Continuous Monitoring of Ground-Motion Parameters

We applied a time-domain method to continuously monitor groundmotion parameters to facilitate rapid determination of the geographical distribution of ground motion and earthquake parameters. This approach minimizes the impact of the sudden increase in the workload on data acquisition facilities and streamlines the operation of a seismic network, especially during a complex earthquake sequence. The incoming continuous time series is processed with the use of various timedomain recursive filters to compute ground-motion velocity, acceleration, energy, Wood-Anderson seismograms, and narrow-band responses at 0.3, 1.0, and 3.0 sec, which are used for computation of response spectral amplitudes. The method is implemented in the Southern California Digital Seismic Network. The method can be implemented at every field station, which would make it possible to process the time series locally and continuously telemeter desired amplitude parameters with sufficient accuracy from a field station to multiple users through a relatively low data-rate communication line (e.g., regular telephone line or limited bandwidth satellite link). Introduction As modem broadband and wide dynamic range seismic instruments have become widely available, real-time monitoring of earthquake ground motion is becoming an important function of regional seismic networks for earthquake hazard mitigation purposes. Relevant amplitude parameters include acceleration, velocity, displacement, energy, WoodAnderson response, and narrow-band responses at various periods to be used for computation of response spectral amplitudes. Traditionally, most seismic networks operate in trigger mode. Even if the data are recorded continuously, computation of amplitude parameters is initiated only when a significant seismic event is detected. In this mode, the processing of data must wait until the recording of the trigger is completed. In addition, the workload on the system increases suddenly during a major earthquake, which could cause a system failure during the time when real-time data are most needed for emergency services. To allow real-time processing and to alleviate the overload problem, we developed a continuous monitoring method that continuously computes ground-motion parameters regardless of whether earthquakes are occurring or not; an earthquake is not treated as a special event but is part of continuous perturbation of ground motion. This minimizes the fluctuation of workload and assures more reliable overall operation of the system. In practice, we replace the traditional frequency-domain analysis with application of a set of time-domain recursive filters and process data, sample by sample, as the signal comes in through telemetry. This is a simple conceptual change, but we believe that it would lead to a significant improvement in seismic network operation in the future. The purpose of this article is to address some technical issues with the hope that a continuous method can be easily implemented, if desired, in other networks with a similar objective: reliable real-time ground-motion monitoring. Also, the method described here can be implemented in a data logger itself. Such a data logger can process data on site to compute desired amplitude parameters (acceleration, velocity, displacement, Wood-Anderson, and spectral responses, etc.) and send them to users. In this case, because the computed amplitude parameters can be sent at a relatively slow rate (e.g., 1 sample/sec), even a relatively slow communication line (e.g., regular digital telephone line) would be adequate. This would allow for easy simultaneous reception of the desired information at multiple locations to avoid a single point failure of the system. We have tested the method on-line in the Southern California Digital Seismic Network (e.g., TERRAscope and TriNet) for nearly 2 yr. This continuous system is currently used for on-line magnitude reporting and for broadcasting ground-motion data (acceleration and velocity) through our web site: www.trinet.org as ShakeMap (Wald et al., 1999). In the following, we describe the method applied to broadband instruments and strong-motion accelerographs. The broadband instruments used here are Streckeisen STS1, STS-2, and Guralp CMG-40T seismometers, and the strong-motion instrument is a Kinemetrics FBA-1 accelerograph. The method should be applicable to other instruments with similar characteristics. The broadband channel with a sampling interval (At) of 0.05 sec is called the vbb 311