5:15 PM - 7:15 PM
[SCG63-P02] Implication of maximum amplitude and P wave arrival time: consideration from comparing velocity seismogram with accelerogram at the same location
Keywords:earthquake early warning , velocity seismograph, saturation, anti-aliasing filter
To increase observation sites for earthquake early warning, we have investigated a method to estimate seismic intensity (on JMA scale) from velocity seismograms, in which conversion from velocity to acceleration, correction of instrumental characteristics, and process for estimation of seismic intensity are conducted. We indicated coincidence of two waveforms from velocity seismograms and accelerograms. When they differ, we should expect some malfunctions in the observation. Monitor of the difference is useful for quality check of input data and early detection of the malfunctions.
One of typical malfunctions is saturation. There are two types: saturation of digitizer, and saturation due to limitation of pendulum stroke in seismograph. Saturation of digitizer is easily detected, because the output is flat in seismogram and the upper-limit is described in specification documents. On the other hand, saturation due to limitation of pendulum stroke is not so easily identified from velocity seismograms directly. Appling filter to obtain relative displacement make it easy to detect it. Saturation of digitizer is sometimes found in velocity seismogram at JMA observation sites, and saturation due to limitation of pendulum stroke is occasionally found in data from NIED Hi-net where the stroke is about 1mm (Obara et al, 2005; Shiomi et al., 2005). Even when saturation due to limitation of pendulum stroke occurs, the apparent maximum amplitude does not necessarily indicate under-estimation: it sometimes indicates over-estimation.
Careful comparison of the two waveforms (from velocity seismogram and accelerogram) sometimes indicates the slight difference in time (2-3 samples for 100 samples/s data). This is visible even for arrival time of spike-like P wave of very local earthquakes. Possible reason of the time difference may be caused by the difference of anti-aliasing filter. Some anti-aliasing filters have characteristics of slight decrease, and some filters have steep decrease with increasing frequency in the range of higher than the cut-off frequency. Impulse responses show that steeper decrease indicates larger apparent time delay. This implies that difference of apparent time delay is caused by the difference of characteristics of anti-aliasing filter.
Maximum amplitude in velocity seismograms may be sometimes contaminated by the limitation of pendulum stroke, and arrival time of P wave may be sometimes delayed apparently due to the anti-aliasing filter.
One of typical malfunctions is saturation. There are two types: saturation of digitizer, and saturation due to limitation of pendulum stroke in seismograph. Saturation of digitizer is easily detected, because the output is flat in seismogram and the upper-limit is described in specification documents. On the other hand, saturation due to limitation of pendulum stroke is not so easily identified from velocity seismograms directly. Appling filter to obtain relative displacement make it easy to detect it. Saturation of digitizer is sometimes found in velocity seismogram at JMA observation sites, and saturation due to limitation of pendulum stroke is occasionally found in data from NIED Hi-net where the stroke is about 1mm (Obara et al, 2005; Shiomi et al., 2005). Even when saturation due to limitation of pendulum stroke occurs, the apparent maximum amplitude does not necessarily indicate under-estimation: it sometimes indicates over-estimation.
Careful comparison of the two waveforms (from velocity seismogram and accelerogram) sometimes indicates the slight difference in time (2-3 samples for 100 samples/s data). This is visible even for arrival time of spike-like P wave of very local earthquakes. Possible reason of the time difference may be caused by the difference of anti-aliasing filter. Some anti-aliasing filters have characteristics of slight decrease, and some filters have steep decrease with increasing frequency in the range of higher than the cut-off frequency. Impulse responses show that steeper decrease indicates larger apparent time delay. This implies that difference of apparent time delay is caused by the difference of characteristics of anti-aliasing filter.
Maximum amplitude in velocity seismograms may be sometimes contaminated by the limitation of pendulum stroke, and arrival time of P wave may be sometimes delayed apparently due to the anti-aliasing filter.