ON THE METROLOGICAL SUPPORT OF THE LONG-PERIOD SEISMOLOGY K.V. Kislov, V.V. Gravirov Institute of Earthquake Prediction Theory and Mathematical Geophysics, Russian Academy of Science (IEPT RAS), Profsoyuznaya str. 84/32 Moscow 117997 Russia, e-mail: kvkislov@yandex.ru Abstract. Modern seismological analyses require well-calibrated instruments. What is the quality of the seismic station? What is the data precision? Procedures obtaining the answers to these questions are half-finished. Seismic stations and their long-period sensors are not measuring instruments. Standard procedures of in-use testing aren't applicable to them. Nor their accuracy cannot be determined. Some Russian stations are part of the Global Seismographic Network (GSN) and International Monitoring System (IMS). These stations should use appropriate methods of verification. These methods presuppose three basic approaches. There are two sets of long-period sensors installed at the same station and the data are compared. They produce original calibration. They bring the reference instrument, which is considered as good, to the station and set it next to compare data. They compare how earthquakes records of other stations correspond to records of this station. These techniques have already been partially applied. The first two approaches appear to be too expensive and time consuming for the Russian stations, which are not in the GSN and the IMS. In our opinion, in-depth development of the third approach, bringing it to a routine maintenance process stations, i.e. the creation of user-friendly programs and algorithms, is the most promising way out of this situation. INTRODUCTION The records of ground motion, measured by broadband seismometers, serve for a wide range of scientific research activities. Accordingly, the knowledge of key parameters of seismometers is important for providing reliable results [1]. It is known [2] that from the standpoint of Federal Agency on Technical Regulation and Metrology (Rosstandart) seismic station can't be considered as a means of measurement. It is device for the qualitative analysis of seismic processes. Metrological problems are shifted to shoulders of users. If you place confidence in data, use them, if the answer is no... You have no other data. Application of various methods of calibration and validation of metrological characteristics of seismometers breaks traceability of measurements [3]. Some Russian stations are part of the Global Seismographic Network (GSN) and International Monitoring System (IMS). These stations should use appropriate methods of verification [4]. And, above all, globality of the studied processes demands the uniformity of measurements worldwide. Let's consider how the metrological procedures are carried out in world practice. Global Seismographic Network (GSN) Test of seismic GSN stations regarding compliance to the metrological norms is made as follows: Validation of orientation and polarities of the sensors. Calibration using coils of the seismometer. Comparison of installed instrument signals to known reference sensor. Installation of the second (auxiliary) constantly operating sensor. Noise analysis. Data analysis. Let us examine these procedures in more detail. 1. Orientation and polarities of the sensors. 183
The uncertainty of true North may be the single greatest source of error in orientation metadata. The procedure to orient a sensor involves the following three steps: a). Determining True North b). Translating True North to a fiducial line c). Orienting sensor to fiducial line For cases in which a sensor of unknown orientation is down a borehole or in an underground vault with no accurate fiducial line available, a second broadband sensor may be emplaced at the surface. Data from both sensors are then analyzed to determine their relative orientation. It was decided to check the orientation and polarity of the pendulum at all stations GSN. Although modern methods allow to orient pendulum with an accuracy of 20 ', it was decided not to reinstall the sensors when they deviate from the north to 3. At the same time the sensors will be considered orthogonal if their relative orientation is within 1.5 degrees of 90. 2. Calibration using coils of the seismometer. A calibration is the analyses of instrument response to signals injected through the calibration coils of the seismometer [5]. The calibration shall be conducted at least once annually at all stations. There are methods to calibrate the instrument by the tides [6, 7]. For broadband seismographs at quiet sites, the tides of the solid Earth are a reliable and predictable test signal (fig.1). They have a predominant period of slightly less than 12 hours and an amplitude in the order of 10-6 m/s2 [12]. [8]. They may be extracted by low-pass filtration with a corner frequency of 1 mhz. By comparison with the predicted tides, the gain and polarity of the seismograph may be checked. However so useful method isn't mentioned in longterm plans of GSN [10]. Fig.1. Recording of tides and natural oscillations of the Earth by seismometer CMG-3TB with the built-in digitizer CMG-DM24 [9]. The relative transfer function of the channel can be calculated on a first approximation on base of impulsive excitation, and, based on amplitudes of the tidal waves it is possible to estimate absolute transfer function of system roughly. 3. Comparison of installed instrument signals to known reference sensor (singular intercomparison). It is assumed network operators will measure the absolute calibration and orientation of deployed sensors by installing a temporary sensor whose absolute orientation and sensitivity is known, and comparing that seismometerʼs output with that from the permanent GSN seismometers. The reference seismometer must be installed (to within 20 ) side by side with the tested one. The temperature mode in the seismic vault has to be restored. It will take the time needed for noise at sites to stabilize after deployment of the sensor. Finally, parallel records of two sensors have to contain multiple earthquakes with a magnitude more, than 6.5. The temporary instrument will then be returned to the network operations center and its response verified after each use [11]. 184
Naturally such verification is ineffective and can be done at the stations only every few years. 4. Installation of the second (auxiliary) constantly operating sensor with similar parameters (mutual control). This gives you the opportunity to make absolute comparisons between different models of seismic sensors. This also provides a method of identifying quality variations between two or more of the same model sensor [12, 13]. It is necessary to calculate the coherence of the deconvolved vertical, N S, and E W components of main and auxiliary sensors. The coherence is calculated for ~ 2-hour-long time windows containing the signals for earthquakes with M W 6.5. For each pair of seismograms, the coherence is calculated in narrow frequency bands around 32 s, 64 s, 128 s and 256 s. If the coherence is greater than 0.95, the value is stored together with the complex scaling factor (represented here as a scaling factor and phase shift) that should be applied to the secondary-sensor data to bring the two time series into the best agreement. In the following, the discussion is based on the assumption that the secondary sensor is properly calibrated and that deviations from a scaling factor of 1.0 and a phase shift of 0 should be attributed to differences between the true and reported response functions of the primary sensor. Duplication of the long-period instruments at all stations is a very expensive method. 5. Noise analysis A next method for investigating the overall performance of the sensors is to monitor background noise levels for all seismic channels, after conversion of the data to ground acceleration. For this purpose it is necessary to calculate hourly rms values of the time-domain seismic signal in narrow frequency bands, and convert the rms values to a power spectral density (PSD) at that frequency using Parseval s theorem. For each month, we will calculate the low-noise value at each frequency by determining the PSD amplitude not exceeded 10% of the time. The PSD data provide a wealth of information about the station and the sensors (fig.2.). Results from this noise analysis are useful for characterizing the performance of existing broadband stations, for detecting operational problems, and for learning about sources of seismic noise within a data set. Fig.2. Monthly PSD of ground acceleration at 72-s period for all long-period (LH) channels at station KIP for the period 1988 2009. Smaller symbols are used for months with fewer available hourly measurements. Each component and sensor is represented by a distinct symbol and color. The red horizontal line indicates the level of Peterson s (1993) Low Noise Model (LNM) at 72 s. 185
The thin vertical lines show the times of epoch boundaries in the station metadata [14]. 6. Data analysis. This method for assessing the quality of the data is the systematic comparison of recorded long-period waveforms with synthetic seismograms calculated for known seismic events [15]. Seismic data for the long-period and very-long-period channels of station are collected. Corresponding synthetic waveforms for all earthquakes in the Global CMT catalog (Global Centroid Moment Tensor database) with MW 6.5 are calculated. Correlation coefficients and optimal scaling factors between observed and synthetic waveforms are calculated for the three types of data used in the standard CMT analysis: body waves, with periods in the range 50 150 sec, mantle waves, with periods in the range 125 350 sec, and surface waves, with periods in the range 50 150 sec. When confronted with the seismograms for an individual earthquake, it is often difficult to assess whether a poor fit is due to incorrect source parameters, inadequate modeling of wave propagation through an Earth model, or some problem with the recorded seismograms. Therefore the scaling factor is only calculated for waveforms with a correlation of 0.75 or greater. The scaling factor is the number by which the synthetic seismogram should be multiplied to maximize the agreement with the observed seismogram. Annual averages of the scaling factors are calculated when four or more individual event scaling estimates are available for the year. From the above procedures only calibration using coils of the seismometer is done under any schedules basis. Management of effective metrological support Amplitude variations of 10% and smaller are interpreted as signals in modern studies that seek to map the attenuating properties of the Earth s interior. Phase anomalies of a few seconds at long periods are similarly interpreted in terms of Earth s elastic structure by numerous authors. Therefore it is important to ensure station instrumentation is functioning as expected and that the response of the system is well characterized by the metadata. When there is confidence that a calibration has provided an accurate estimate of the instrument response at a station, the appropriate Data Collection Center then should update the station s metadata. Sustaining our fleet of equipment and infrastructure with its spectrum of ages requires a continuous engagement in monitoring the state of the network. It is clear that the most important and effective metrology procedure is data analysis. This method doesn't answer a question of a measurement error of the ground motion. However, it is the best approximation of the transfer functions of the channels for seismic events. This is the best of what scientists were able to come up. We should add the remote calibration of instruments (including calibration using the tides), the noise analysis and the validation of orientation and polarities of the sensors using Web access. The exact geographical (latitude and longitude) for each site is also being confirmed with GPS techniques for comparison with early map based determinations. All procedures should be determined for long term Web access. The method could and should be developed. The data analysis has to take place in the center of information processing, by means of completely automated procedure. The synthetic seismograms for all stations must be prepared after each earthquake with a big magnitude (more than 6.5). Results of comparing with real data should be embodied in metadata. Thus the metadata must be available to users on an equal basis (together) with the seismic data. So, in the processing centers for each station must be taken measures: The remote calibration of instruments (including calibration using the tides) annually. Noise analysis (deviations from long-term noise characteristics of data streams) annually for each month. Data analysis (comparison of recorded long-period waveforms with synthetic seismograms) after each earthquake with a big magnitude (more than 6.5). Validation of orientation and polarities of the sensors based on comparison between observed and synthetic seismograms (the azimuth residuals show polarization anomalies). Documentation of the current waveform quality problems. Review and correct current metadata for all stations, as necessary. 186
Finding sources of waveform quality problems and development timetable for correction. This will require additional resources and personnel, to fully implement these methods, to develop standards, procedures, programs for routine data processing. If a station produces anomalous calibration results, or appears to be providing rapidly changing response information, we will calibrate stations more frequently until a resolution of the problem can be found. Web-published metrics not only offer a clear status and history of sensor data quality for the scientist using the data, but also better enable network operators to monitor quality, to bring engineering expertise to problems identified, and for making decisions on the allocation of resources for field trips. This metrological scheme in the long-period seismology will allow to have a constant confidence in the reliability of seismic data. CONCLUSION There is an urgent need to solve the problem of metrological assurance in long-period seismometry, because the quality of seismic measurements has a decisive influence on a number of the most important indicators in various branches: the economic efficiency of prospecting for useful minerals, the confidence in the operation of the system for predicting earthquakes etc. In order to verify that seismic instruments meet the above demands and other user requirements it is important from a testing standpoint, that one be able to measure the self-noise of seismic sensors. The lack of systematic calibrations, and inspection of calibration results, makes it difficult to identify instrument problems. Metrological assurance of Russian long-period seismology should be brought into line with international standards. Assurance of measurements' uniformity and accuracy is necessary owing to globality of studied processes. However procedures which are applied for GSN stations are far from perfect (sites not visited often enough for good calibration, sensors operating out of specification, metadata not updated). It is necessary to adopt the principles and, at the same time reduce the cost and simplify the procedure, and to make them more efficient by using the automatic data analysis. This will require additional staff and special seismological metrological service. It is necessary to develop an industry-specific standards, programs, and procedures. The capability of remote automatic calibration via the telemetry link is essential for routine monitoring of sensor state-of-health, documenting the sensor response, and tracking data quality. Along with other methods of inspection stations, it'll allow to stay abreast of the quality of data for each station and of the seismic network as a whole. The special attention should be paid to metadata, i.e. their availability and possible errors descriptions of station parameters: orientation of the sensors, response functions, polarities. Regardless of the cause, it is necessary to document and publicize the lack of accurate and reliable station characteristics, especially when it is not obvious from simple inspection of the data that a problem exists. Information about the state of GSN data quality metrics must be on a par with the data themselves. This information must be as routinely, easily available to the user community as it is for the network operators themselves. Confidence in data requires knowledge of their quality. REFERENCES 1. Krivtsov E.P., Sinel'nikov A.E., Yankovskii A.A. (1993), Supporting unified seismometry measurements. Measurement Techniques, 36(8), 895-898. 2. Zakharchenko N. Z. (2007), Scientific and legal aspects of seismometry metrology, in: Proc. Ist regional technical conf. Problems of Integrated Geophysical Monitoring of the Russian Far East (Petropavlovsk-Kamchatsky, 11-17 November 2007), GS RAS, 122-126. 3. Sargsyan V.K., Sargsyan R.E. (2009), On some metrological problems in seismometry, Proceedings of NAS RA & ESUA - Series of Technical Sciences,. Т. LXII, N3, 330-336. 4. Mishatkin V. N., Zakharchenko N. Z. (2009), The problem of the seismic stations certification, in: Proc. IInd regional technical conf. Problems of Integrated Geophysical Monitoring of the Russian Far East (Petropavlovsk-Kamchatsky, 11-17 October 2009), GS RAS, 278-282. 5. Wielandt E. (2003), Seismometry, in IASPEI International Handbook of Earthquake and Engineering Seismology edited by W. H. K. Lee, H. Kanamori and P. C. Jennings, 44pp. 187
6. Osipov K.S. (1992), Adaptive analysis of non-stationary time series in the study of seismic waves in the period range 0.5-5 hours. Extended Abstract of a dissertation for the degree of Candidate in Physics and mathematics, S-t Petersburg. 7. Davis, P and Berger, J. (2007), Calibration of the Global Seismographic Network Using Tides. Seismological Research Letters; 78(4), 454-459. 8. Wielandt E. (2012), Seismic sensors and their calibration. Chapter 5, in: New Manual of Seismological Observatory Practice (NMSOP-2), IASPEI, ed. By Bormann, P., GFZ German Research Centre for Geosciences, Potsdam, 46 pp. 9. http://www.vulcan-seismicsystems.com/images_guralp/l_guralp_cmg-3t_portable_rus.pdf 10. GSN Calibration Policy (2010), http://www.iris.edu/hq/files/programs/gsn/gsnqual/gsn_calibration_policy_rev1.pdf 11. Tasič I., Runovc F. (2014), The development and analysis of 3D transformation matrices for two seismometers. Journal of Seismology, 18(3), 575-586. 12. Ringler A.T. and Hutt C.R. (2010), Self-Noise Models of Seismic Instruments. Seismological Research Letters, 81(6), 972-983. 13. Hutt R.C., Evans R.J., Followill F., Nigbor L.R., Wielandt E. (2009), Guidelines for standardized testing of broadband seismometers and accelerometers. USGS Open-File Report 2009 1295, U. S. Geological Survey, 66 pp. 14. Ekström G., Nettles M. (2010), Performance of the GSN station KIP-IU, 1988-2009. A report in a series documenting the status of the Global Seismographic Network, 15pp. 15. Pavlov V. (2002), A convenient technique for calculating synthetic seismograms in a layered half-space, in: Proc. 4 th International Conference «Problems of Geocosmos» (St. Petersburg, 3-8 June 2002), Р.320-323. 188