An Automatic Event Detector at the Tonto Forest Seismic Observatory

Abstract

The on‐line operation of an automatic event detector has been evaluated at the Tonto Forest Observatory short‐period seismic array. For 31 seismometers and one fixed threshold, the 90 percent incremental detection threshold on the Kuril Island beam, centered at Δ=70 degrees, is [Formula: see text] with an experimentally determined false alarm rate of 0.17 per day. This compares favorably with the capabilities of a human operator. Storms in the Kurils significantly affect the distribution of amplitudes of the F-statistic detection trace, and we estimate that most of the false alarms observed at the operating threshold can be traced to the statistical bias introduced by this storm‐generated energy. If the threshold were adjusted to maintain a constant false alarm rate, the maximum effect on the threshold magnitude would be [Formula: see text].

Single Channel Seismic Event Detection

“…This detector may be said to be as "standard" as the IBM detector, and some normalization of this type seems to be a useful idea. Blandford (1974) found that an F-detector operating on a beam for one month also performed in agreement with theory and had no trouble with spikes, dropouts, local events or noise bursts. The third detector is that of simply summing the raw power in the signal band and then converting to a unit normal.…”

“…Vanderkulk et al evaluated by means of theoretical analyses the degradation which would result from various approximations to the optimum procedure and concluded that the difference would be small. Further research on detection proceeded to questions of array detectors which might be resistant to false alarms due to spikes and locals, Blandford (1970Blandford ( , 1972Blandford ( , 1974, to general questions concerning array and network detection, Wirth et al (1971), Shumway (1971), Ringdal et al (1972), Blandford and Wirth (1973) and ; to methods of holding the false alarm rate constant, Lacoss (1972), and to various ad-hoc methods of recognizing spikes and local events, von Seggern (1977).…”

Multivariate acoustic detection of small explosions using Fisher's combined probability test

“…Each detector measures a different signal characteristic: signal coherence over an array over a short time span, a rapid increase in power, and a broadband short-duration signal. The first discriminant is an array F-detector that measures the ratio of beam power on an array (where the beam is formed by time-aligning and stacking the waveforms from each array element over all directions of arrival such that the beam power is maximized) to the residual power (Blandford, 1974). By using a short time-window of 1 s, this discriminant identifies short-duration/high-frequency signals that may be consistent with explosions.…”

<title>Infrasonic observations of bolides on October 4, 1996</title>

“…In addition to doing single station analyses, we also event associated the signals between the various arrays in order to look for events with common wave sources. We used four signal detection methods, including peak crosscorrelation of the signal, the f statistic (Blandford, 1974), lag closure and the composite signal power. We found 28 signals using the standard cross-correlation approach when a value of 0.75 was chosen for the detection threshold, cross-correlation coefficient.…”