Intercomparisons Between Geodetic and Seismic Algorithms for ShakeAlert

Date: 4/25/2019

Time: 02:45 PM

Room: Puget Sound

The current ShakeAlert earthquake early warning system under development by the USGS and university partners consists of several seismic source characterization algorithms, namely EPIC and Finder. However, seismic algorithms tend to saturate for the largest earthquakes, leading to an under-prediction of strong ground motions. For almost a decade now, many groups have demonstrated the utility of adding high-rate, real-time GNSS data for the robust characterization of large earthquakes (M > 7). Three GNSS-based algorithms, BEFORES, G-FAST and GlarmS, are currently under development for ShakeAlert. These algorithms have been tested on an array of earthquakes, however, intercomparisons in ‘apples-to-apples’ tests have not been performed. This task is the primary objective of the ShakeAlert Geodetic Algorithms Testing and Implementation Subcommittee (GATIS). The GATIS group has been devising a framework for effective testing and metrics definitions so that streamlined comparisons can be made between all the geodetic algorithms and the current seismic algorithms. Metrics can be segregated into three categories, ground motions, source models, and timeliness, and each category is subdivided into many different metrics. For example, timeliness is separated into warning times, MMI threshold exceedance times and computation time. For our initial tests, we have decided to focus on the El Mayor-Cucapah, Napa, and Tohoku-oki earthquakes. From these initial tests, our testing framework will be fully streamlined to enable tests across an array of real and synthetic earthquakes.

Presenting Author: Brendan W. Crowell

Authors