There is no purpose-built tsunami warning system in the US whose goal is to issue warnings to the coastline immediately adjacent to a large earthquake. For these regions, destructive tsunami waves will arrive in the first 5-20 minutes following a significant quake. The goal of TlarmS is to develop the algorithms and models necessary to speed up tsunami warning to the point where we can issue expected tsunami intensity maps within 2-3 minutes of the earthquake.

Maps showing testing of TlarmS

Testing the TlarmS algorithm

In the left hand panel, the maps labeled "Non-real time" are what actually happened during those 4 earthquakes - a post-event reconstruction of each tsunami. The maps show what tsunami heights were exceeded at each prefecture or county and the black curves next to them show on a mile-by-mile basis what the tsunami was at the coastline. The right column shows the tsunami warning maps TlarmS could have issued 2-3 minutes after the earthquake start, had TlarmS been running in real time.

To accomplish this, the Berkeley Seismo Lab's Diego Melgar and colleagues are using the results from rapid earthquake source models being implemented for ShakeAlert and other warning systems around the world. TlarmS uses any source information, complex or simple, from point-source magnitude estimates to centroid moment tensor and full finite fault solutions, to drive a physically realistic rapid tsunami propagation model that can forecast expected impacts at the near-source coast.

Currently, the TlarmS group is in the process of systematically testing the performance and reliability of the fast tsunami intensity forecasts by running TlarmS through a large suite of M8-M9 Cascadia earthquake scenarios developed in house. The Berkeley Seismo Lab is in close collaboration with NOAA tsunami warning centers and working so that in the near future the TlarmS approach can become operational and used routinely by operational agencies, not just in Cascadia, but around the world.