- Email: [email protected]
Analysis of Superconducting Gravimeter measurements at MunGyung station, Korea
Journal of Geodynamics 47 (2009) 180–190
Contents lists available at ScienceDirect
Journal of Geodynamics journal homepage: http://www.elsevier.com/locate/jog
Analysis of Superconducting Gravimeter measurements at MunGyung station, Korea Jeong Woo Kim a,b,∗ , Juergen Neumeyer b,c , Tae Hee Kim b,d , Ik Woo e , Hyuck-Jin Park b , Jeong-Soo Jeon f , Ki-Dong Kim b a
Department of Geomatics Engineering, University of Calgary, Calgary, AB, Canada Department of Geoinformation Engineering, Sejong University, Seoul, Republic of Korea Geoscientist, Potsdam , Germany d Department of Polar Science, Sokendai, Tokyo, Japan e College of Ocean Science and Technology, Kunsan National University, Kunsan, Republic of Korea f Geology and Geoinformation Division, Korea Institute of Geoscience and Mineral Resources, Daejon, Republic of Korea b c
a r t i c l e
i n f o
Article history: Received 29 January 2008 Received in revised form 29 July 2008 Accepted 31 July 2008 Keywords: Superconducting Gravimeter SG calibration Noise magnitude Tidal analysis Environmental gravity effects Co-seismic gravity change
a b s t r a c t In 2005, Korea’s first Superconducting Gravimeter (SG) was installed at MunGyung (MG). For the installation of the SG, a new measuring hut with a grounded pier was built. The SG has been successfully recording since April 2005. To monitor the groundwater level changes around the SG, a borehole was drilled in September 2005. For analyzing the gravity data, the SG calibration factor of −64.548 gal/V (gal = 10−8 m/s2 ) was determined by a parallel registration with an absolute gravimeter. The noise magnitude (NM) at the MG site was determined and compared with other SG stations and the New Low Noise Model. The MG site can be classified within the NM of other SG stations. Additionally, noise caused by misaligned instrumental tilt has been investigated and spectrally analyzed. Covering the instrument with a polystyrene box has resulted in a decrease of this noise. After preprocessing, the tidal analysis has been performed and compared with the Wahr–Dehant tidal model parameters. Non-tidal gravity variations, caused by atmospheric pressure and ground water level variations, were calculated. According to the FES2002 ocean tide model, the ocean loading at MG site was calculated. All known gravity effects (i.e. solid Earth tides, polar motion, atmospheric pressure and ground water level induced gravity variations, and the ocean loading gravity effect) were then subtracted from the SG measured gravity data. The remaining gravity variations (gravity residuals) were spectrally analyzed for detection of unknown gravity effects. Co-seismic gravity changes were detected by the SG for earthquakes with epicenters within a distance of about 100 km from the MG station. The largest detected co-seismic gravity change was 0.06 gal. © 2008 Elsevier Ltd. All rights reserved.
1. Introduction Korea’s first Superconducting Gravimeter (SG) was installed at MunGyung (MG) Observatory in 2005, where geodetic and geophysical instruments are combined for the Korea National Laboratory Project “Optimal Data Fusion of Geophysical and Geodetic Measurements for Geological Hazards Monitoring and Prediction” supported by the Ministry of Science and Technology of Korea (Kim et al., 2005, 2007). The remote controlled SG (GWR Serial No. 045) of the new generation is equipped with a Helium gas liquefying
∗ Corresponding author at: Department of Geomatics Engineering, University of Calgary, Calgary, AB, Canada. Tel.: +1 403 220 4858; fax: +1 403 284 1980. E-mail address: [email protected] (J.W. Kim). 0264-3707/$ – see front matter © 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.jog.2008.07.008
compressor. Therefore, it has no Helium boil off compared to the older instruments. To install the SG a new measuring hut with a grounded pier was built. The SG has been successfully recording gravity variations since April 2005, with a sampling rate of 1 s. Precise atmospheric ground pressure data are being collected simultaneously. To monitor the groundwater level changes, a borehole was drilled in September 2005. The distance between the borehole and the SG is 3 m. This study is mostly directed at the tidal analysis and co-seismic gravity changes in connection with atmosphere, hydrosphere, and ocean loading induced gravity variations. Because of primarily instrumental tilt compensation problems the influence of misaligned tilt on gravity data is checked. Gravity and atmospheric pressure data from the observation period April 2005 to January 2007 and the groundwater level data since September 24, 2006 were analyzed. Before analyzing, the SG was calibrated
J.W. Kim et al. / Journal of Geodynamics 47 (2009) 180–190
181
Fig. 1. Location (left) and overview of MunGyung (MG) superconducting gravity observatory (center) inside KIGAM’s seismological station (right).
by comparison with an absolute gravimeter, and the noise magnitude was determined which expresses the quality of the site and the recorded data.
3. SG sensor calibration and noise study
2. SG site and observatory descriptions
The state-of-the-art in the calibration of SGs is the derivation of the calibration coefficient from a parallel registration of SG and absolute gravimeter (e.g. FG5) or a well calibrated spring gravimeter (e.g. LaCoste & Romberg or Scintrex). The calibration coefficient can be estimated with an accuracy of about ±1 nm s−2 /V, which corresponds to about ±0.1–0.2% nm s−2 /V by parallel registration with an absolute gravimeter (Hinderer et al., 1991, 1998; Francis and van Dam, 2002). With the inertial calibration platform (BKG Frankfurt, Germany), an accuracy of ±0.2 nm s−2 /V can be achieved (Falk et al., 2001). A theoretical approach for determination of the calibration (scaling) coefficient is based on theoretical tides as reference. The determination of the scaling coefficient is performed by regression analysis between the raw gravity data and theoretical tides, based on Hartmann–Wenzel tidal catalogue HW95 with and accuracy of 1 ngal (Hartmann and Wenzel, 1995) and the Wahr–Dehant Earth tide model (Wenzel, 1996). Atmospheric pressure and ocean loading induced gravity variations, according to Sections 4.2 and 4.3, have been added to the theoretical Earth tides, because the raw gravity data also include these signals. The scaling coefficient was determined to be CFTh = −64.56 gal/V. Neumeyer et al. (2002) showed that this method of scaling factor determination is in good agreement with a calibration based on absolute gravimeter recordings.
MunGyung is located nearly at the center of S. Korea (Fig. 1). Its longitude and latitude are 128.2147◦ E and 36.6402◦ N, respectively; and, its altitude is 107.5 m above sea level. Administratively, it is located at 222-1 Gusan-li, Hogye-myon, MunGyung, GyungSang Province, Korea. Since the bedrock of the MG SG Observatory is limestone, hydrologic effects are vigorous and they are considered accordingly. The MG SG Observatory was established at one of the seismological stations built and operated by the Korea Institute of Geoscience & Mineral Resources (KIGAM). Inside this newly built government facility, a GPS (Global Positioning System) ground station is also being operated with borehole seismic accelerometer (Fig. 2). It was necessary to dig into the ground until we reached the basement rock, which was not exposed; and, we drilled a 5 m ironcored concrete pillar into the basement rock (Fig. 2 upper). On the top of the pillar, we constructed a 1 m × 2 m × 1 × m concrete platform padded by polystyrene, which prevents vibration transmission from the observatory building. The MG SG observatory was registered as the 21st observatory of the Global Geodynamics Project (GGP) (Crossley et al., 1999), and it formed an East Asian cluster along with other 6 SGs (3 in Japan, 1 in China, 2 in Taiwan) (Fig. 3).
3.1. Sensor calibration
Fig. 2. Construction of MunGyung Superconducting Gravimeter pillar and platform (upper). Gravimeter Dewar and Data Acquisition System, as well as borehole accelerometer and GPS, are shown (lower).
182
J.W. Kim et al. / Journal of Geodynamics 47 (2009) 180–190
Fig. 3. Asian superconducting gravity observatories for the Global Geodynamics Project (GGP).
In October 2007, a parallel registration with the absolute gravimeter FG5 from Industrial Technology Research Institute (ITRI) at Hsinchu, Taiwan was carried out from October 8 at 11:50:49 UTC to October 10 at 00:44:15 UTC. In a linear regression analysis between the FG5 drop recordings (15 s drop period) and the raw SG data (15 s sampling period), a calibration coefficient of CFFG5 = −64.548 ± 0.224 gal/V was determined. Outliers larger than ±3 had been removed from both data sets previously. The difference between the two coefficients, CFTh and CFFG5 , is 0.012 gal/V, which lies within the error bar of the determined calibration coefficient. For further calculations, CFFG5 has been used. 3.2. Noise characteristics The investigation of weak gravity effects absolutely requires a low noise level at the site. The signal-to-noise ratio (S/N) of the recorded gravity data depends on the instrumental noise as well. The instrumental noise is small in the inspected frequency band compared to the noise at the site. To estimate the noise level at an SG site, the noise magnitude (NM) is often used (Banka and Crossley, 1999). For the MG SG, the NM was calculated from seismi-
cally quite gravity data in the frequency band from 0.1 to 100 mHz by: m/s2 NM = 10 · log(PSD)(related to √ in dB) Hz
(1)
The power spectral density (PSD) was calculated from 1 s intervals of raw gravity data, corrected for air pressure induced gravity for a length of 7 days between January 1 and 7, 2007. These are data after solving the tilt problem of the SG, described in Section 3.3. The PSD was applied with a subdivisions length of the raw gravity date of 12 h and an overlapping of 50%, and the data were overlaid with a Hanning window. The lost of energy by using the Hanning window is compensated by multiplication of the Fourier compo8/3 (Bendat and nents in the PSD procedure with a factor of Piersol, 1986). In Fig. 4, PSD (left axis) and the allocated NM (right axis) are shown as functions of frequency in mHz. Additionally, the NM, according to the New Low Noise Model (NLNM) (Peterson, 1993), is also shown. For comparison, the NM was calculated for the Sutherland (SU) site in South Africa from raw gravity air pressure corrected data collected between May 15 and 31, 2005, from the lower gravity sensor (data from the GGP database http://ggp.gfzpotsdam.de). The NM at the MG station was less than −170 dB
Fig. 4. Power spectral density (PSD) and noise magnitude (NM) at MunGyung (MG) and South Africa’s Sutherland (SU) sites in comparison with the New Low Noise Model (NLNM).
J.W. Kim et al. / Journal of Geodynamics 47 (2009) 180–190
183
Fig. 5. Geometry of misalignment of SG measurements relative to the Earth’s mass center.
within the frequency range between 1 and 10 mHz, and close to −160 dB between 0.1 and 1 mHz. Within these frequency ranges, the data from the SU site is closer to the NLNM. Above 10 mHz, the NM at MG is lower than SU. This frequency range is normally filtered out because the inserting range of the free oscillation of the Earth starts below 10 mHz. The noise peak in the range from 30 to 100 mHz is mainly caused by the marine micro seismic (Widmer-Schnidrig, 2003). A comparison of the noise magnitude with other SG stations (Rosat et al., 2004) shows the NM at MG is in the same order of magnitude as the other analyzed SG stations of the GGP network shown in Fig. 3. 3.3. Instrumental tilt noise The test mass (1 inch Nb sphere) of the SG sensor is suspended by the magnetic force generated by superconducting coils. The SG only measures the vertical component of gravity gv exactly, if the gravity sensor axis is aligned to the vertical direction of the Earths mass center. As shown in Fig. 5, if there is a misalignment with an angle between the two axes, the measured gravity, gm , becomes: gm = gv − gT
(2)
Fig. 6. Tilt Y power and balance recording.
#1). In addition, the SG was covered with a polystyrene box on September 21, 2006 (Event #2) to maintain a more stable ambient temperature and protect the ThL from blowing of the air conditioning system. Fig. 6 shows the recording of tilt power and balance signals before and after Events #1 and #2. There was a small decrease of tilt power variation after Event #1 and a decrease from about ±50 to 20 mV after Event #2. The tilt balance signal came up to about ±25 mV before Event #1. After Event #1, the balance signal was reduced to about ±10 mV. A further decrease of the balance signal could be achieved by box installation (Event #2). The medium balance signal was reduced to ±5 mV. Tilt Y tilt power and balance have behaviour which is very similar to tilt X. The frequency dependent decrease of the tilt balance signal is shown in Fig. 7. The strong peaks between 0.017 and 0.1 h (1 and 6 min) and at 12 h and 24 h recorded with the old thermo-mechanical levellers (old ThL) disappeared after Event #1 (new Thl) installation of the new ones (new ThL). An additional decrease can be seen after box installation (Box Inst) at 12 h and 24 h for tilt X and tilt Y and for tilt Y for the whole inspected period range of 0.017–60 h.
Since gm = gv cos , the tilt induced gravity signal, gT , becomes: gT = gv (1 − cos )∼gv ·
2 2
(3)
The automatic tilt compensation system is designed to control the tilt angle, , to be minimal. It consists of the two vertical sensitive pendulum tilt sensors, X and Y, which are orthogonally arranged within the gravity sensing unit of the SG. If gT shall be smaller than 10−12 g, the tilt angle must be ≤ 1.4 rad (Goodkind, 1999). Tilt power and balance signals are proportional to , and they are recorded with the gravity signal. In the tilt setup procedure, the tilt is aligned to a minimum by micrometer screws. During the SG operation, the automatic tilt compensation system sets the tilt angle to a minimum by thermo-mechanical levellers (ThL). If the tilt compensation system does not work properly, the misaligned tilt causes noise in the gravity data (Iwano and Fukuda, 2004; Imanishi, 2005). The thermo-mechanical levellers are also sensitive to the changes in ambient temperature. Therefore, the SG sensor room temperature should be stable within the range of about ±1.5 ◦ C and the air conditioning system should not blow to the thermo-mechanical levellers. Since the automatic tilt system of the MG SG was found not to work properly for months, it was new adjusted and the blowing of the air conditioning system was reduced on June 14, 2006 (Event
Fig. 7. Amplitude spectrum of tilt X and tilt Y balance recordings.
184
J.W. Kim et al. / Journal of Geodynamics 47 (2009) 180–190
nance. From the raw gravity data, spikes larger than 0.2 gal were removed. Offsets larger than 1 gal were also removed if they did not originate from either atmosphere or groundwater level induced gravity variations. These preprocessed gravity (ıgpre ) and atmospheric pressure (ap) data were filtered and decimated to a 5 min sampling rate for further analysis using the program DECIMATE from ETERNA package 3.3. 4.2. Air pressure induced gravity The admittance function for atmospheric pressure has been determined by regression analysis between preprocessed and Earth tide (ET) reduced gravity data and the atmospheric ground pressure data (ap) measured at MG site. Fig. 8. Power spectral density of gravity residuals after Event #1 and Event #2.
The influence of the misaligned tilt on gravity recordings is shown in Fig. 8. It displays the PSD of the gravity residuals (ıgres ) (see Section 5) 7 days before and after Event #1 and Event #2. The PSD was calculated with a subdivisions length of the gravity residuals of 7 h, an overlapping of 50%, and the data were overlaid with a Hanning window. There are peaks at the same periods (between 4 and 7, 8, and 9, and around 12 min), as in the spectrum of tilt balance. These peaks became smaller or disappeared after Event #1. After Event #2, only a remarkable reduction of the peaks between 4 and 5 min was achieved.
ıgcor = ıgpre − ET
This method yields a satisfactory reduction of the atmospheric pressure effect. Physical approaches with 3-D atmospheric pressure data models essentially improve the reduction in long periodic tidal band (Neumeyer et al., 2004, 2007; Boy and Hinderer, 2006). Results at SG stations in Europe show that the seasonal surface pressure independent gravity effect is about 2 gal, which should be considered in future. Unfortunately, no data were available for calculation of the 3-D atmospheric pressure correction. The single admittance coefficient for the ap (ap adm) has been determined to be −0.32 gal/h Pa. The atmospheric pressure induced gravity, as shown in Fig. 9, was calculated by: ıgair = ap · ap adm
4. Data reductions The raw gravity data contain instrumental and other perturbations as well as atmosphere and hydrosphere including the ocean induced gravity variations. For a precise tidal analysis, frequency analysis of the gravity residuals, and the investigation of co-seismic gravity changes these perturbations and gravity variations must be removed from the data.
(4)
(5)
The maximal gravity change (ıgair ) yielded 14 gal, which was derived from the maximal air pressure change (ıap) of 44 h Pa. The maximal annual gravity changes during the observation period yielded 6.5 gal, derived from the maximal annual air pressure variation of 20 h Pa. Fig. 9 also shows the strong seasonal atmospheric pressure induced variations in this area. 4.3. Ocean tide loading induced gravity
4.1. Preprocessing In the preprocessing, spikes and offsets due to instrumental and other perturbations were carefully removed from the raw SG recordings (ıgraw ) using the programs Tsoft (Tsoft, 2002), DETIDE and DESPIKE from the ETERNA 3.3 Earth tide processing package (Wenzel, 1996). This includes offsets in data due to SG mainte-
Based on the global ocean tide model FES2002 (Lefevre et al., 2002; Le Provost et al., 2002), the ocean loading for the main tidal waves in semidiurnal, diurnal and long periodic tidal bands have been calculated, based on Francis and Mazzega (1990). The gravity variations induced by ocean tide loading (ıgol ) were calculated in time domain. Their maximal amplitude during the
Fig. 9. Atmospheric pressure induced gravity effect (ıgair ) and annual wave of ıgair .
J.W. Kim et al. / Journal of Geodynamics 47 (2009) 180–190
185
Fig. 10. Amplitudes and phases of ocean loading at MunGyung (MG), Wuhan (WU) and Sutherland (SU) sites.
observation period was about 4 gal. Additionally, the ocean loading vectors (amplitudes and phases) were determined for the tidal constituents. Fig. 10 shows the amplitudes and phases for the main diurnal and semidiurnal tidal waves for the MG site in comparison with the Sutherland SG station in South Africa ( = 32.38◦ S, = 20.81◦ E) and the Wuhan (WU) SG site in China ( = 30.515◦ N, = 114.49E). At the MG station, the largest amplitudes caused the partial waves (O1 = 1.04 gal, K1 = 1.22 gal and M2 = 1.02 gal). The MG site is more affected by the ocean tides than WU because MG is located closer to the ocean (ocean distance about 100 km) and therefore more affected by the regional ocean loading effect, which can be quite different as shown for the SU site (distance to the ocean about 200 km) with larger amplitudes in the semidiurnal band.
4.4. Groundwater level variation induced gravity Water circulation in the surroundings of the SG causes variation in the gravitational attraction and Earth’s surface deformation similar to those from the atmosphere. Precipitation causes changes in soil moisture and groundwater level. In several cases, a good correlation between gravity and groundwater level variations were shown (Kroner, 2001; Virtanen, 2001; Harnisch and Harnisch, 2002;). The gravity effect of these variations is determined by regression analysis. The regression coefficient varies between about 1 and 10 gal/m, depending on the hydrological conditions. This is a simple model, which does not reflect the real hydrological gravity signal very accurately. Both soil moisture and groundwater level data reflect local effects, as well as signals on regional or continental scale. Additionally, topography and local hydrological structure have a big impact on hydrological loading (e.g. Boy et al., 2005; Meurers et al., 2005; Kroner and Jahr, 2006). All these facts constrict a clear separation of hydrological gravity contributions from different scales.
For determination of the hydrological gravity effect we calculated the admittance coefficient between the gravity data ıgpre
ET ap
= ıgpre − ET − ıgair
(6)
and the groundwater level variation (ıh gwl) (Fig. 11) measured in the borehole at the MG SG site by regression analysis. The admittance coefficient (adm gwl) was determined to be 0.29 gal/m. The hydrological gravity effect based on groundwater level induced gravity (ıggwl ) was calculated by: ıggwl = ıh gwl · adm gwl.
(7)
During the observation period from September 2005 to January 2007, the maximal groundwater level variation was 2.73 m. This corresponds to the maximal groundwater level induced gravity variation of 0.7 gal only. It is possible the borehole is situated within a compressed aquifer that is not representative of the ground water level changes around the SG. More boreholes at representative places are necessary to model the ground water level influence more precisely. Because there is a very small correlation between ground water level variation and gravity, we assumed the determined groundwater level induced gravity variation does not reflect the real behaviour of the groundwater level influence. Therefore, the Bouguer plate modelling has been applied according to (Torge, 1989): ıggwl = 2 · · G · w · Ps · ıh gwl
(8)
with a gravitational constant G = 6.673 × 10−11 m3 kg−1 s−2 , water density w = 103 kg/m3 , water filled pore space Ps = 0.05, and ıh gwl = 2.73 m. When we assumed a water filled pore space of 5%, the maximal gravity change yielded 5.7 gal. From this calculation we can derive an admittance coefficient of 2.1 gal/m. This admittance coefficient has been applied to further data evaluation. This result also gives only a very rough estimation of the effect.
186
J.W. Kim et al. / Journal of Geodynamics 47 (2009) 180–190
Fig. 11. ıh gwl at MG site.
The spectrum of groundwater level variation (Fig. 12) shows peaks at 8, 12, and 24 h, which may be caused by atmospheric pressure changes. But there were also peaks at about 12.43, 25.82 and around 105 h for unknown reasons. For better modelling, a local hydrology model, which considers the local hydrological cycle, should be developed around the SG site. Input data for this model should be precipitation, soil moisture and groundwater level variations measured at representative locations.
The tidal analysis was performed for an interval of 21 months of SG preprocessed gravity (ıgpre ) and atmospheric pressure data. Fig. 13 shows the determined Earth tide parameters ı and (ET). For comparison, the Wahr–Dehant model (WD model) tidal parameters (Dehant, 1987; Dehant et al., 1999) are also pictured. The deviations of the observed from the model parameters can be seen clearly. The major reason for the deviations is the influence of the ocean loading. Thus, the tidal analysis was also performed with ocean tide loading reduced preprocessed gravity (ıgpre ol ):
4.5. Tidal analysis
ıgpre
For the determination of the Earth tide parameters ı (ratio observed to theoretical partial tidal amplitude) and (phase lead), the program ANALYSE from the ETERNA 3.3 package was used. The ANALYSE program is based on the least square method using the Wahr–Dehant tidal model and different tidal potential catalogues (Wenzel, 1996). The most accurate catalogue is the HW95 with a resolution of 1 ngal (Hartmann and Wenzel, 1995). The accuracy of the determined tidal parameters depends on the SG calibration accuracy and the reduction of the non-tidal gravity effects (atmosphere, hydrology and ocean) within the period range of the tidal waves. With a calibration accuracy of ±0.2%, the tidal amplitude factors can be determined as ı = ±0.002.The phase shift of the recorded SG data can be determined with an accuracy around ±0.5 s; thus, the phase lead can then be determined with an accuracy as ±0.001◦ .
The black columns (ET OL) show ı and of ocean tide loading reduced preprocessed gravity. These parameters come much closer to the model, except for the diurnal wave S1. The strong deviation of ı and for S1 from the model may be caused by the influence of daily variations of the atmospheric pressure and daily groundwater level variations (strong peak in the spectrum of ıggwl at 24 h in Fig. 12). The first high precision tidal amplitudes, expressed by the amplitude factors ı and phase leads , have been determined for the MG site and the Korean region with a standard deviation of ±0.07 gal. The remaining small deviations to the model may be caused by insufficient modelling of the gravity variation induced by atmosphere, hydrosphere and ocean. The precise determined tidal parameters can be used for Earth tide reduction of relative and absolute gravity, other precise measurements like satellite
ol
= ıgpre − ıgol
Fig. 12. Amplitude spectrum of ıh gwl at MG site.
(9)
J.W. Kim et al. / Journal of Geodynamics 47 (2009) 180–190
187
Fig. 13. Tidal analysis results at the MG site from preprocessed gravity (ıgpre ) (ET), ocean tide loading reduced preprocessed gravity (ıgpre ol ) (ET OL), in comparison with the WD model.
positioning, GPS-, Laser- and radio-interferometric methods, and further investigations (e.g. verification of global and regional ocean tide models) (Baker and Bos, 2003). 5. Frequency analysis of the gravity residuals The frequency analysis is performed on the gravity residuals by: ıgres = ıgraw − ET − ıgair − ıgol − ıggwl
(10)
Raw gravity data were reduced from all known gravity effects caused by Earth tides, air pressure (∼14 gal), ocean loading (∼4 gal)), ground water level (∼5.7 gal) and polar motion (∼5 gal) for detecting unknown frequencies using the power spectrum. It was applied with a subdivisions length of the gravity residual data of 3 days and an overlapping of 70%, and the data were multiplied with a Hanning window.
The spectrum of ıgres in Fig. 14 shows strong peaks at frequencies 1–7 cpd and 11 cpd. The same peaks are in the power spectral density of the atmospheric pressure. These are the well known peaks in gravity residuals (e.g. Kroner, 1997). They correspond to the harmonics of the solar heated atmospheric wave S1. These waves could not be taken into account by using the single admittance coefficient for calculation of the air pressure induced gravity. Therefore, they remain in the gravity residuals. By using the frequency dependent admittance these waves also cannot be removed completely from the gravity residuals (Crossley et al., 1995; Neumeyer, 1995). For checking frequencies which do not have their origins in the ocean loading the power spectral density of tide gauge measurements (TG) from the Korean station, East Hoopo (36◦ 40 28 N; 129◦ 27 20 E) with a distance to the MG station of 109 km was added. Local tide gauge measurements correlate well with the ocean loading and can improve the ocean loading correction compared to global ocean tide models (Neumeyer et al., 2005). The PSD
Fig. 14. Power spectrum of gravity residuals (ıgres ) and tide gauge measurements (TG).
188
J.W. Kim et al. / Journal of Geodynamics 47 (2009) 180–190
Table 1 Summary of earthquake events within a distance of about 160 km to the MunGyung (MG) superconducting gravity observatory. Earthquake event EQ#1 EQ#2 EQ#3 EQ#4
Date (Yr-M-D) 2007-01-20 2006-04-29 2006-01-19 2005-06-29
Time (UTC) 11:56:53 02:01:12 03:35:34 15:25:04
Magnitude 4.8 3.5 3.2 3.1
of TG shows the diurnal (1 cpd) and semidiurnal (2 cpd) ocean tides which are not completely removed in the gravity residuals ıgres . We also have around 3 and 4 cpd and at larger frequencies common peaks in both spectra caused by the ocean loading. More detailed analyses about this subject were made e.g., by Anderson (1999), Boy et al. (2004), and Khan and Hoyer (2004). Furthermore, there are several peaks in ıgres outside these frequencies, e.g. f1 = 1.3543 cpd, f2 = 5.5743 cpd, f3 = 6.47053 cpd, f4 = 7.38707 cpd, f5 = 8.47461 cpd, f6 = 8.71401 cpd, f7 = 9.89047 cpd and f8 = 11.8262 cpd. These peaks are not in the spectra of the atmospheric pressure, ground water level, and ocean loading induced gravity. The sources for these frequencies are unknown. They must be further investigated. 6. Co-seismic gravity change Gravity changes due to earthquakes have been observed with absolute gravimeters e.g. Tanaka et al. (2001) and Superconducting Gravimeters e.g. Goodkind (1999) and Imanishi et al. (2004). Tanaka et al. (2001) observed gravity changes of 6 gal after an earthquake of magnitude 6.1 with an absolute gravimeter. Goodkind (1999) analyzed SG gravity changes at Fairbanks, Alaska, corresponding to earthquakes within 500 km distance of the SG. He analyzed a decreasing of gravity before the earthquakes (May 16, 1993 magnitude 4.0 depth 106 km distance 126 km) and a rising after the earthquake. These measured gravity changes accounted for some gals, but were in contradiction to the straightforward models of earthquakes (Lambert and Bower, 1991). From these results, the large gravity changes could be overlaid by gravity variations of other sources. Nevertheless, this study gave input to further studies. Imanishi et al. observed co-seismic gravity changes from the Tokachi-oki earthquake (2003-09-25) at 19:50:08 with magnitude of 8 at location = 41.78◦ N, = 144.079◦ E close to the Hokkaido Island at a depth of 32 km. Co-seismic gravity changes have been recorded at the SG stations of Essashi (0.58 gal, distance 3.4◦ ), Matsushiro (0.1 gal, distance 6.9◦ ) and Kyoto (0.07 gal, distance
Latitude (◦ North) 37.68 37.09 37.21 36.66
Longitude (◦ East) ◦
128.59 129.92◦ 129.63◦ 129.63◦
Depth (km)
Distance (km)
13.1 7.8 Unknown 6.4
119.4 159.8 80.7 127
9.4◦ ). These results are in good agreement with the calculated results of dislocation models. Generally, we have to consider that the sphere (test mass) of the SG is partly out of feedback and registration range during strong earthquakes. The measurement range of the SG covers about ±900 gal. The automatic resetting of the sphere into range may also cause an offset. This is the uncertainty in interpreting coseismic gravity changes measured with SGs. On the other hand, the analysis period should not be too large, in order to reduce an interaction with gravity variations of other sources. We analyzed 4 earthquakes within a distance of about 100 km to the MG station (Table 1). During the largest earthquake (EQ#1), the maximal seismic signal was at the end but still in feedback 811 gal. The signals of the other earthquakes were within feedback too. Therefore, we could exclude an instrumental effect caused by out of feedback of the sphere. The recorded raw gravity data of the earthquakes (1 s sampling rate) were reduced from theoretical Earth tides, air pressure induced gravity, ocean loading effect, polar motion and the linear drift. For detection of a co-seismic gravity change, an offset in the data close to the event time of the earthquake, we selected a data set for about 1 h before and 1 h after the earthquake. The data were divided into two blocks. Block1 starts 1 h before the earthquake and ends 1 sample (1 s) before recording of the earthquake (event time t0 ). Block two starts 2 min after the event time and end after 1 h. The first 2 min of the recorded earthquake were not used for fitting because of the large amplitudes. After 2 min the amplitudes were ±2.5 gal in maximum. The data of the two blocks were separately fitted by a quadratic function Eq. (11) to consider the different nonlinear trends before and after the earthquake. y(t) = a + b · t + c · t 2
(11)
Then we calculated the gravity of the fitted functions at the event time t0 for block1 y1(t0 ) and y2(t0 ) for block2. The co-seismic
Fig. 15. Earthquake recorded at the MG station (black) and the fitted functions before and after the earthquake (white).
J.W. Kim et al. / Journal of Geodynamics 47 (2009) 180–190
gravity change ∂gco-seis was determined by ∂gco-seis = y2(t0 ) − y1(t0 )
(12)
The calculated gravity change is ∂gco-seis = 0.06 gal. Fig. 15 shows the recorded earthquake and the fitted curves before an after the earthquake. Because other effects can cause a gravity change, we analyzed the atmospheric pressure and the groundwater level data for a step at and around the event time of the earthquakes. We could not find a step in these data. The gravity changes due to the other earthquakes resulted in EQ#2 = 0.025 gal, EQ#3 = 0.025 gal and EQ#4 = −0.01 gal. When we consider the accuracy of the SG in time domain of about ±0.01 gal, the detected gravity changes of EQ#4 can not be stated as relevant and EQ#2 and EQ#3 are close to the SG accuracy. But after EQ#1, there is a significant gravity change. Of course, for confirmation of the observation results, modelling with a dislocation model (e.g., Sun and Okubo, 1998) also has to be carried out. A pre-seismic signal could not be detected for these earthquakes. 7. Outlook The new SG in Korea makes the worldwide SG network in the Asian region denser and it contributes to the research of local, regional and global gravity variations. The present results encourage for more detailed investigations in order to improve the present and discover new ones. For these goals also data from the Asian SG sub-network should be used. Further investigations should be concentrated on: • Analysis of more recorded earthquakes for co-seismic gravity changes and comparison with dislocation models. • Studying of the hydrology induced gravity variations in combination with the Asian SG sub-network and hydrological models. • Improving the ocean loading correction in combination with tide gauge measurements. • Looking for special frequencies in the power spectrum of the gravity residuals to detect gravity effect linked to the Earth’s interior in combination with the Asian sub-network. Acknowledgments This study was supported by National Research Lab project (M10302-00-0063) of the Ministry of Science and Technology of Korea. We like to thank Prof. Cheinway Hwang and his research group from the Industrial Technology Research Institute (ITRI) at Hsinchu, Taiwan, for carrying out the absolute measurements at the MG station. References Anderson, O.B., 1999. Shallow water tides in the northwest European self region from TOPEX/POSEIDON altimetry. J. Geophys. Res. 104 (C4), 7729–7774. Baker, T.F., Bos, M.S., 2003. Validating Earth and ocean tide models using tidal gravity measurements. Geophys. J. Int. 152, 468–485. Banka, D., Crossley, D., 1999. Noise levels of superconducting gravimeters at seismic frequencies. Geophys. J. Int. 139, 87–97. Bendat, J.S., Piersol, A.G., 1986. Random Data—Analysis and Measurement Procedures. A Wiley-Interscience Publication JOHN WILEY & SONS, New York, Chichester, Brisbane, Toronto, Singapore. Boy, J.-P., Llubles, M., Ray, R., Hinderer, J., Florsch, N., Rosat, S., Lyard, F., Letelier, T., 2004. Non-linear oceanic tides observed by superconducting gravimeters in Europe. J. Geodyn. 38, 391–405. Boy, J.P., Hinderer, J., Ferhat, G., 2005. Gravity changes and crustal deformation due to hydrology loading. Geophys. Res. Abstr. 7, 07166. Boy, J.P., Hinderer, J., 2006. Study of the seasonal gravity signal in superconducting gravity data. J. Geodyn. 41, 227–233. Crossley, D., Hinderer, J., Casula, O., Francis, O., Hsu, H.T., Imanishi, Y., Jentzsch, G., Kääriäinen, J., Merriam, J., Meurers, B., Neumeyer, J., Richter, B., Sato, D., Shihuya, K., van Dam, T., 1999. Network of superconducting gravimeters benefits a number of disciplines. EOS. Trans. Am. Geophys. Union 80 (11), 125–126.
189
Crossley, D.J., Jensen, O.G., Hinderer, J., 1995. Effective barometric admittance and gravity residuals. Phys. Earth Planet. Int. 90, 221–241. Dehant, V., 1987. Tidal parameters for an inelastic Earth. Phys. Earth Planet. Int. 49, 97–116. Dehant, V., Defraigne, P., Wahr, J., 1999. Tides for a convective Earth. J. Geophys. Res. 104B1, 1035–1058. Falk, R., Harnisch, M., Harnisch, G., Novak, I., Richter, B., 2001. Calibration of the Superconducting Gravimeter SG 103 C023, CD029 and CD030. J. Geodetic Soc. Japan 47/1, 22–27. Francis, O., Mazzega, P., 1990. Global charts of ocean tide loading effects. J. Geophys. Res. 95, 11411–11424. Francis, O., van Dam, T., 2002. Evaluation of the precision of using absolute gravimeters to calibrate superconducting gravimeters. Metrologia 39, 485–488. Goodkind, J.M., 1999. The superconducting gravimeter. Rev. Sci. Inst. 70/11, 4131– 4152. Harnisch, G., Harnisch, M., 2002. Seasonal variations of hydrological influences on gravity measurements at Wettzell. Bull. D’Inf. Marees Terr. 137, 10849–10861. Hartmann, T., Wenzel, H.G., 1995. Catalogue HW95 of the tide generating potential. Marees Terr. Bull. d’Inform. 123, 9278–9301. Hinderer, J., Florsch, N., Mäkinen, J., Legros, H., Faller, J.E., 1991. On the calibration of a superconducting gravimeter using absolute gravity measurements. Geophys. J. Int. 106 (2), 491–497. Hinderer, J., Amalvict, M., Franzis, O., Mäkinen, J., 1998. On the calibration of superconducting gravimeters with the help of absolute gravity measurements. In: Ducarme, B., Paquet, P. (Eds.), Proc. 13th Int. Symp. Earth Tides. Brussels, 1997, pp. 557–564. Imanishi, Y., Sato, T., Higashi, T., Sun, W., Okubo, S., 2004. A network of superconducting gravimeters detects submicrogal coseismic gravity changes. Science 306, 476–478, doi:10.1126/science1101875. Imanishi, Y., 2005. On the possible cause of long period instrumental noise (parasitic mode) of a superconducting gravimeter. J. Geod. 78, 683–690. Iwano, Fukuda, 2004. Superconducting observations without a tilt compensation system. Phys. Earth Planet. Int. 147, 343–351. Khan, S.A., Hoyer, J., 2004. Shallow-water tides in Japan from superconducting gravimetry. J. Geod. 78, 150–245. Kim, T.H., Neumeyer, J., Woo, I., Park, H.J., Kim, J.W., 2007. Installation and data analysis of superconducting gravimeter in MunGyung, Korea; preliminary results. Korea Econ. Environ. Geol. 40 (4), 445–459. Kim, J.W., et al., 2005. Geohazard monitoring with space and geophysical technology—an introduction to the KJRS. Kor. J. Remote Sensing 21 (1), 3–13 (Special Issue). Kroner, C., 1997. Reduktion von Luftdruckeffekten in zeitabhängigen Schwerebeobachtungen, Dissertation. Technische Universität Clausthal. Kroner, C., 2001. Hydrological effects on gravity data of the Geodynamic Observatory Moxa. J. Geodyn. Soc. Jpn. 47 (1), 353–358. Kroner, C., Jahr, T., 2006. Hydrological experiments around the superconducting gravimeter at Moxa observatory. J. Geodyn. 41, 268–275. Lambert, A., Bower, D., 1991. Constraints on the usefulness of gravimetry for detecting precursory crustal deformations. Tectonophysics 193 (4), 369–375. Le Provost, C., Lyard, F., Lefevre, F., Roblou, L., 2002. FES 2002—a new version of the FES tidal solution series. In: Abstract Volume Jason-1 Science Working Team Meeting, vol. 528, Biarritz, France. Lefevre, F., Lyard, F.H., Le Provost, C., Schrama, E.J.O., 2002. FES99: a global tide finite element solution assimilating tide gauge and altimetric information. J. Atmos. Oceanic Technol. 19, 1345–1356. Meurers, B., Van Camp, M., Petermans, T., Verbeeck, K., Vanneste, K., 2005. Investigation of local atmospheric and hydrological gravity signals in Superconducting Gravimeter time series. Geophys. Res. Abstr. 7, 07463. Neumeyer, J., 1995. Frequency dependent atmospheric pressure correction on gravity variations by means of cross spectral analysis. Marees Terrestres Bulletin d’Informationes, Bruxelles, 122, 9212–9220. Neumeyer, J., Barthelmes, F., Combrinck, L., Dierks, O., Fourie, P., 2002. Analysis results from the SG registration with the dual sphere superconducting gravimeter at SAGOS (South Africa). Marees Terrestres Bulletin d’Informations, Bruxelles, 135, 10607–10616. Neumeyer, J., Hagedoorn, J., Leitloff, J., Schmidt, T., 2004. Gravity reduction with three-dimensional atmospheric pressure data for precise ground gravity measurements. J. Geodyn. 38, 437–450. Neumeyer, J., del Pino, J., Dierks, O., Sun, H.-P., Pflug, H., 2005. Improvement of ocean loading correction on gravity data with additional tide gauge measurements. J. Geodyn. 40 (1), 104–111. Neumeyer, J., Schmidt, T., Stoeber, C., 2007. Improved determination of the atmospheric attraction with 3D air density data and its reduction on ground gravity measurements. In: Int. Association of Geodesy Symposia, vol. 130, Dynamic Planet, Cairns Australia. Springer Berlin Heidelberg, New York, pp. 541– 548. Peterson, J., 1993. Observation and modelling of background seismic noise. Open File Report 93-332. U.S. Department of Interior, Geological Survey, Albuquerque, New Mexico. Rosat, S., Hinderer, J., Crossley, D., Boy, J.P., 2004. Performance of superconducting gravimeters from long-period seismology to tides. J. Geodyn. 38, 461–476. Sun, W., Okubo, S., 1998. Surface potential and gravity changes due to internal dislocations in a spherical earth—II. Application to a finite fault. Geophys. J. Int. 132 (1), 79–99.
190
J.W. Kim et al. / Journal of Geodynamics 47 (2009) 180–190
Tanaka, Y., Okubo, S., Machida, M., Kimura, I., Kosuge, T., 2001. First detection of absolute gravity change caused by earthquake. Geophys. Res. Lett. 28 (15), 2979–2981. Torge, W., 1989. Gravimetry. de Gruyter, Berlin, New York. Tsoft, 2002. http://www.astro.oma.be/SEISMO/TSOFT/tsoft.html. Virtanen, H., 2001. Hydrological studies at the gravity station Metshovi Finland. J. Geodetic. Soc. Jpn. 47 (1), 328–333.
Wenzel, H.-G., 1996. The nanogal software: earthtide data processing package ETERNA 3.3. Marees Terrestres Bulletin d’Informations, Bruxelles, 124, 9425–9439. Widmer-Schnidrig, R., 2003. What can Superconducting Gravimeters contribute to normal mode seismology? Bull. Seism. Soc. Am. 93 (3), 1370–1380.
Contents lists available at ScienceDirect
Journal of Geodynamics journal homepage: http://www.elsevier.com/locate/jog
Analysis of Superconducting Gravimeter measurements at MunGyung station, Korea Jeong Woo Kim a,b,∗ , Juergen Neumeyer b,c , Tae Hee Kim b,d , Ik Woo e , Hyuck-Jin Park b , Jeong-Soo Jeon f , Ki-Dong Kim b a
Department of Geomatics Engineering, University of Calgary, Calgary, AB, Canada Department of Geoinformation Engineering, Sejong University, Seoul, Republic of Korea Geoscientist, Potsdam , Germany d Department of Polar Science, Sokendai, Tokyo, Japan e College of Ocean Science and Technology, Kunsan National University, Kunsan, Republic of Korea f Geology and Geoinformation Division, Korea Institute of Geoscience and Mineral Resources, Daejon, Republic of Korea b c
a r t i c l e
i n f o
Article history: Received 29 January 2008 Received in revised form 29 July 2008 Accepted 31 July 2008 Keywords: Superconducting Gravimeter SG calibration Noise magnitude Tidal analysis Environmental gravity effects Co-seismic gravity change
a b s t r a c t In 2005, Korea’s first Superconducting Gravimeter (SG) was installed at MunGyung (MG). For the installation of the SG, a new measuring hut with a grounded pier was built. The SG has been successfully recording since April 2005. To monitor the groundwater level changes around the SG, a borehole was drilled in September 2005. For analyzing the gravity data, the SG calibration factor of −64.548 gal/V (gal = 10−8 m/s2 ) was determined by a parallel registration with an absolute gravimeter. The noise magnitude (NM) at the MG site was determined and compared with other SG stations and the New Low Noise Model. The MG site can be classified within the NM of other SG stations. Additionally, noise caused by misaligned instrumental tilt has been investigated and spectrally analyzed. Covering the instrument with a polystyrene box has resulted in a decrease of this noise. After preprocessing, the tidal analysis has been performed and compared with the Wahr–Dehant tidal model parameters. Non-tidal gravity variations, caused by atmospheric pressure and ground water level variations, were calculated. According to the FES2002 ocean tide model, the ocean loading at MG site was calculated. All known gravity effects (i.e. solid Earth tides, polar motion, atmospheric pressure and ground water level induced gravity variations, and the ocean loading gravity effect) were then subtracted from the SG measured gravity data. The remaining gravity variations (gravity residuals) were spectrally analyzed for detection of unknown gravity effects. Co-seismic gravity changes were detected by the SG for earthquakes with epicenters within a distance of about 100 km from the MG station. The largest detected co-seismic gravity change was 0.06 gal. © 2008 Elsevier Ltd. All rights reserved.
1. Introduction Korea’s first Superconducting Gravimeter (SG) was installed at MunGyung (MG) Observatory in 2005, where geodetic and geophysical instruments are combined for the Korea National Laboratory Project “Optimal Data Fusion of Geophysical and Geodetic Measurements for Geological Hazards Monitoring and Prediction” supported by the Ministry of Science and Technology of Korea (Kim et al., 2005, 2007). The remote controlled SG (GWR Serial No. 045) of the new generation is equipped with a Helium gas liquefying
∗ Corresponding author at: Department of Geomatics Engineering, University of Calgary, Calgary, AB, Canada. Tel.: +1 403 220 4858; fax: +1 403 284 1980. E-mail address: [email protected] (J.W. Kim). 0264-3707/$ – see front matter © 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.jog.2008.07.008
compressor. Therefore, it has no Helium boil off compared to the older instruments. To install the SG a new measuring hut with a grounded pier was built. The SG has been successfully recording gravity variations since April 2005, with a sampling rate of 1 s. Precise atmospheric ground pressure data are being collected simultaneously. To monitor the groundwater level changes, a borehole was drilled in September 2005. The distance between the borehole and the SG is 3 m. This study is mostly directed at the tidal analysis and co-seismic gravity changes in connection with atmosphere, hydrosphere, and ocean loading induced gravity variations. Because of primarily instrumental tilt compensation problems the influence of misaligned tilt on gravity data is checked. Gravity and atmospheric pressure data from the observation period April 2005 to January 2007 and the groundwater level data since September 24, 2006 were analyzed. Before analyzing, the SG was calibrated
J.W. Kim et al. / Journal of Geodynamics 47 (2009) 180–190
181
Fig. 1. Location (left) and overview of MunGyung (MG) superconducting gravity observatory (center) inside KIGAM’s seismological station (right).
by comparison with an absolute gravimeter, and the noise magnitude was determined which expresses the quality of the site and the recorded data.
3. SG sensor calibration and noise study
2. SG site and observatory descriptions
The state-of-the-art in the calibration of SGs is the derivation of the calibration coefficient from a parallel registration of SG and absolute gravimeter (e.g. FG5) or a well calibrated spring gravimeter (e.g. LaCoste & Romberg or Scintrex). The calibration coefficient can be estimated with an accuracy of about ±1 nm s−2 /V, which corresponds to about ±0.1–0.2% nm s−2 /V by parallel registration with an absolute gravimeter (Hinderer et al., 1991, 1998; Francis and van Dam, 2002). With the inertial calibration platform (BKG Frankfurt, Germany), an accuracy of ±0.2 nm s−2 /V can be achieved (Falk et al., 2001). A theoretical approach for determination of the calibration (scaling) coefficient is based on theoretical tides as reference. The determination of the scaling coefficient is performed by regression analysis between the raw gravity data and theoretical tides, based on Hartmann–Wenzel tidal catalogue HW95 with and accuracy of 1 ngal (Hartmann and Wenzel, 1995) and the Wahr–Dehant Earth tide model (Wenzel, 1996). Atmospheric pressure and ocean loading induced gravity variations, according to Sections 4.2 and 4.3, have been added to the theoretical Earth tides, because the raw gravity data also include these signals. The scaling coefficient was determined to be CFTh = −64.56 gal/V. Neumeyer et al. (2002) showed that this method of scaling factor determination is in good agreement with a calibration based on absolute gravimeter recordings.
MunGyung is located nearly at the center of S. Korea (Fig. 1). Its longitude and latitude are 128.2147◦ E and 36.6402◦ N, respectively; and, its altitude is 107.5 m above sea level. Administratively, it is located at 222-1 Gusan-li, Hogye-myon, MunGyung, GyungSang Province, Korea. Since the bedrock of the MG SG Observatory is limestone, hydrologic effects are vigorous and they are considered accordingly. The MG SG Observatory was established at one of the seismological stations built and operated by the Korea Institute of Geoscience & Mineral Resources (KIGAM). Inside this newly built government facility, a GPS (Global Positioning System) ground station is also being operated with borehole seismic accelerometer (Fig. 2). It was necessary to dig into the ground until we reached the basement rock, which was not exposed; and, we drilled a 5 m ironcored concrete pillar into the basement rock (Fig. 2 upper). On the top of the pillar, we constructed a 1 m × 2 m × 1 × m concrete platform padded by polystyrene, which prevents vibration transmission from the observatory building. The MG SG observatory was registered as the 21st observatory of the Global Geodynamics Project (GGP) (Crossley et al., 1999), and it formed an East Asian cluster along with other 6 SGs (3 in Japan, 1 in China, 2 in Taiwan) (Fig. 3).
3.1. Sensor calibration
Fig. 2. Construction of MunGyung Superconducting Gravimeter pillar and platform (upper). Gravimeter Dewar and Data Acquisition System, as well as borehole accelerometer and GPS, are shown (lower).
182
J.W. Kim et al. / Journal of Geodynamics 47 (2009) 180–190
Fig. 3. Asian superconducting gravity observatories for the Global Geodynamics Project (GGP).
In October 2007, a parallel registration with the absolute gravimeter FG5 from Industrial Technology Research Institute (ITRI) at Hsinchu, Taiwan was carried out from October 8 at 11:50:49 UTC to October 10 at 00:44:15 UTC. In a linear regression analysis between the FG5 drop recordings (15 s drop period) and the raw SG data (15 s sampling period), a calibration coefficient of CFFG5 = −64.548 ± 0.224 gal/V was determined. Outliers larger than ±3 had been removed from both data sets previously. The difference between the two coefficients, CFTh and CFFG5 , is 0.012 gal/V, which lies within the error bar of the determined calibration coefficient. For further calculations, CFFG5 has been used. 3.2. Noise characteristics The investigation of weak gravity effects absolutely requires a low noise level at the site. The signal-to-noise ratio (S/N) of the recorded gravity data depends on the instrumental noise as well. The instrumental noise is small in the inspected frequency band compared to the noise at the site. To estimate the noise level at an SG site, the noise magnitude (NM) is often used (Banka and Crossley, 1999). For the MG SG, the NM was calculated from seismi-
cally quite gravity data in the frequency band from 0.1 to 100 mHz by: m/s2 NM = 10 · log(PSD)(related to √ in dB) Hz
(1)
The power spectral density (PSD) was calculated from 1 s intervals of raw gravity data, corrected for air pressure induced gravity for a length of 7 days between January 1 and 7, 2007. These are data after solving the tilt problem of the SG, described in Section 3.3. The PSD was applied with a subdivisions length of the raw gravity date of 12 h and an overlapping of 50%, and the data were overlaid with a Hanning window. The lost of energy by using the Hanning window is compensated by multiplication of the Fourier compo8/3 (Bendat and nents in the PSD procedure with a factor of Piersol, 1986). In Fig. 4, PSD (left axis) and the allocated NM (right axis) are shown as functions of frequency in mHz. Additionally, the NM, according to the New Low Noise Model (NLNM) (Peterson, 1993), is also shown. For comparison, the NM was calculated for the Sutherland (SU) site in South Africa from raw gravity air pressure corrected data collected between May 15 and 31, 2005, from the lower gravity sensor (data from the GGP database http://ggp.gfzpotsdam.de). The NM at the MG station was less than −170 dB
Fig. 4. Power spectral density (PSD) and noise magnitude (NM) at MunGyung (MG) and South Africa’s Sutherland (SU) sites in comparison with the New Low Noise Model (NLNM).
J.W. Kim et al. / Journal of Geodynamics 47 (2009) 180–190
183
Fig. 5. Geometry of misalignment of SG measurements relative to the Earth’s mass center.
within the frequency range between 1 and 10 mHz, and close to −160 dB between 0.1 and 1 mHz. Within these frequency ranges, the data from the SU site is closer to the NLNM. Above 10 mHz, the NM at MG is lower than SU. This frequency range is normally filtered out because the inserting range of the free oscillation of the Earth starts below 10 mHz. The noise peak in the range from 30 to 100 mHz is mainly caused by the marine micro seismic (Widmer-Schnidrig, 2003). A comparison of the noise magnitude with other SG stations (Rosat et al., 2004) shows the NM at MG is in the same order of magnitude as the other analyzed SG stations of the GGP network shown in Fig. 3. 3.3. Instrumental tilt noise The test mass (1 inch Nb sphere) of the SG sensor is suspended by the magnetic force generated by superconducting coils. The SG only measures the vertical component of gravity gv exactly, if the gravity sensor axis is aligned to the vertical direction of the Earths mass center. As shown in Fig. 5, if there is a misalignment with an angle between the two axes, the measured gravity, gm , becomes: gm = gv − gT
(2)
Fig. 6. Tilt Y power and balance recording.
#1). In addition, the SG was covered with a polystyrene box on September 21, 2006 (Event #2) to maintain a more stable ambient temperature and protect the ThL from blowing of the air conditioning system. Fig. 6 shows the recording of tilt power and balance signals before and after Events #1 and #2. There was a small decrease of tilt power variation after Event #1 and a decrease from about ±50 to 20 mV after Event #2. The tilt balance signal came up to about ±25 mV before Event #1. After Event #1, the balance signal was reduced to about ±10 mV. A further decrease of the balance signal could be achieved by box installation (Event #2). The medium balance signal was reduced to ±5 mV. Tilt Y tilt power and balance have behaviour which is very similar to tilt X. The frequency dependent decrease of the tilt balance signal is shown in Fig. 7. The strong peaks between 0.017 and 0.1 h (1 and 6 min) and at 12 h and 24 h recorded with the old thermo-mechanical levellers (old ThL) disappeared after Event #1 (new Thl) installation of the new ones (new ThL). An additional decrease can be seen after box installation (Box Inst) at 12 h and 24 h for tilt X and tilt Y and for tilt Y for the whole inspected period range of 0.017–60 h.
Since gm = gv cos , the tilt induced gravity signal, gT , becomes: gT = gv (1 − cos )∼gv ·
2 2
(3)
The automatic tilt compensation system is designed to control the tilt angle, , to be minimal. It consists of the two vertical sensitive pendulum tilt sensors, X and Y, which are orthogonally arranged within the gravity sensing unit of the SG. If gT shall be smaller than 10−12 g, the tilt angle must be ≤ 1.4 rad (Goodkind, 1999). Tilt power and balance signals are proportional to , and they are recorded with the gravity signal. In the tilt setup procedure, the tilt is aligned to a minimum by micrometer screws. During the SG operation, the automatic tilt compensation system sets the tilt angle to a minimum by thermo-mechanical levellers (ThL). If the tilt compensation system does not work properly, the misaligned tilt causes noise in the gravity data (Iwano and Fukuda, 2004; Imanishi, 2005). The thermo-mechanical levellers are also sensitive to the changes in ambient temperature. Therefore, the SG sensor room temperature should be stable within the range of about ±1.5 ◦ C and the air conditioning system should not blow to the thermo-mechanical levellers. Since the automatic tilt system of the MG SG was found not to work properly for months, it was new adjusted and the blowing of the air conditioning system was reduced on June 14, 2006 (Event
Fig. 7. Amplitude spectrum of tilt X and tilt Y balance recordings.
184
J.W. Kim et al. / Journal of Geodynamics 47 (2009) 180–190
nance. From the raw gravity data, spikes larger than 0.2 gal were removed. Offsets larger than 1 gal were also removed if they did not originate from either atmosphere or groundwater level induced gravity variations. These preprocessed gravity (ıgpre ) and atmospheric pressure (ap) data were filtered and decimated to a 5 min sampling rate for further analysis using the program DECIMATE from ETERNA package 3.3. 4.2. Air pressure induced gravity The admittance function for atmospheric pressure has been determined by regression analysis between preprocessed and Earth tide (ET) reduced gravity data and the atmospheric ground pressure data (ap) measured at MG site. Fig. 8. Power spectral density of gravity residuals after Event #1 and Event #2.
The influence of the misaligned tilt on gravity recordings is shown in Fig. 8. It displays the PSD of the gravity residuals (ıgres ) (see Section 5) 7 days before and after Event #1 and Event #2. The PSD was calculated with a subdivisions length of the gravity residuals of 7 h, an overlapping of 50%, and the data were overlaid with a Hanning window. There are peaks at the same periods (between 4 and 7, 8, and 9, and around 12 min), as in the spectrum of tilt balance. These peaks became smaller or disappeared after Event #1. After Event #2, only a remarkable reduction of the peaks between 4 and 5 min was achieved.
ıgcor = ıgpre − ET
This method yields a satisfactory reduction of the atmospheric pressure effect. Physical approaches with 3-D atmospheric pressure data models essentially improve the reduction in long periodic tidal band (Neumeyer et al., 2004, 2007; Boy and Hinderer, 2006). Results at SG stations in Europe show that the seasonal surface pressure independent gravity effect is about 2 gal, which should be considered in future. Unfortunately, no data were available for calculation of the 3-D atmospheric pressure correction. The single admittance coefficient for the ap (ap adm) has been determined to be −0.32 gal/h Pa. The atmospheric pressure induced gravity, as shown in Fig. 9, was calculated by: ıgair = ap · ap adm
4. Data reductions The raw gravity data contain instrumental and other perturbations as well as atmosphere and hydrosphere including the ocean induced gravity variations. For a precise tidal analysis, frequency analysis of the gravity residuals, and the investigation of co-seismic gravity changes these perturbations and gravity variations must be removed from the data.
(4)
(5)
The maximal gravity change (ıgair ) yielded 14 gal, which was derived from the maximal air pressure change (ıap) of 44 h Pa. The maximal annual gravity changes during the observation period yielded 6.5 gal, derived from the maximal annual air pressure variation of 20 h Pa. Fig. 9 also shows the strong seasonal atmospheric pressure induced variations in this area. 4.3. Ocean tide loading induced gravity
4.1. Preprocessing In the preprocessing, spikes and offsets due to instrumental and other perturbations were carefully removed from the raw SG recordings (ıgraw ) using the programs Tsoft (Tsoft, 2002), DETIDE and DESPIKE from the ETERNA 3.3 Earth tide processing package (Wenzel, 1996). This includes offsets in data due to SG mainte-
Based on the global ocean tide model FES2002 (Lefevre et al., 2002; Le Provost et al., 2002), the ocean loading for the main tidal waves in semidiurnal, diurnal and long periodic tidal bands have been calculated, based on Francis and Mazzega (1990). The gravity variations induced by ocean tide loading (ıgol ) were calculated in time domain. Their maximal amplitude during the
Fig. 9. Atmospheric pressure induced gravity effect (ıgair ) and annual wave of ıgair .
J.W. Kim et al. / Journal of Geodynamics 47 (2009) 180–190
185
Fig. 10. Amplitudes and phases of ocean loading at MunGyung (MG), Wuhan (WU) and Sutherland (SU) sites.
observation period was about 4 gal. Additionally, the ocean loading vectors (amplitudes and phases) were determined for the tidal constituents. Fig. 10 shows the amplitudes and phases for the main diurnal and semidiurnal tidal waves for the MG site in comparison with the Sutherland SG station in South Africa ( = 32.38◦ S, = 20.81◦ E) and the Wuhan (WU) SG site in China ( = 30.515◦ N, = 114.49E). At the MG station, the largest amplitudes caused the partial waves (O1 = 1.04 gal, K1 = 1.22 gal and M2 = 1.02 gal). The MG site is more affected by the ocean tides than WU because MG is located closer to the ocean (ocean distance about 100 km) and therefore more affected by the regional ocean loading effect, which can be quite different as shown for the SU site (distance to the ocean about 200 km) with larger amplitudes in the semidiurnal band.
4.4. Groundwater level variation induced gravity Water circulation in the surroundings of the SG causes variation in the gravitational attraction and Earth’s surface deformation similar to those from the atmosphere. Precipitation causes changes in soil moisture and groundwater level. In several cases, a good correlation between gravity and groundwater level variations were shown (Kroner, 2001; Virtanen, 2001; Harnisch and Harnisch, 2002;). The gravity effect of these variations is determined by regression analysis. The regression coefficient varies between about 1 and 10 gal/m, depending on the hydrological conditions. This is a simple model, which does not reflect the real hydrological gravity signal very accurately. Both soil moisture and groundwater level data reflect local effects, as well as signals on regional or continental scale. Additionally, topography and local hydrological structure have a big impact on hydrological loading (e.g. Boy et al., 2005; Meurers et al., 2005; Kroner and Jahr, 2006). All these facts constrict a clear separation of hydrological gravity contributions from different scales.
For determination of the hydrological gravity effect we calculated the admittance coefficient between the gravity data ıgpre
ET ap
= ıgpre − ET − ıgair
(6)
and the groundwater level variation (ıh gwl) (Fig. 11) measured in the borehole at the MG SG site by regression analysis. The admittance coefficient (adm gwl) was determined to be 0.29 gal/m. The hydrological gravity effect based on groundwater level induced gravity (ıggwl ) was calculated by: ıggwl = ıh gwl · adm gwl.
(7)
During the observation period from September 2005 to January 2007, the maximal groundwater level variation was 2.73 m. This corresponds to the maximal groundwater level induced gravity variation of 0.7 gal only. It is possible the borehole is situated within a compressed aquifer that is not representative of the ground water level changes around the SG. More boreholes at representative places are necessary to model the ground water level influence more precisely. Because there is a very small correlation between ground water level variation and gravity, we assumed the determined groundwater level induced gravity variation does not reflect the real behaviour of the groundwater level influence. Therefore, the Bouguer plate modelling has been applied according to (Torge, 1989): ıggwl = 2 · · G · w · Ps · ıh gwl
(8)
with a gravitational constant G = 6.673 × 10−11 m3 kg−1 s−2 , water density w = 103 kg/m3 , water filled pore space Ps = 0.05, and ıh gwl = 2.73 m. When we assumed a water filled pore space of 5%, the maximal gravity change yielded 5.7 gal. From this calculation we can derive an admittance coefficient of 2.1 gal/m. This admittance coefficient has been applied to further data evaluation. This result also gives only a very rough estimation of the effect.
186
J.W. Kim et al. / Journal of Geodynamics 47 (2009) 180–190
Fig. 11. ıh gwl at MG site.
The spectrum of groundwater level variation (Fig. 12) shows peaks at 8, 12, and 24 h, which may be caused by atmospheric pressure changes. But there were also peaks at about 12.43, 25.82 and around 105 h for unknown reasons. For better modelling, a local hydrology model, which considers the local hydrological cycle, should be developed around the SG site. Input data for this model should be precipitation, soil moisture and groundwater level variations measured at representative locations.
The tidal analysis was performed for an interval of 21 months of SG preprocessed gravity (ıgpre ) and atmospheric pressure data. Fig. 13 shows the determined Earth tide parameters ı and (ET). For comparison, the Wahr–Dehant model (WD model) tidal parameters (Dehant, 1987; Dehant et al., 1999) are also pictured. The deviations of the observed from the model parameters can be seen clearly. The major reason for the deviations is the influence of the ocean loading. Thus, the tidal analysis was also performed with ocean tide loading reduced preprocessed gravity (ıgpre ol ):
4.5. Tidal analysis
ıgpre
For the determination of the Earth tide parameters ı (ratio observed to theoretical partial tidal amplitude) and (phase lead), the program ANALYSE from the ETERNA 3.3 package was used. The ANALYSE program is based on the least square method using the Wahr–Dehant tidal model and different tidal potential catalogues (Wenzel, 1996). The most accurate catalogue is the HW95 with a resolution of 1 ngal (Hartmann and Wenzel, 1995). The accuracy of the determined tidal parameters depends on the SG calibration accuracy and the reduction of the non-tidal gravity effects (atmosphere, hydrology and ocean) within the period range of the tidal waves. With a calibration accuracy of ±0.2%, the tidal amplitude factors can be determined as ı = ±0.002.The phase shift of the recorded SG data can be determined with an accuracy around ±0.5 s; thus, the phase lead can then be determined with an accuracy as ±0.001◦ .
The black columns (ET OL) show ı and of ocean tide loading reduced preprocessed gravity. These parameters come much closer to the model, except for the diurnal wave S1. The strong deviation of ı and for S1 from the model may be caused by the influence of daily variations of the atmospheric pressure and daily groundwater level variations (strong peak in the spectrum of ıggwl at 24 h in Fig. 12). The first high precision tidal amplitudes, expressed by the amplitude factors ı and phase leads , have been determined for the MG site and the Korean region with a standard deviation of ±0.07 gal. The remaining small deviations to the model may be caused by insufficient modelling of the gravity variation induced by atmosphere, hydrosphere and ocean. The precise determined tidal parameters can be used for Earth tide reduction of relative and absolute gravity, other precise measurements like satellite
ol
= ıgpre − ıgol
Fig. 12. Amplitude spectrum of ıh gwl at MG site.
(9)
J.W. Kim et al. / Journal of Geodynamics 47 (2009) 180–190
187
Fig. 13. Tidal analysis results at the MG site from preprocessed gravity (ıgpre ) (ET), ocean tide loading reduced preprocessed gravity (ıgpre ol ) (ET OL), in comparison with the WD model.
positioning, GPS-, Laser- and radio-interferometric methods, and further investigations (e.g. verification of global and regional ocean tide models) (Baker and Bos, 2003). 5. Frequency analysis of the gravity residuals The frequency analysis is performed on the gravity residuals by: ıgres = ıgraw − ET − ıgair − ıgol − ıggwl
(10)
Raw gravity data were reduced from all known gravity effects caused by Earth tides, air pressure (∼14 gal), ocean loading (∼4 gal)), ground water level (∼5.7 gal) and polar motion (∼5 gal) for detecting unknown frequencies using the power spectrum. It was applied with a subdivisions length of the gravity residual data of 3 days and an overlapping of 70%, and the data were multiplied with a Hanning window.
The spectrum of ıgres in Fig. 14 shows strong peaks at frequencies 1–7 cpd and 11 cpd. The same peaks are in the power spectral density of the atmospheric pressure. These are the well known peaks in gravity residuals (e.g. Kroner, 1997). They correspond to the harmonics of the solar heated atmospheric wave S1. These waves could not be taken into account by using the single admittance coefficient for calculation of the air pressure induced gravity. Therefore, they remain in the gravity residuals. By using the frequency dependent admittance these waves also cannot be removed completely from the gravity residuals (Crossley et al., 1995; Neumeyer, 1995). For checking frequencies which do not have their origins in the ocean loading the power spectral density of tide gauge measurements (TG) from the Korean station, East Hoopo (36◦ 40 28 N; 129◦ 27 20 E) with a distance to the MG station of 109 km was added. Local tide gauge measurements correlate well with the ocean loading and can improve the ocean loading correction compared to global ocean tide models (Neumeyer et al., 2005). The PSD
Fig. 14. Power spectrum of gravity residuals (ıgres ) and tide gauge measurements (TG).
188
J.W. Kim et al. / Journal of Geodynamics 47 (2009) 180–190
Table 1 Summary of earthquake events within a distance of about 160 km to the MunGyung (MG) superconducting gravity observatory. Earthquake event EQ#1 EQ#2 EQ#3 EQ#4
Date (Yr-M-D) 2007-01-20 2006-04-29 2006-01-19 2005-06-29
Time (UTC) 11:56:53 02:01:12 03:35:34 15:25:04
Magnitude 4.8 3.5 3.2 3.1
of TG shows the diurnal (1 cpd) and semidiurnal (2 cpd) ocean tides which are not completely removed in the gravity residuals ıgres . We also have around 3 and 4 cpd and at larger frequencies common peaks in both spectra caused by the ocean loading. More detailed analyses about this subject were made e.g., by Anderson (1999), Boy et al. (2004), and Khan and Hoyer (2004). Furthermore, there are several peaks in ıgres outside these frequencies, e.g. f1 = 1.3543 cpd, f2 = 5.5743 cpd, f3 = 6.47053 cpd, f4 = 7.38707 cpd, f5 = 8.47461 cpd, f6 = 8.71401 cpd, f7 = 9.89047 cpd and f8 = 11.8262 cpd. These peaks are not in the spectra of the atmospheric pressure, ground water level, and ocean loading induced gravity. The sources for these frequencies are unknown. They must be further investigated. 6. Co-seismic gravity change Gravity changes due to earthquakes have been observed with absolute gravimeters e.g. Tanaka et al. (2001) and Superconducting Gravimeters e.g. Goodkind (1999) and Imanishi et al. (2004). Tanaka et al. (2001) observed gravity changes of 6 gal after an earthquake of magnitude 6.1 with an absolute gravimeter. Goodkind (1999) analyzed SG gravity changes at Fairbanks, Alaska, corresponding to earthquakes within 500 km distance of the SG. He analyzed a decreasing of gravity before the earthquakes (May 16, 1993 magnitude 4.0 depth 106 km distance 126 km) and a rising after the earthquake. These measured gravity changes accounted for some gals, but were in contradiction to the straightforward models of earthquakes (Lambert and Bower, 1991). From these results, the large gravity changes could be overlaid by gravity variations of other sources. Nevertheless, this study gave input to further studies. Imanishi et al. observed co-seismic gravity changes from the Tokachi-oki earthquake (2003-09-25) at 19:50:08 with magnitude of 8 at location = 41.78◦ N, = 144.079◦ E close to the Hokkaido Island at a depth of 32 km. Co-seismic gravity changes have been recorded at the SG stations of Essashi (0.58 gal, distance 3.4◦ ), Matsushiro (0.1 gal, distance 6.9◦ ) and Kyoto (0.07 gal, distance
Latitude (◦ North) 37.68 37.09 37.21 36.66
Longitude (◦ East) ◦
128.59 129.92◦ 129.63◦ 129.63◦
Depth (km)
Distance (km)
13.1 7.8 Unknown 6.4
119.4 159.8 80.7 127
9.4◦ ). These results are in good agreement with the calculated results of dislocation models. Generally, we have to consider that the sphere (test mass) of the SG is partly out of feedback and registration range during strong earthquakes. The measurement range of the SG covers about ±900 gal. The automatic resetting of the sphere into range may also cause an offset. This is the uncertainty in interpreting coseismic gravity changes measured with SGs. On the other hand, the analysis period should not be too large, in order to reduce an interaction with gravity variations of other sources. We analyzed 4 earthquakes within a distance of about 100 km to the MG station (Table 1). During the largest earthquake (EQ#1), the maximal seismic signal was at the end but still in feedback 811 gal. The signals of the other earthquakes were within feedback too. Therefore, we could exclude an instrumental effect caused by out of feedback of the sphere. The recorded raw gravity data of the earthquakes (1 s sampling rate) were reduced from theoretical Earth tides, air pressure induced gravity, ocean loading effect, polar motion and the linear drift. For detection of a co-seismic gravity change, an offset in the data close to the event time of the earthquake, we selected a data set for about 1 h before and 1 h after the earthquake. The data were divided into two blocks. Block1 starts 1 h before the earthquake and ends 1 sample (1 s) before recording of the earthquake (event time t0 ). Block two starts 2 min after the event time and end after 1 h. The first 2 min of the recorded earthquake were not used for fitting because of the large amplitudes. After 2 min the amplitudes were ±2.5 gal in maximum. The data of the two blocks were separately fitted by a quadratic function Eq. (11) to consider the different nonlinear trends before and after the earthquake. y(t) = a + b · t + c · t 2
(11)
Then we calculated the gravity of the fitted functions at the event time t0 for block1 y1(t0 ) and y2(t0 ) for block2. The co-seismic
Fig. 15. Earthquake recorded at the MG station (black) and the fitted functions before and after the earthquake (white).
J.W. Kim et al. / Journal of Geodynamics 47 (2009) 180–190
gravity change ∂gco-seis was determined by ∂gco-seis = y2(t0 ) − y1(t0 )
(12)
The calculated gravity change is ∂gco-seis = 0.06 gal. Fig. 15 shows the recorded earthquake and the fitted curves before an after the earthquake. Because other effects can cause a gravity change, we analyzed the atmospheric pressure and the groundwater level data for a step at and around the event time of the earthquakes. We could not find a step in these data. The gravity changes due to the other earthquakes resulted in EQ#2 = 0.025 gal, EQ#3 = 0.025 gal and EQ#4 = −0.01 gal. When we consider the accuracy of the SG in time domain of about ±0.01 gal, the detected gravity changes of EQ#4 can not be stated as relevant and EQ#2 and EQ#3 are close to the SG accuracy. But after EQ#1, there is a significant gravity change. Of course, for confirmation of the observation results, modelling with a dislocation model (e.g., Sun and Okubo, 1998) also has to be carried out. A pre-seismic signal could not be detected for these earthquakes. 7. Outlook The new SG in Korea makes the worldwide SG network in the Asian region denser and it contributes to the research of local, regional and global gravity variations. The present results encourage for more detailed investigations in order to improve the present and discover new ones. For these goals also data from the Asian SG sub-network should be used. Further investigations should be concentrated on: • Analysis of more recorded earthquakes for co-seismic gravity changes and comparison with dislocation models. • Studying of the hydrology induced gravity variations in combination with the Asian SG sub-network and hydrological models. • Improving the ocean loading correction in combination with tide gauge measurements. • Looking for special frequencies in the power spectrum of the gravity residuals to detect gravity effect linked to the Earth’s interior in combination with the Asian sub-network. Acknowledgments This study was supported by National Research Lab project (M10302-00-0063) of the Ministry of Science and Technology of Korea. We like to thank Prof. Cheinway Hwang and his research group from the Industrial Technology Research Institute (ITRI) at Hsinchu, Taiwan, for carrying out the absolute measurements at the MG station. References Anderson, O.B., 1999. Shallow water tides in the northwest European self region from TOPEX/POSEIDON altimetry. J. Geophys. Res. 104 (C4), 7729–7774. Baker, T.F., Bos, M.S., 2003. Validating Earth and ocean tide models using tidal gravity measurements. Geophys. J. Int. 152, 468–485. Banka, D., Crossley, D., 1999. Noise levels of superconducting gravimeters at seismic frequencies. Geophys. J. Int. 139, 87–97. Bendat, J.S., Piersol, A.G., 1986. Random Data—Analysis and Measurement Procedures. A Wiley-Interscience Publication JOHN WILEY & SONS, New York, Chichester, Brisbane, Toronto, Singapore. Boy, J.-P., Llubles, M., Ray, R., Hinderer, J., Florsch, N., Rosat, S., Lyard, F., Letelier, T., 2004. Non-linear oceanic tides observed by superconducting gravimeters in Europe. J. Geodyn. 38, 391–405. Boy, J.P., Hinderer, J., Ferhat, G., 2005. Gravity changes and crustal deformation due to hydrology loading. Geophys. Res. Abstr. 7, 07166. Boy, J.P., Hinderer, J., 2006. Study of the seasonal gravity signal in superconducting gravity data. J. Geodyn. 41, 227–233. Crossley, D., Hinderer, J., Casula, O., Francis, O., Hsu, H.T., Imanishi, Y., Jentzsch, G., Kääriäinen, J., Merriam, J., Meurers, B., Neumeyer, J., Richter, B., Sato, D., Shihuya, K., van Dam, T., 1999. Network of superconducting gravimeters benefits a number of disciplines. EOS. Trans. Am. Geophys. Union 80 (11), 125–126.
189
Crossley, D.J., Jensen, O.G., Hinderer, J., 1995. Effective barometric admittance and gravity residuals. Phys. Earth Planet. Int. 90, 221–241. Dehant, V., 1987. Tidal parameters for an inelastic Earth. Phys. Earth Planet. Int. 49, 97–116. Dehant, V., Defraigne, P., Wahr, J., 1999. Tides for a convective Earth. J. Geophys. Res. 104B1, 1035–1058. Falk, R., Harnisch, M., Harnisch, G., Novak, I., Richter, B., 2001. Calibration of the Superconducting Gravimeter SG 103 C023, CD029 and CD030. J. Geodetic Soc. Japan 47/1, 22–27. Francis, O., Mazzega, P., 1990. Global charts of ocean tide loading effects. J. Geophys. Res. 95, 11411–11424. Francis, O., van Dam, T., 2002. Evaluation of the precision of using absolute gravimeters to calibrate superconducting gravimeters. Metrologia 39, 485–488. Goodkind, J.M., 1999. The superconducting gravimeter. Rev. Sci. Inst. 70/11, 4131– 4152. Harnisch, G., Harnisch, M., 2002. Seasonal variations of hydrological influences on gravity measurements at Wettzell. Bull. D’Inf. Marees Terr. 137, 10849–10861. Hartmann, T., Wenzel, H.G., 1995. Catalogue HW95 of the tide generating potential. Marees Terr. Bull. d’Inform. 123, 9278–9301. Hinderer, J., Florsch, N., Mäkinen, J., Legros, H., Faller, J.E., 1991. On the calibration of a superconducting gravimeter using absolute gravity measurements. Geophys. J. Int. 106 (2), 491–497. Hinderer, J., Amalvict, M., Franzis, O., Mäkinen, J., 1998. On the calibration of superconducting gravimeters with the help of absolute gravity measurements. In: Ducarme, B., Paquet, P. (Eds.), Proc. 13th Int. Symp. Earth Tides. Brussels, 1997, pp. 557–564. Imanishi, Y., Sato, T., Higashi, T., Sun, W., Okubo, S., 2004. A network of superconducting gravimeters detects submicrogal coseismic gravity changes. Science 306, 476–478, doi:10.1126/science1101875. Imanishi, Y., 2005. On the possible cause of long period instrumental noise (parasitic mode) of a superconducting gravimeter. J. Geod. 78, 683–690. Iwano, Fukuda, 2004. Superconducting observations without a tilt compensation system. Phys. Earth Planet. Int. 147, 343–351. Khan, S.A., Hoyer, J., 2004. Shallow-water tides in Japan from superconducting gravimetry. J. Geod. 78, 150–245. Kim, T.H., Neumeyer, J., Woo, I., Park, H.J., Kim, J.W., 2007. Installation and data analysis of superconducting gravimeter in MunGyung, Korea; preliminary results. Korea Econ. Environ. Geol. 40 (4), 445–459. Kim, J.W., et al., 2005. Geohazard monitoring with space and geophysical technology—an introduction to the KJRS. Kor. J. Remote Sensing 21 (1), 3–13 (Special Issue). Kroner, C., 1997. Reduktion von Luftdruckeffekten in zeitabhängigen Schwerebeobachtungen, Dissertation. Technische Universität Clausthal. Kroner, C., 2001. Hydrological effects on gravity data of the Geodynamic Observatory Moxa. J. Geodyn. Soc. Jpn. 47 (1), 353–358. Kroner, C., Jahr, T., 2006. Hydrological experiments around the superconducting gravimeter at Moxa observatory. J. Geodyn. 41, 268–275. Lambert, A., Bower, D., 1991. Constraints on the usefulness of gravimetry for detecting precursory crustal deformations. Tectonophysics 193 (4), 369–375. Le Provost, C., Lyard, F., Lefevre, F., Roblou, L., 2002. FES 2002—a new version of the FES tidal solution series. In: Abstract Volume Jason-1 Science Working Team Meeting, vol. 528, Biarritz, France. Lefevre, F., Lyard, F.H., Le Provost, C., Schrama, E.J.O., 2002. FES99: a global tide finite element solution assimilating tide gauge and altimetric information. J. Atmos. Oceanic Technol. 19, 1345–1356. Meurers, B., Van Camp, M., Petermans, T., Verbeeck, K., Vanneste, K., 2005. Investigation of local atmospheric and hydrological gravity signals in Superconducting Gravimeter time series. Geophys. Res. Abstr. 7, 07463. Neumeyer, J., 1995. Frequency dependent atmospheric pressure correction on gravity variations by means of cross spectral analysis. Marees Terrestres Bulletin d’Informationes, Bruxelles, 122, 9212–9220. Neumeyer, J., Barthelmes, F., Combrinck, L., Dierks, O., Fourie, P., 2002. Analysis results from the SG registration with the dual sphere superconducting gravimeter at SAGOS (South Africa). Marees Terrestres Bulletin d’Informations, Bruxelles, 135, 10607–10616. Neumeyer, J., Hagedoorn, J., Leitloff, J., Schmidt, T., 2004. Gravity reduction with three-dimensional atmospheric pressure data for precise ground gravity measurements. J. Geodyn. 38, 437–450. Neumeyer, J., del Pino, J., Dierks, O., Sun, H.-P., Pflug, H., 2005. Improvement of ocean loading correction on gravity data with additional tide gauge measurements. J. Geodyn. 40 (1), 104–111. Neumeyer, J., Schmidt, T., Stoeber, C., 2007. Improved determination of the atmospheric attraction with 3D air density data and its reduction on ground gravity measurements. In: Int. Association of Geodesy Symposia, vol. 130, Dynamic Planet, Cairns Australia. Springer Berlin Heidelberg, New York, pp. 541– 548. Peterson, J., 1993. Observation and modelling of background seismic noise. Open File Report 93-332. U.S. Department of Interior, Geological Survey, Albuquerque, New Mexico. Rosat, S., Hinderer, J., Crossley, D., Boy, J.P., 2004. Performance of superconducting gravimeters from long-period seismology to tides. J. Geodyn. 38, 461–476. Sun, W., Okubo, S., 1998. Surface potential and gravity changes due to internal dislocations in a spherical earth—II. Application to a finite fault. Geophys. J. Int. 132 (1), 79–99.
190
J.W. Kim et al. / Journal of Geodynamics 47 (2009) 180–190
Tanaka, Y., Okubo, S., Machida, M., Kimura, I., Kosuge, T., 2001. First detection of absolute gravity change caused by earthquake. Geophys. Res. Lett. 28 (15), 2979–2981. Torge, W., 1989. Gravimetry. de Gruyter, Berlin, New York. Tsoft, 2002. http://www.astro.oma.be/SEISMO/TSOFT/tsoft.html. Virtanen, H., 2001. Hydrological studies at the gravity station Metshovi Finland. J. Geodetic. Soc. Jpn. 47 (1), 328–333.
Wenzel, H.-G., 1996. The nanogal software: earthtide data processing package ETERNA 3.3. Marees Terrestres Bulletin d’Informations, Bruxelles, 124, 9425–9439. Widmer-Schnidrig, R., 2003. What can Superconducting Gravimeters contribute to normal mode seismology? Bull. Seism. Soc. Am. 93 (3), 1370–1380.