This information can bring practical benefits to important areas ranging from seismic hazard studies to regional tectonics and improving mineral exploration research and processes.

“We are currently simulating the waveforms and comparing them with the observed ones,” says Dr Saygin.

“The next step is to incorporate a misfit minimisation scheme, which will perturb the model to generate simulated waveforms, which will match the observed ones. This is a multi-iteration approach, which will require many thousands of core hours.”

The resulting models will be crucial in increasing understanding of the tectonic framework of the region and in improving the earthquake locations for nuclear test ban monitoring, rapid earthquake impact assessment, and tsunami warning systems.

Improving these systems could bring a range of scientific, human and economic benefits.

Comments are closed.