Long Term – Real time Background Noise monitoring Around BR235 Nurcan Meral Ozel, Serdar Kocak, Korhan U. Semin, Ocal Necmioglu, T. Cem Destici, Ugur Teoman, Robert Kemerait*
Boğaziçi University Kandilli Observatory and Earthquake Research Institute, Belbaşı Nuclear Tests Monitoring Centre, Ankara, TURKEY
*Air Force Technical Applications Center, USA Contact: [email protected] or [email protected]
CTBT Science & Technology 2011 Conference 8 – 10 June 2011, Vienna, AUSTRIA
Abstract
Turkish NDC is monitoring the quarry activities in the vicinity of array elements in coordination
with AFTAC. The mining activity around one of the array site BR235 was analyzed in the period
of August 2008 and December 2010. The power density spectrum of BR235 long period data was
compared with the other elements of the same array in order to evaluate the overall noise effect
on the BR235 data. Power density spectrum analysis allows us to determine the precise frequency
characteristics of the background noise, which will help us to assess the station sensitivity. The
long period data are important for nuclear explosion monitoring, primarily for estimating Ms
magnitude; consequently measuring the mb:Ms discriminant. Our preliminary results show some
difference in the 10 – 30 second period range amplitudes for these time periods of the data. A
detailed analysis of recorded quarry blasts and activities, together with the preliminary results of
the noise will be presented by this research.
Introduction
Bogazici University and Kandilli Observatory and Earthquake Research Institute (KOERI) act as
the Turkish National Data Center and is operating IMS Primary Seismic Station (PS-43) under
Belbasi Nuclear Tests Monitoring Center for the verification of compliance with the
Comprehensive Nuclear-Test-Ban Treaty (CTBT) since February 2000. The NDC is responsible
for operating two arrays which are part of the International Monitoring System (IMS), as well as
for transmitting data from these stations to the International Data Centre (IDC) in Vienna.
PS-43 is composed of two sub-arrays (Ankara and Keskin). The medium-period array with about
38 km diameter located in Ankara and the short-period array with about 3 km diameter located in
Keskin. Each array has a broadband element located at the middle of the circular geometry. Short
period instruments (Geotech 23900A) are installed at depth 30 meters from the surface while
medium (Geotech KS54000) and broadband instruments (Geotech KS54000) are installed at
depth 60 meters from surface. Medium period instruments are operating with 4 sps (sample-per-
second) sampling rate, broadbands at 40 sps and short period ones are operating at 20 sps.
Freewave Wireless Data Transceivers which form the basis of RDL provide intra-site
communications from all instrument sites. The data from both arrays comes to CRB (Central
recording Building) and data is sent to IDC via satellite (VSAT) whereas KOERI (Kandilli
Observatory and Earthquake Research Institute) and AFTAC (American Air Force Technical
Application Center) receive the data via telephone lines.
Figure 3: Comparison of number of quarry blasts recorded by BR235 and declared by company
in the period of August 2008 – December 2010. Figure 1: Configuration and location area of Medium-Period array.
Quarry Blast Activities Around Site BR235
Since the beginning of 2007, there have been 7 different applications for constructing stone
quarries very near to the site BR235. One of the companies (REMTUR) especially operates closer
than 1.5 km to this site. The exact dates and times of the explosions have been requested from the
quarry company. The quarry’s effect on the data is closely monitored by Turkish NDC and
AFTAC since August of 2008. In this study, we tried to assess the effect of the quarry generated
noise on the BR235 array data. Figure 2 shows us the relative location of quarry and the site
BR235.
Figure 2: Boundaries and the distance of the stone quarry from the BR235 is shown here.
As we can see from Figure 2, the stone quarry operates too close to our station. Its known that
quarry blasts that are very near the seismic sensors can cause problems in data quality. Therefore,
we have decided to make a noise analysis in order to estimate the overall effect of the blasts on
seismic data.
According to our agreement with the quarry company, we have been given a list of blast date and
times for each month by the company since August 2008. Following figures show our analysis of
the data for the given dates and times, number of blasts. If we look at the histogram graphs, we
see an increase in the number of explosions every year. Recorded explosion quantity lowers
during winter season as expected. Also the company started to make 3 explosions per week since
April 2010, which explains the increase in the numbers in 2010.
Quarry Blast Noise Analysis
Data Analysis
The dates and times of explosions given by the company have been checked by reviewing the
corresponding data. We have analyzed whether an explosion occurred on the given time and date
or not. The noise effects on the BR235 data were analyzed with following methods;
- Power Spectrum Density (PSD) and Probability Density Functions (PDF) calculation
We have compared the BR235 data with other elements of the array in order to evaluate the
noise effect on other sites.
Probability Density Functions
We have used the algorithm developed by McNamara and Buland, in order to estimate the true
variation of the background noise at a given site. In most noise studies, body and surface waves
from earthquakes, or system transients and instrumental glitches such as data gaps, clipping,
spikes, mass re-centers or calibration pulses are removed. These signals are included in our
processing because they are generally low probability occurrences that do not contaminate high
probability ambient seismic noise observed in the Probability Density Functions (PDF). In fact,
examination of artifacts related to station operation and episodic cultural noise allows us to
estimate both the overall station quality and a baseline level of earth noise at each site.
For each channel, raw frequency distributions are constructed by gathering individual PSDs in the
following manner:
1) binning periods in 1/8 octave intervals; and
2) binning power in 1 dB intervals.
Figure 4: Example of a PDF
figure. Colour bar on the right
shows the probability values.
HNM and LNM are reference
high and low noise models by
Peterson(1993). We can easily
see culturel, microseismic noises
and the system artifacts
(calibrations, gaps vs.) System
artifacts are low probability
occurences (purple colour).
Figure 6: Monthly PSD, PDF estimation of
BR235 and 3 other array sites. Here we show
the analysis of May 2010, one of the most
active months in 2010. There are 10
explosions in total, which can be easily seen
at BR235 PSD graph (upper-left figure).
PSDs of the explosions are higher than the
NHM model values in most of the frequency
band. There is a 30 dB difference with the
PSDs of the explosions and the background
noise. Other stations also seems affected in
1-2 Hz frequency band by the explosions.
Figure 5: An explosion near BR235 on June 30, 2009 at 13:05 (a) shows the zoomed waveform and the amplitude spectrum of the explosion at
BR235, (b) the same explosion seen across the array, (c) the power spectrum of the explosion for different sites.
(a) (b) (c)
Figure 7: Yearly PSD and PDF background noise results for BR235, BR234 and BR236. The black line shows the average noise levels for each
station.
PSDs of the explosions.
During this study, we have started with daily analysis, waveforms were reviewed for the recorded explosions and compared with the explosion
dates. Recorded explosions were saved for further analysis such as amplitude spectrum, PSD and PDF calculation. Statistical information were
extracted about the number of quarry explosions. The data was processed with PQLX software in order to get PSD and PDF graphs. Figure 5, 6, and
7 shows us some of the results obtained from PQLX software.
Conclusions
- Since 2008, the number of quarry explosions show an increase each year. Winter time there are less explosions but during summer there are
more than one explosions per day, usually within 10 minutes. The actual explosion times and the declared times generally doesn’t match in 2009
and 2010.
- If there is an earthquake occured at the same time with the explosion, the data becomes degraded for the duration of the explosion (20-30sec)
- PSDs show that the data is affected in 1-2 Hz. Frequency band. But since BR235 is a medium-period station, we are interested in 0.03-0.1 Hz.
Band for Ms magnitude calculation. There is a very small effect in this frequency band.
- Yearly noise analysis in Figure 7 indicates that in general BR235 is quiter than the other sites within the array.
References Bendat, J.S. and A.G. Piersol (1971). Random data: analysis and measurement procedures. John Wiley and Sons, New York, 407p. McNamara, D.E. and R.P. Buland, Ambient Noise Levels in the Continental United States, Bull. Seism. Soc. Am., 94, 4, 1517-1527, 2004. Peterson, J., Observation and modeling of seismic background noise, U.S. Geol. Surv. Tech. Rept., 93-322, 1-95, 1993.
Each raw frequency distribution bin is then normalized by the total number of PSDs to construct a
Probability Density Function (PDF). The probability of occurrence of a given power at a
particular period is plotted for direct comparison to the Peterson high and low noise models
(NHNM, NLNM). (McNamara and Buland, 2004)
Power Spectrum Density
Hour-long, continuous, and over-lapping (50%) time series data are processed. There is no
removal of earthquakes, system transients and/or data glitches. The instrument transfer function is
removed from each segment, yielding ground acceleration for easy comparison to the
NLNM/NHNM by Peterson. Additional data preparation includes;
- Removing the mean;
- Removing the long period trend;
- Tapering using a 10% sine function; and
- Transforming using an FFT algorithm (Bendat and Piersol, 1971).