School Yard Seismology
Yu. Fedorenko1, E.S. Husebye 1 and E. Boulaenko 2,|
1 Inst. of Solid Earth Phys., University of Bergen, Allé gt. 41, 5007 Bergen, Norway
2 Inst. for Analytical Instrumentation, Saint-Petersburg, Russia |
Background -
Novel Seismograph Design -
Real Testing of "Geophone" Seismograph -
School Yard Seismology -
Conclusions -
Acknowledgements -
References -
Background
Advancements in solid earth physics are driven primarily by access to huge amount of high quality wavefield data as recorded by globally distrubuted seismograph stations. Since these broadband stations are spaced hundreds of kilometers apart the spatial resolution are also of the order of hundreds of kilometers. For localized studies of crust and upper mantle structures significantly smaller station separations are required say as once realized through the large aperture arrays NORSAR and LASA. In this regard, instrumentation and operational costs become critical factors.Seismologist, in particular in academia, have tried for decades to design a simple, robust and inexpensive seismograph stations which would produce wavefield recordings of adequate quality for research ((Husebye & Thoresen, 1984; Husebye et al., 1984). Again, low operational costs are essential for network operations for more than a few years that is beyond a funding period of typical 3 - 5 years. In this research note, we describe our efforts in designing an inexpensive 3-component short period seismograph station which first deployment is in the school yard of Åsane Gymnas - a high school in Bergen. The motivation for school installation is two-fold; 1) force science interest by students through 'do and learn' experiments tied to an exciting natural phenomenon like earthquakes and 2) to operate a seismic network for research while minimizing running costs. The interest among Norwegian schools in participating is highly satisfactorily so our vague ideas on such an undertaking just a few months ago must now be rated a Project including Web pages and news services. In the last chapter we will outline the structure and future plans of this high school educational project with its routing in observational seismology.
Novel Seismograph Design
Initially, our efforts here were initiated by the need for extensive testing of signal detectors, phase pickers and signal recognition for seismic source identification (Fedorenko et al., 1998, 1999, Fedorenko and Husebye, 1999). In principle it should be simple to retrieve such data from various kind of national and international data centers but in practice not; not at least regarding short period (SP) recordings (monitoring data). Most effective way of handling such tasks are simply to pay research visits to the data center itself. With this motivation but no desire nor resources at hand for reinventing the seismograph, we used standard, inexpensive instrument components for our novel 3-component seismograph design outlined below. It is novel in the sense that station operation constitute an integral part of the school PC-Internet system. Also, students are given access to professional processing software like PITSA so they can undertake signal analysis of their own say investigating crustal structure locally. The choice of instrumentation was motivated by easy commercial access to inexpensive but robust geophones which through proper preamplifier design may mimic SP seismometers as used in say CTBT monitoring. Broadband instruments are more bulky, expensive and more difficult to operate by amateurs; if really needed students can access individual BB stations, or IRIS and ORFEUS data bases.Geophones Substituting Seismometers
Even in 1960-ties academic institutions experimented with small, inexpensive geophones around $60 each as a substitute for bulky, expensive seismometers (around $2000 each). In the 1990-ties more widespread use of geophones in the role of seismometer become more common in particular where compact size is needed like in ocean-bottom seismometer (OBS) surveys. Our current research interests are for 0.5-40 Hz of the seismic wavefield so we settled for the GS-11D geophones from the Geospace Corp. Huston, TX which has natural undamped frequency at 4.5 Hz. Naturally, a geophone is not a seismometer but may be modified to perform in such a manner. This was achieved in combination with our own elaborated preamplifier with an approximately flat acceleration response in the 0.5 - 40 Hz frequency range (Fig 1).The motivation for an approximately flat acceleration response is twofold; firstly because this is an advantage/requirement when using the efficient wavelet transform for noise suppression (Husebye & Fedorenko, 1999) and secondly a ground acceleration is a common parameter in seismic loading and risk studies. Most SP seismometers measure ground velocity but since response curves are not always flat the output is in fact a mixture of ground velocities and accelerations.
Figure 1. Approximately flat acceleration response for the GS-11D geophone in combination with our own designed preamplifier. In essence, low-cost geophone transformed to high-cost short period seismometer by smart "response" hardware.
Analog to Digital (A/D) Converter
Recently, costly and precise 24-bits A/D have been introduced into seismograph stations which is sort of sampling " overkill" since there is no customary preamplifiers or commutators with dynamic range close to 24-bits. On the other hand, the commonly used 16 bits A/D has a somewhat limited dynamic range at 96 db thus occasionally causing clipping of local EQ recordings. However, it is a well known fact that low-pass filtering of quantization noise by oversampling improves the accuracy of the A/D - converter. We have taken advantage of this thus in effect converting the standard 16 bits A/D converter to one as accurate as 19 - bits with effective dynamic range not worse than 18 bits (Fig. 2) deemed adequate for SP seismic recording systems. Presently the primary sampling rate is 3150 Hz for each channel, while output sampling rate is 50 Hz. In our A/D design we have also built in an effective spike suppressor; in some environments like buildings with much electronic equipment such spikes are a real nuisance.
Figure 2. Quantization noise reduction by oversampling. Upper line at both panels represents ADC (A/D converter) output noise without oversampling, fs/fout = 1. The next lines correspond to oversampling fs/fout = [4,16,64], respectively. Notice that quantization error with oversampling by 64 is approximately 8 times lower (3 bits gain) than without oversampling. Quantization ADC steps are in units of noise standard deviation, ADC error RMS is [1/L-1 ... (nj-njq)2]1/2, where nj is the initial noise, njq is the quantized noise and L is number of samples in the noise realization.
GPS, Radio Clocks and Internet Timing Systems
Effective timing of seismic recordings are important since in very localized surveys, time anomalies of the order of 0.1 sec may be of significance. The most accurate and versatile of these timing systems are the GPS-clock (price around $ 300) which accuracy is at least 0.001 sec or 1 msec. An occasional draw back here is that an external antenna is needed and then expose station to possible unfriendly human interests. Radio clocks (about $40 ) may also be used but the transmitters in Rugby, UK and DCF-77 near Potsdam, Germany are not always accessible in NW Europe due to wave propagation effects. Disruptions in timing signals are not uncommon. The optimal timing solution would be to use the Internet that is to extract timing information from time servers which in combination with proper managing software can be obtained with the maximal error about 5 msec and mean error of 0.5 msec. Since many schools at least in Norway maintain permanent Internet connection this timing option is very attractive since it is free. We plan to install "our" 3-comp. seismic stations in Central America and here GPS timing will be used.Non-continuous transmissions may cause problems necessitating so called soft system time adjustment. With this is meant that if external say GPS clock is lost for even a few minutes the time error may exceed the sampling period. In other words, occasionally the 'true' sampling rate may exceed the preset sampling rate unless the soft system time adjusting is part of the total timing system as in case of our design.
CPU and Data Storage
We use an inexpensive CPU unit (Intel 486 or Intel 586 at a cost of $150 ) which is adequate for signal detection processing in several frequency bands. Via Internet or modem/phone detected signal parameters can be transmitted to the Hub while continuous waveform storage is on a 8 Gigabyte disk (cost about $ 80). Accessing waveform data at the station site via the Internet is very attractive due to cost savings and as such necessary for any "school project". Total cost of our novel seismograph now being operational in the Åsane school yards is less than $ 1000. - work efforts and CPU cost excluded.Real Testing of "Geophone" Seismograph
The above design features are incorporated in our new seismograph which size is minimalistic as compared to conventional seismometers (Fig. 3). To test relative performance we installed our station on the same site as one of the stations in Norwegian Seismograph Network (NSN). A local EQ recording at both of these stations are shown in Fig. 4. Since ASK (Kinemetrics Ranger) records ground velocity while our ground acceleration the records are not identical - as expected. Ideally, we would like to compare strong local event recordings with those of a nearby BB station but so far this has not been feasible. As regards seismic event detectability a direct comparison between the two station is not feasible since we use a more advanced 2-D signal detector operating in 3 frequency bands producing relatively many signal detections.
Figure 3. The "School Yard" 3-component seismograph station as installed on solid rock at Åsane Gymnas near Bergen. The three 4.5 Hz geophones are clearly seen while the electronic card is the preamplifier. Instrument box size is 9x11 cm while the field casing box is 20x23 cm. Recordings by this school yard seismograph are shown in Fig. 4.
Figure 4. OUR and ASK station recordings of local earthquake; differences in waveforms are attributed to differences in sensor responces.
School Yard Seismology
There is much ongoing efforts aimed at stimulating interest among high school students with a view to also ensure better enrollments in geoscience disciplines at universities (e.g., see the ORFEUS edu-links). From a professional seismological point of view such efforts would be laudable if school yard seismograph deployments will contribute constructively to local seismicity monitoring and crustal tomography studies with high quality wavefield recordings. The crucial element for success is that every participating school have their own seismograph station producing earthquake records of professional quality. Likewise, software for signal analysis must also be available including tasks like signal detection, event location and Richter magnitude estimation - for a start. To test this 'hypothesis' we have as a pilot project installed our novel seismograph station in the school yard of the Åsane Gymnas; we proceeded in the following manner:- Wooden frame 23x20 cm "fasten" to exposed rock with fast drying cement. The instrument box measures 9x11x5 cm. The site itself is hidden under vegetation to prevent theft and damages.
- Cable with grounding connects the 3-comp. seismometer to the CPU and A/D-converter units in the physics laboratory room inside the school. This CPU has as permanent Internet connection ensuring free data transfer.
- For student signal processing the PITSA software package has been installed.
As mentioned school installations also aim at stimulating student interest in geoscience per se but also as a mean to add a new dimension to physics and mathematics teaching. For example, picking P- and S-phase arrivals from local event recordings and then performing epicenter locations are popular exercises. More advanced problems are tied to spectral decompositions, filtering and wavefield polarization analysis. Using data literally from their own backyard are strongly motivating for students interest here which may be helpful in tracking down local quarry blasting sites.
Indeed, to "protect" ourselves from over eager students we try to organize our involvements in a rational way. Firstly, for Åsane the teacher N. A. Eldholm has set up a Tiger Team of students (3x2-persons groups) each of which are given specific working tasks. One team would participate in software development for simplified seismogram analysis of local recordings, another aim at locating local explosions and earthquakes while the third team would present geoscience structural knowledges pertinent to the Bergen area. Each of these teams would be responsible for liaisons with fellow students. A particular popular task is that of making the mandatory high school thesis problems geoscience oriented. We are also asked to create sort of News Letter service for information sharing and inter-school cooperations. Very exciting and encouraging but also demanding - hopefully also rewarding in the future.
Conclusions
We have demonstrated that it is technical feasible to construct, deploy and operate a low cost 3-comp. seismograph (less than $ 1000.) which recording performances match those of a conventional, national network SP seismograph costing at least $ 5000. Through Internet connection the need for radio/GPS clock is eliminated and likewise data transfer costs. Suitable instrumentation sites are often problematic but far less so if installed on exposed rocks in school yards. Such sites may be relatively noise but in practice not much of a problem since station density is most important in local geodynamics and tomography studies.Network operation and data base organization may be costly undertakings but school yard installations with enthusiastic students in the roles of managers and analysts appear to be a solution to such problems. Perhaps, most important in the context of promoting science in high schools is in fostering science enthusiasm among students actively recording and analyzing earthquakes. Perhaps appropriate to quote one of Åsane "tigers"; "... now I see why maths and physics are important - how can I otherwise understand my school seismograms !"
Further plans: High school curriculas are very tight and modest in geosciences. However, a station at your doorstep create interest in terms of active participation when EQ occur either locally or disasters at far away places. In practical terms this means that math and physics problems are tied to their own EQ recordings. We will help with project work - sort of research on high school level.
Project success depend foremost on local teachers: our contributions here is a Challenging Web page(s) including factual info on geodisiplines and naturally seismology, providing analysis tools like PITSA for signal processing, GMT mapping tool for displays etc. In a longer time perspective, summer schools would be arranged at the Hub (University of Bergen) for teachers to master signal processing, basic seismology and foremost ways of incorporating geosciences in mathematics and physics problem exercises. An other important aspect is to give a hand with selected students Project work say locating local earthquakes, estimating crustal thickness, analysis of macroseismic questionnaires and so forth - goal is to retain science interest through our SEIS-SCHOOL project for many years to come.
Acknowledgements
We want to express our gratitude to Nils A. Eldholm and Sturle Kalstad, Åsane Gymnas and O. Dahle, Sotra VGS for inspiration and enthusiasm in our school yard experiment described here and not at least the learning spirit of the Tiger Team in Seismology at these schools. The research reported here was supported by the Defence Treat Reduction Agency, Department of Defence, USA under Grant DSWA 01-98-C-0159.References
- Fedorenko, Yu. V., E. S. Husebye, B. Heincke and B. O. Ruud, 1998. Recognizing explosion sites without seismogramm readings: neural network analysis of envelope-transformed multistation SP recordings 3-6 Hz. Geophys. J. Int., 133, F1-F6.
- Fedorenko, Yu. V., E. S. Husebye and B. O. Ruud, 1999. Explosion site recognition; neural net discriminator using three-component stations. Phys. Earth Planet. Int., 113, 131-142.
- Fedorenko, Yu. V. and E. S. Husebye, 1999. First breaks - automatic phase picking of P- and S- onsets in seismic records. Geophys. Res. Lett., 26, 3249-3253.
- Husebye, E. S. and E. Thoresen, 1984. Personal seismometry now! EOS, 65, 441-442.
- Husebye, E.S., S.F. Ingate and E. Thoresen, 1984. Seismic arrays for everyone. Terra Cognita, 4, 414 - 422.