Seismic array processing and computational infrastructure for improved monitoring of Alaskan and Aleutian seismicity and volcanoes

Abstract

We constructed a near-real-time system, called Iceworm, to automate seismic data collection, processing, storage, and distribution at the Alaska Earthquake Information Center (AEIC). Phase-picking, phase association, and interprocess communication components come from Earthworm (U.S. Geological Survey). A new generic, internal format for digital data supports unified handling of data from diverse sources. A new infrastructure for applying processing algorithms to near-real-time data streams supports automated information extraction from seismic wavefields. Integration of Datascope (U. of Colorado) provides relational database management of all automated measurements, parametric information for located hypocenters, and waveform data from Iceworm. Data from 1997 yield 329 earthquakes located by both Iceworm and the AEIC. Of these, 203 have location residuals under 22 km, sufficient for hazard response. Regionalized inversions for local magnitude in Alaska yield $\rm M\sb{L}$ calibration curves $\rm (logA\sb0)$ that differ from the Californian Richter magnitude. The new curve is $\rm 0.2\ M\sb{L}$ units more attenuative than the Californian curve at 400 km for earthquakes north of the Denali fault. South of the fault, and for a region north of Cook Inlet, the difference is $\rm 0.4\ M\sb{L}.$ A curve for deep events differs by $\rm 0.6\ M\sb{L}$ at 650 km. We expand geographic coverage of Alaskan regional seismic monitoring to the Aleutians, the Bering Sea, and the entire Arctic by initiating the processing of four short-period, Alaskan seismic arrays. To show the array stations' sensitivity, we detect and locate two microearthquakes that were missed by the AEIC. An empirical study of the location sensitivity of the arrays predicts improvements over the Alaskan regional network that are shown as map-view contour plots. We verify these predictions by detecting an $\rm M\sb{L}$ 3.2 event near Unimak Island with one array. The detection and location of four representative earthquakes illustrates the expansion of geographic coverage from array processing. Measurements at the arrays of systematic azimuth residuals, between $5\sp\circ$ and $50\sp\circ$ from 203 Aleutian events, reveal significant effects of heterogeneous structure on wavefields. Finally, algorithms to automatically detect earthquakes in continuous array data are demonstrated with the detection of an Aleutian earthquake.