Development of a tsunami early warning system for the South China Sea

• Published in: Ocean engineering 2015-05, Vol.100, p.1-18
• Main Authors: Lin, Simon C., Wu, Tso-Ren, Yen, Eric, Chen, Hsin-Yen, Hsu, John, Tsai, Yu-Lin, Lee, Chun-Juei, Liu, Philip, L.-F.
• Format: Article
• Language: English
• Subjects: COMCOT; Manila subduction; Parallel computation; South China Sea; Tsunami early warning
• ISSN: 0029-8018; 1873-5258
• DOI: 10.1016/j.oceaneng.2015.02.003

The Manila subduction zone is identified as one of the most hazardous tsunami source regions, and the countries around the South China Sea are under threat from tsunami hazard. However, the number of early-warning tsunami buoys being deployed in this area is far fewer than that in the Pacific Ocean. This study investigates a feasible approach for establishing a tsunami early warning system in the South China Sea without deploying buoys. The idea is to integrate existing earthquake early warning systems with a fast computing system for estimating tsunami threats. This study presents an efficient and low-cost tsunami fast computing system for early warning. The widely validated tsunami model, COMCOT is chosen as the kernel. The COMCOT source code has been optimized and parallelized in order to meet the requirements of real-time simulation. The optimized model, iCOMCOT performs at least 10 times faster than the original COMCOT. In addition, a flexible and user-friendly grid/cloud-based portal service has been built, which is also made available for mobile devices. As for the automatic generation of the tsunami sources, a new Source-Scaling relationship, which has been validated by recent mega-earthquakes, is implemented. The 2011 Tohoku tsunami is adopted as a case for validation and demonstration. The modelling results of Manila trench, chosen to demonstrate the application of iCOMCOT, show that the western coast of the Philippines is prone to get tsunami attacks. Furthermore, tsunami from the north segment of Manila trench tend to strike southern Taiwan, Hong Kong and Macau area; tsunami from the middle segment, Vietnam. •We developed a low-cost fast calculation system for tsunami early warning.•We optimized and parallelized the original COMCOTsource code.•The optimized model, iCOMCOTperforms at least 10 times faster than the original COMCOT.•The iCOMCOT was available for mobile devices with a friendly user interface.•The 2011 Tohoku Tsunami event was chosen for validation and demonstration.