PROCEDURES FOR ANALYSIS OF DIGITAL TELEMETERED SEISMIC ARRAY DATA

Abstract

Monitoring of earthquakes provides basic data for seismological investigations and research. The advancement of digital technology has made it possible to record the seismic signals from many seismometer stations simultaneously in digital mode at a centralized site for rapid as well as on-line analysis of the recorded digital data. Such data has now become available from the Sample Short Aperture Telemetered Digital Seismic Array in Garhwal Himalaya (TDSA). The data acquisition system in TDSA uses Short Term Average (STA) and Long Term Average (LTA) event detection technique for the recording of the triggered events on 9 track half inch magnetic tapes feeding common time reference from OMEGA world standard time. The main objective of this study is to develop the methods and techniques of digital seismic data processing from array of seismic stations. The complete ground motion time histories were monitored as received from individual stations, initially, to find the optimal values of trigger parameters i.e., STA and LTA periods, threshold level for trigger ratio (STA/ LTA) of individual seismometer outstations, trigger window, pre and post event recording periods and maximum event recording period. These parameters were selected on the lower side so as not to miss any seismic event resulting in recording of many false alarms (Sharma et al., 1987). The background noise level of individual seismometer outstations, array geometry, line of sight conditions of the stations with respect to the central recording station and environmental disturbances were kept in view in selecting the trigger parameters. TDSA being located in a rough terrain and quite remote area, the system has not run on continuous basis due to lack of uninterrupted power supply. In the present study the data for total run time of 781 hours during the period 13-02-89 to 07-05-89 comprising of 530 triggered events has been analyzed and an attempt has been made to optimize the post processing of the recorded events. The initial exercise carried out consisted of development of the software package for Transfer of Magnetic tape Data (TMD) on DEC-2050 computer system (Sharma, 1990) from the data tapes written as blocks of 4096 bytes in industrial compatible format. Based on bit manipulations 11 i TMD software reads two 16 bit words (PDF 11/23) at a time as single 36 bit word (DEC 2050) and then according to the prescribed format the data is demultiplexed and saved in different files changing the binary mode into the integer mode. TMD takes care of the parity bit by supplying the value of previous sample to the corrupt sample in the time series. TMD provides digital seismogram files for each station and a separate file for date and time of the first sample, sampling rate and the total number of samples of each seismogram. The data files are then transferred on PC XT/ AT for post processing of the recorded data. Another software package for automatic determination of on-set time of the first arrivals of seismic signals at different stations has been developed under the present study (Sharma et a!., 1990). The automatic Phase Picker (APP) incorporates the function of an event detector and also obtains precise values of first arrivals for determining hypocentral parameters. APP developed in this study is based on a characteristic function which is a modified time series of ground motion time history. The characteristic function enhances the changes in the amplitudes and/or frequencies of the original time series. STA representing the current signal value and LTA representing the back ground noise level are computed over the characteristic function in a manner to enhance the signal to noise ratio. Once the first arrival is picked, the algorithm enters into confirmatory tests for seismic event and if the time series is passed through the tests then the onset time of the first arrival, up/ down first motion characteristic, maximum amplitude and coda length are recorded. The tests performed are frequency dispersion test, coda shape test and the test for arrival times at different seismometer outstations which is applied after the times from each channel are picked. A modified version of APP software package has been developed in which the threshold value for the trigger ratio is decided by the background noise level. The results of both the techniques are comparable. For the processing of the recorded data, single pole high pass and low pass digital filters are applied to remove DC bias and the high frequency content from the recorded time series. The software package APP is then run to first select the seismic events from the triggered events. APP further computes the onset times of the first arrival of the seismic pluses, up/down first motion characteristic, maximum amplitude and the coda length. For the verification of the results obtained by the APP, the filtered digital seismograms are plotted on the screen of the PC for visual inspection and the onset time and characteristics of seismic signals are marked. This procedure, called visual iv analysis, can achieve the accuracy in picking the first arrival to 2.77 milli sec. However, any error made in marking the right sample on the screen shall affect the accuracy of the first arrival. APP has marked 17 events as seismic events out of 530 triggered events recorded at three or more stations. Only one seismic event is rejected as false alarm due to failing of third confirmatory test in which some noise peak has come just before the first seismic arrival and could not be filtered by the filtering process. All the triggered events are processed by visual analysis and the same 18 events as marked by APP are confirmed to be the only seismic events. The higher number of false alarms are mainly due to environmental disturbances (transmission errors) giving rise to sudden increase in the amplitudes up to ±32768 micro volts and the choice of trigger parameter values so as not to miss low magnitude events. Qualitywise with respect to the times of first arrival of seismic phases as noted down by visual analysis, APP gives the same arrival time for 36 seismograms whereas a difference of 0.0 to 0.5 sec in case of 19 seismograms and 0.5 to 1.0 sec in case of 7seismograms is noticed. In 2 seismograms the first arrival is marked by the visual analysis only and in 5 seismograms the first arrival is picked only by APP. Acomparison of the results of APP with other reported phase picker algorithms (Stewart, 1971, Allen, 1978) shows that APP is better in quality and quantity in respect of first arrival time pickings. The first arrival times obtained by the procedures laid down for the analysis of the events are used to determine the hypocential parameters for the 18 seismic events from HYPO 71 computer program (Lee and Lahr, 1975). Source parameters like stress drop, seismic moment and source dimension etc. have been computed for some events using Brune's model ( Brune,1970) of seismic source. Based on the data from these events, an empirical seismic moment - magnitude relationship has been proposed for this region. The shallow microearthquake events with low stress drops indicate the low resisting power of the upper crust to withhold the strain energy arising due the plate motions in the region.