Intercomparison of Relative and Absolute Gravimeters Olivier FRANCIS
The right gravimeter for the right job (signal to be detected) Absolute/relative Stability (accuracy)/precision (repetability) Spatial (portability) Is the signal periodic or a trend? Choice of the gravimeter: relative instrument Absolute instrument Absolute/relative instrument
Instrument specification from the manufacturer Absolute gravimeters Meter Accuracy Precision Repeatabili ty microgal microggal/ microgal Sqrt(Hz) A10 10 100 10 FG5 2 10 1
Instrument specifications Scintrex Reading resolution: 1 microgal Standard deviation less than 5 microgal LaCoste-Romberg 0.1 microgal resolution Superconducting gravimeter???
Signal characteristics Periodical signal (tides) Trend Network (static) Intercomparisons But first calibration of the relative gravity meters
Calibration methods for relative gravimeters - Baselines: BIPM, Hanover, Canada, Luxembourg and Algeria (mobile instruments) - Oscillating platforms (perturbations, high-frequency calibration) - Moving masses (heavy cylindrical ring, ) - Adjustment on a tidal amplitude (it takes time ) - Direct comparison with absolute measurements -
Calibration of Superconducting gravimeters
Precision on the calibration factor
Tidal factors using 47 days
Spring gravimeters calibration Time domain : linear regression between SG and LCR data Frequency domain : tidal factors
Calibration of a spring gravimeters Superconducting gravimeter LCR 906 Membach station
Time domain Calibration Factor (nm.s**-2/volt) -8020-8030 -8040-8050 -8060 +0.1% -0.1% 0 1 2 3 4 5 6 7 8 9 10 11 12 weeks Each dot represents the LCR906 calibration factor calculated using 2 weeks of data. The data sets overlap by one week. The dotted lines are the values of the calibration factor at ±0.1% of the calibration factor calculated using three months of data (solid line)
Frequency domain Amplitude and phase of the tidal waves resulting from the analysis of 85.5 days of simultaneously recorded SG-C021 and LCR906 observations in Membach. R.M.S. (LCR) = 2 or 3 times R.M.S. (SG)
Phase calibration Van Camp M., Vauterin P., Wenzel H.-G, Schott P. and Francis O., Accurate transfer function determination for superconducting gravimeters, Geophysical Research Letters, VOL.27, NO.1, 37-40, 2000. Precision 0.01 second
Intrumental Drift
The Membach Station 3 km 6 km Reservoirs capacities: Max. 24 10 6 m³ Min. 10 10 6 m³
The Membach Gallery, the SG-C021 and AG Gravimeters SG GWR-C021 continuously measuring since August, 1995 AG FG5-202 frequently measuring since January, 1996 +/- 47 m Natural Slope 2.4 m Galery 140 m 76 m Vesdre River
Gravity residuals from the superconducting gravimeter and the absolute gravimeter 360 SG-C021 360 1 year AG-FG5 320 96 sets 2 000 to 20 000 drops ~ 1 to 8 days 320 FG5#202-9810467000 280 240 280 240 GWR#C021 relative 200 200 40 nm/s² or 4 µgal 160 Slow trend: - 0.6 ± 0.1 µgal / yr 160 + 3.0 ± 0.5 mm/yr (free air + Bouguer) 1-Jan-96 31-Dec-96 1-Jan-98 1-Jan-99 2-Jan-00 1-Jan-01 2-Jan-02 2-Jan-03
Power Spectrum Density The PSD AG time series is shown when the microseismic noise is very high (red), normal (blue) and quiet (purple: sampling rate = 1/5 s; green: 1/10 s). The dark green represents the spectrum from the 96 AG values. The SG spectrum is divided into 4 parts: black: using the hourly values from Jan 1996 to June 2004 after removing the SG instrumental drift by fitting a first order polynomial on the difference between the SG and the AG gravity data; light blue: idem but without removing the drift; grey: using 10 s data during 31 days (January 2004); orange: using 1 s data during 24 h in April 1999. The environmental noise dominates SG observations for frequencies lower than 10-3 Hz and frequencies lower than 1-2.5 10-5 Hz (1-0.5 day period) for the AG measurements. The peaks due to the microseismic noise appear clearly, in spite of the low-pass filter [Van Camp et al., 2000]. The peak at 8 10-3 Hz (122 s period) is due to a free oscillation of the levitating sphere, typical of the SGs. The peaks at 1.1, 2.3, 3.4, 4.6 and 5.8 10-5 Hz (1, 2, 3, 4 and 5 cpd)
What did we learn? Relative gravimeters need to be calibrated (0.1% in amplitude, 0.01 sec. in phase) Relative gravimeters drift SG should be checked against an AG LCR: interpretation the trend could be risky Tidal factors (SG, Scintrex, LCR, FG5)
Comparsion between SG, Scintrex CG3M and CG5
FFT Residuals CG3M/CG5
Intercomparison of Absolute gravimeters Postdam (1909-1971) Paris (Sèvres, BIPM, 1980) Walferdange (2003)
Walferdange: November 3-73 7 2003 15 meters from 13 countries 5 types of absolute gravimeters First time : - simultaneous observations - Estimate of the error due to the operators
Instruments A-10#008 FG5 fiber interferometer IMGC#02 FG5 bulk interferometer Jilag#6
Operators error
Instruments Offset
Instruments Offset Sigma = 1.6 microgal
Comparison with previous intercomparisons
ICAG-2005 : relative gravimeters
Error Budget Main Error Source in dg rms ugal Measures to reduce the errors in dg rms ugal Temperature ~ 3 Air-condition < +- 0.5 deg. < 1 Displacement ~ 5 Short-symmetric distances < 1.5 Tides ~ 1 Short-symmetric-Time-intervals <<.5 Ground noises ~ 2 Filtering and multi-readings < 1.5 Reading/round off ~ 2 Repeated reading-occupation < 1 Zero-drift ~ 5 Short-equal-intervals, multiclosures, Polynomial approach < 2 Scale ~ 2 Quasi-zero gravity ties <<.5 Sensor H; h/v gradient ~ 2 Height fixed tripods/reduction <.5 Typing error ~! Fixed-traceable schedules ~ 0 Others ~ 2 - - - ~ 2 All above ~ 9 N-meters, M-measures Average ~ 4/ NM
Measurements Montage
Measurements
Raw data analysis - Uncertainty of a dg - direct Statistics without outliers or adjustment a measured dg=r A -R A1 : Uncertainty of a dg : σ dg A1 Uncertainty of mean value of dg N gravimeters, M measures each: dg Tie by the error budget estimation: σ dg ~ 4 ugal by the statistics of the triangle closures by the statistics of the deviations vs the mean value of dg by comparing to the ICAG01 and FG5-108 absolute g U dg A = σ dg MN
Raw data : Triangle Closure A1 A dg2 dg1 dg3 B2 B6 B3 B B1 B5 B4 A2 Triangle closures are designed A non-0 closure is a true error - theoretical case: =dg1+dg2-dg3 = 0 - Real case: dg1+dg2-dg3 = 0 Uncertainty for a tie of a meter : σ dg ~
Raw data : Triangle Closure S008 - -1 st order- A A1 A2-2.6 B B1 B2 0.1 B B2 B6-2.8 B B6 B3 1.0 B B3 B4 0.7 B B4 B5-0.7 B B5 B1 0.3-2 nd order- C1 C2 B 1.3 C1 C2 A 1.2 C1 B A -1.0 C2 B A -0.9 σ( )= 1.4 ugal 8 6 4 2 0-2 -4-6 N = 66 Mean = 0.06 Min = -6.6 ; Max = 5.7 σ = 2.2 ugal 1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 Uncertainty by closure statistics : σ dg( ) ~ σ( ) = 2.2 < 4 ugal
Raw Data: Deviations + = a dg Mean of all the dg of a tie S008 σ =1.5 ugal N = 157 Uncertainty by deviation statistics σ dg(+) ~ σ (+) = 2.4 < 4 ugal S245 σ =2.2 ugal S424 σ =2.6 ugal 3 Sci+1 Lcr N = 591 σ =2.4 ugal
Raw Data: Compared to Absolute g Preliminary adjustment of 4 gravimeters (solution 15 Aug.*) fixed B at 90 cm with ICAG01 A-B scale extended to C1-B-C2 1st order 2nd order Point dg Relative 15 Aug.* ICAG01 Dec. 2001 FG5-108 30 Aug 5 Sept* Dif. Abs-Rel A 25 701.2 25 701.4 0.2 A2 25 706.6 25 707.8 1.2 B 28 018.8 28 018.8 0.0 B3 28 001.7 28 001.2-0.5 C1-B -4737.5* -4737.7 0.2 C2-B 4020.7 * 4020.4-0.3 C2-C1 8758.2 * 8758.1-0.1 σ = 0.5 4 σ ( Abs Rel) = 0.5 < = 2ugal 4 (N = 4 gravimeters) * Relative result was calculated and published on internet before the absolute determinations (draft result)
Raw Data analysis: Summary by error budget estimation : ~ 4 ugal by closures : σ dg ( ) ~ σ( ) = 2.2 < 4 ugal by deviations : σ dg (+) ~ σ (+) = 2.4 < 4 ugal by comparing to absolute g : σ(abs-rel) = 0.5 < 4/2 ugal Uncertainty of mean value of N gravimeters, M measures each and in the worst case (1st order tie or gradient) : U dg σ dg 4 ~ < < < NM NM 4 N ugal (N > 9, M > 3)
«It s a capital mistake to theorize without data» Sherlock Holmes
Algérie - 2000