Two versions of the AS-1 vertical seismometer, built by The Amateur Seismologist of Montrose, California, were operated alongside an STS-2 seismometer in the underground tunnels at the Albuquerque Seismological Laboratory. The older shorter-period design AS-1 was set to a mechanical period of 1 second; the newer longer-period design AS-1 was set to a period of 2.5 seconds. Vertical component data from all three instruments were simultaneously recorded on a Quanterra 24-bit data acquisition system over a period of several weeks. Several teleseismic events in the magnitude range from 6 to 7 were recorded. Noise levels, SNRs, and long term performance characteristics of the two AS-1 instruments are presented, using the STS-2 as a reference instrument. The suitability of the AS-1 instruments for education and outreach purposes is discussed.
Two AS-1 vertical seismometers were setup in the ALQ underground tunnel alongside a standard STS-2 broadband seismometer. The older design AS-1 (one-piece base) was set to a mechanical free period of 1 second, and is called the "AS-1 (short period)" instrument here. The newer AS-1, called the "AS-1 (long period)," is constructed using a two-piece base, and was set to a mechanical free period of 2.5 seconds. All three seismometers were covered with insulating covers to protect them from air drafts and for thermal stability. The AS-1 seismometers were connected through their standard "Black box" analog amplifier/filters to a Quanterra 24-bit data logger. (The 8-bit and 12-bit digitizers inside the AS-1 "black boxes" were not used.) The STS-2 vertical component was also connected to the Quanterra. Data sampled at 20 samples per second from all three instruments were recorded for 11 days. We present a noise analysis covering all 11 days: March 29 to April 9, 1999.
The older AS-1 instrument (one-piece base) was set to a free period of 1 second. The newer AS-1 (2-piece base) was set to a free period of 2.5 seconds. The oil damping cups on both AS-is were filled with 30-weight motor oil. The basic response-determining elements for the AS-1s are: the free period of each instrument, the oil damping, and the 3-pole filter in the AS-1 "black box". For the STS-2, we assume that the transfer function supplied by the manufacturer is correct (flat to velocity from about 120 seconds to approximately 70 Hz).
The signal from a M~ 6.4 earthquake on March 31, 1999, south of Panama, recorded by all three instruments, was analyzed to more accurately determine the transfer functions of the AS-1s. Spectra for this event, corrected for each instrument's response, were calculated and compared (see Figure 1a and Figure 1b
The AS-I trial transfer functions (based on the response-determining elements) were adjusted to achieve spectra approximately matching that from the STS-2 for this signal, which contains energy well above ALQ background noise over a broad range of frequencies. As we see in Figure 1, the match is not perfect. There is a ~3db error in the amplitude response of the 1-sec AS-1 from 2.5 seconds to 15 seconds period, resulting in the corrected earthquake amplitude spectra being offset in this period band. We suspect that non-linear electrolytic capacitors used in the AS-1 "black box" filter are the culprit, resulting in a non-linear circuit. To emulate the black box we used two single-pole low-pass filters at 6 seconds and a one-pole high-pass filter at 20 seconds.
The STS-2 amplitude response and the amplitude responses determined for both AS-1 instruments (with their respective "black boxes") are shown in Figure 2.
Spectra for the entire 11-day test period, corrected for instrument transfer
functions, are calculated. There are 518 different 1-hour segments, with 50% overlap. These spectra are presented in
along with the median (bold white curve). We use the median as a robust estimate of the total noise power, as it effectively removes spurious signals and earthquake signals. Several things are immediately obvious:
|Vault STS-2: The instrument with the lowest noise exhibits the widest range of power levels at nearly all periods, indicating that it is responding to changing background noise conditions and both smaller and larger events.|
|Original AS-1 (short period): The instrument with the highest noise at periods longer than 10 seconds exhibits only a very narrow range of power levels in this period range, indicating that it is not responding to changing background noise conditions. The simultaneous data from the STS-2 indicates that the actual earth background noise is too small for this instrument to observe in this band. The same is true for periods smaller than about 0.5 second.|
|New AS-1 (long period): A well-known fact about mechanical seismometers is that, at periods longer than the natural period of the instrument, sensitivity increases as the square of the instrument free period. Therefore this 2.5-second instrument should theoretically have a long period signal-to-noise ratio (SNR) that is (2.5/1)2 = 6.25 higher than the 1-second AS-1, other things being equal. Figure 3 bears this out, as the long period AS-1 noise power is lower, and demonstrates more scatter, than the short period AS- 1 at periods longer than about 10 seconds.|
The medians from Figure 3 are plotted in Figure 4
for direct comparison. Some of the difference between the short period AS-1 (red curve) and the STS-2 (black curve) in the period range from approximately 2.5 seconds to 15 seconds is likely due to the problem in determining a linear transfer function for the short period AS-1.
depicts the difference in total noise power between the STS-2 and each AS-1.
Above are three columns of 24-hour seismograms that we call "heliplots," for data from March31, 1999. The left column contains unfiltered seismograms, one from each instrument, plotted at the same magnification at 8 seconds period. The middle column consists of long period seismograms produced by low-pass filtering the same 24-hour period of data, plotted at a common magnification at 20 seconds period. Finally, the right-hand column contains short period seismograms, band-pass filtered around 1 second, and plotted at a common magnification at 1 second period.
We see in the Unfiltered heliplots that the signal from a M~ 6.4 event at a distance of 36 degrees (Panama) is clearly seen on both AS-1 instruments. The surface waves from this event are best seen by the STS-2, which has better sensitivity and lower noise in the long period band.
In the center column of Low-Pass Filtered at 20 Seconds helipiots, we see further evidence of better long period noise performance of the long period version of the AS-1 (compared to the short period AS-1).
We see in the right column of heliplots, which are Band-pass Filtered at 1 Second, that a few body phases from a number of smaller events are also recorded.
Noise Pulses: All of the AS-1 heliplots demonstrate spurious noise pulses. However, the short period AS-1 seems to have fewer pulses than the long period AS-1, possibly indicating better mechanical stability.
11-Day Time Series: Figure 6
contains the entire 11-day unfiltered time series used in the noise analysis. Several events are clearly seen on all three instruments.
The simple design of the AS-1 seismometer makes it ideal for use in the classroom for demonstrating the simple principles of seismometry. This is not the case with many other types of instruments, such as miniaturized broadband feedback instruments in which the mechanical workings are not so obvious or are hidden inside a metal case. The AS-1 is capable of detecting signals from moderate to large teleseisms, as well as smaller local and regional events, even after having been dis-assembled and re-assembled by a high school earth science or physics class. The disadvantage of the AS-1, compared to a typical broadband seismometer, is its limited sensitivity and higher noise in the long period band.
The IRIS Education and Outreach Program has decided to place an order for 40 AS- is for distribution to participating schools. IRIS ordered the short period (one-piece base design) as it is simpler to work with and is more stable than the long period version. The short period version should well serve both educational and observational purposes, even though its long period noise is somewhat higher than that of the long period version.