A REAL-TIME EARTHQUAKE MOMENT TENSOR SCANNING CODE FOR THE ANTELOPE SYSTEM (BRTT, INC)

Abstract:

While all seismic observatories routinely determine hypocentral location and local magnitude within a few minutes of an earthquake's occurrence, the ability to estimate seismic moment and sense of slip in a similar time frame is less widespread. This is unfortunate, because moment and mechanism are critical parameters for rapid hazard assessment; for larger events, moment magnitude is more reliable due to the tendency of local magnitude to saturate, and certain mechanisms such as off-shore thrust events might indicate earthquakes with tsunamigenic potential. In order to increase access to this capability, we have developed a continuous moment tensor scanning code for Antelope, the ubiquitous open-architecture seismic acquisition and processing software in use around the world. The scanning code, which uses an algorithm that has previously been employed for real-time monitoring at the University of California, Berkeley, is able to produce full moment tensor solutions for moderate events from regional seismic data. The algorithm monitors a grid of potential sources by continuously cross-correlating pre-computed synthetic seismograms with long-period recordings from a sparse network of broad-band stations. The code package consists of 3 modules. One module is used to create a monitoring grid by constructing source-receiver geometry, calling a frequency-wavenumber code to produce synthetics, and computing the generalized linear inverse of the array of synthetics. There is a real-time scanning module that correlates streaming data with pre-inverted synthetics, monitors the variance reduction, and writes the moment tensor solution to a database if an earthquake detection occurs. Finally, there is an 'off-line' module that is very similar to the real-time scanner, with the exception that it utilizes pre-recorded data stored in Antelope databases and is useful for testing purposes or for quickly producing moment tensor catalogs for long time series. The code is open source and is available from a Github repository. This presentation will outline the structure of the software package, will discuss the capabilities and limitations of the code, and will show example solutions from running the code at the Alaska Earthquake Center, including results from the M_w = 5.8 Skwentna, Alaska earthquake of June 24, 2015.