NH23C-1912
Using the GPU based Model Tsunami-HySEA for the Italian CTSP

Tuesday, 15 December 2015
Poster Hall (Moscone South)
Jose M. Gonzalez Vida Sr1, Manuel J. Castro2, Jorge Macias3, Marc de la Asuncion2, Irene Molinari4, Daniele Melini4, Fabrizio Romano5, Roberto Tonini4, Stefano Lorito5 and Alessio Piatanesi6, (1)University of Malaga, Applied Mathematics, Malaga, Spain, (2)University of Malaga, Mathematical Analysis, Malaga, Spain, (3)University of Malaga, Mathematical Analysis, Malag, Spain, (4)Istituto Nazionale di Geofisica e Vulcanologia, Rome, Italy, (5)National Institute of Geophysics and Volcanology, Rome, Italy, (6)Istituto Nazionale di Geofisica e Vulcanologia, Palermo, Italy
Abstract:
The Istituto Nazionale di Geofisica e Vulcanologia of Italy (INGV) in collaboration with the EDANYA Group (University of Málaga) are proposing a FTRT (Faster Than Real Time) tsunami simulation approach that is being implemented in the NEAMTWS Italian CTSP, namely the Centro Allerta Tsunami (CAT), which is in pre-operational stage starting from 1 October 2014, in the 24/7 seismic monitoring room at INGV.

We here present the different versions and capabilities of the NTHMP benchmarked HySEA model, developed by EDANYA Group. HySEA implements a multi-GPU version that can compute in several minutes the maximum amplitude and arrival time of the main tsunami wave in about 17,000 selected locations along the Mediterranean. At the same time, HySEA is implemented in nested meshes with different resolution and multi-GPU environment, which allows much faster than real time (below few minutes) inundation simulations. The performances of the code allows as well the preparation of a huge number of different pre-calculated scenarios that are being used for PTHA and for warning applications.

Acknowledgements. This research has been partially supported by the Junta de Andalucía research project TESELA (P11-RNM7069), the Spanish Government Research project DAIFLUID (MTM2012-38383-C02-01). Also these results have received funding from the Italian Flagship Project RITMARE, from the INGV-DPC agreement, All.B2, and from EU FP7 project ASTARTE, Assessment, Strategy and Risk Reduction for Tsunamis in Europe grant n° 603839 (Project ASTARTE). The multi-GPU computationswere performed at the Laboratory of Numerical Methods (University of Malaga).

<< Previous Abstract | Next Abstract