Connecting earthquake source products to local tsunami warning

Tuesday, 15 December 2015
Poster Hall (Moscone South)
Diego Melgar and Richard M Allen, University of California Berkeley, Berkeley, CA, United States
Issuing warning of a tsunami in advance of its arrival to the coastlines immediately adjacent to large earthquakes remains a challenging problem. The heterogeneous development state of regional geophysical monitoring infrastructure across subduction zones worldwide means that a flexible approach to warning, capable of ingesting multiple data types and earthquake source products, is the most appealing. We will present results from the study of 3 recent large events that have been observed with diverse geophysical measurements; the 2011 Mw9.0 Tohoku-oki, the 2010 Mw8.8 Maule and 2014 Mw8.2 Iquique events. First, we will show that earthquake slip models derived from combination of land (GPS and strong motion) as well as off-shore (tide gauges, ocean-bottom pressure, and GPS buoy) can be coupled to tsunami propagation models to produce simulations that closely match the measured run-up at the local coastlines. Using these models as a baseline for validation we will demonstrate a methodology that takes advantage of simpler, but more readily available earthquake source products such as rapid point-source magnitude estimates from coastal GPS observations and regional moment tensors. We will show that while trading-off precision for speed, these simpler earthquake source models produce inundation forecasts reliable enough to be used for warning within minutes of earthquake onset. Most subduction zones around the world already have some geophysical infrastructure and are producing some form of real-time earthquake source product, our results strongly argue that by coupling these data products to tsunami propagation models local tsunami warning is possible at most subduction zones with already available infrastructure.