Introduction
Since 1973, a group of geologists and geophysicists from the University of Wisconsin-Milwaukee have studied the large floating fresh-water ice sheets formed during the winter months on northern Lake Michigan and Green Bay, Wisconsin, U.S.A. The continuing studies include: (1) a study of variations in lake-ice fracture patterns as related to snow cover, ice thickness, crystal orientation and size, and air, water and ice temperature, and (2) a series of seismo-acoustic investigations of the elastic properties of fresh-water lake ice. Relatively little work of this type has been done on fresh-water naturally-occurring lake ice on large lake bodies such as the Great Lakes of the north-central U.S.A. The data derived from the studies are important to an understanding of lake-ice deformational processes and stress propagation within the ice.
Monitoring Seismic Events
The long-term monitoring of transient seismic phenomena is one method for the investigation of body stresses, stress changes and displacements (Reference Lasca, Lasca, Taylor and WillisLasca and others, 1975). The need for actual wave shapes and accurate arrival times, however, requires the use of recording instrumentation with a pass band on the order of D.C. to too Hz, and time resolution on the order of i ras. The requirements effectively eliminate analogue paper recorders and require the use of magnetic tape recordings. While low-speed magnetic lape recorders can provide the necessary band-width and time resolution, the recovery of a few transient events from a long-term field tape requires the attended laboratory play-back of the tape. Thus it is necessary for the operator to scan the entire field tape, possibly with time compression through the use of higher play-back speeds, and to select each event manually. In actual practice this may require several passes through the lape, which severely limits the utility and reliability of the original data. To improve data reliability and to eliminate manual recovery of seismic event data from continuously recorded field tapes, an automated seismic recovery system was designed.
Automated Seismic Recovery System
An automated means of signal recovery or storage may be achieved by (1) monitoring the output of one or several input data channels and (2) starting the recording process only after the output level has reached some predefined level. The number of false starts due to noise bursts can be reduced by the use of R.M.S. amplitude in some predefined time window. The majority of seismic wave-forms, however, are characterized by peak amplitudes arriving after the initial arrivals. Tor reliable operation the predefined R.M.S. starting level must center on the high-amplitude, late arrivals. Therefore, to record entire wave trains, it is necessary that the recording process be initiated prior to any output from the monitoring system. In practice this requires a delay between the incoming data and the permanent recording of these data or what is commonly termed "signal recovery in retrospect".
The system for recovery in retrospect described here is relatively inexpensive (approximately $700), utilizes standard components, and is intended as a simple addition to an existing field system rather than a complete replacement. The required circuitry may be used directly in the field or in the laboratory for unattended recovery of data from a field tape.
A block diagram of the total data-collection system is shown in Figure 1. The required delay is achieved through a NIMBUS Digital Data Delay which will accept up to 8 analogue inputs; this system is produced by Nimbus Instruments of West Sacramento, California. In operation the incoming data are converted to digital form in the delay circuit and delayed for a time dependent upon the digitizing frequency. Following this delay, the signals are put out as discrete voltages at the sample rate. The initial low-pass filtering of the input is necessary to prevent aliasing, or the generation of lower-frequency signals as a result of improperly sampled higher frequencies. The final low-pass filter is required for a continuous analogue signal. The maximum allowable frequency, delay value, and number of input channels are set by initial design parameters. The system provides a maximum frequency of 125 Hz with a delay of 1.024 s on four channels of input data.
Block diagram of the automated seismic recovery system used for signal recovery in retrospect.
The event detector (Fig.1) is shown in greater detail in Figure 2. The detector consists of TTL integrated circuits, beginning with a Schmiit-trigger fed through an attenuator. The trigger output lights an LED to assist threshold adjustment, and simultaneously fires the first one-shot multivibrator. Λ half-second delay is programmed into the one-shot multivibrator to assure proper recording of low-level signals arriving after the trigger initiating burst. The first multivibrator starts the recorder via a reed-relay. The recorder then records the output of the NIMBUS circuit or all incoming data 1.024 s prior to the start of the recording process. As the first multivibrator returns to normal state, it fires a second one-shot multi-vibrator, programmed at one second, to keep the recorder running, thus insuring that data still in the NIMBUS circuit are recorded. Both multivibrators are re-triggerable so that a multiple- or long-duration event is completely recorded.
Block diagram of the event detector.
In operation the recovery system is highly reliable and simple to operate. It may be used (1) to initiate high speed (on the order of 100 cm/s) paper recordings in the field, (2) to store transient events on magnetic tape while in the field, or (3) for unattended recovery of transient events from a continuous field recording. The latter approach is the most desirable as the field recording may be searched at several detection thresholds, and the field-system weight is minimal. We have used the system for the completely unattended recovery of 43 seismic events, recorded at four different sites, occurring during a continuously recorded 72 h period. Manual recovery of the events from the continuous field tapes would have required at least two man weeks.
Acknowledgement
The research which led to the development of the automated seismic monitoring system was supported by N,S,F. Grant No. KG-41011 for the study of the elastic properties of freshwater ice.
