HR: 1330h
AN: G32A-0743    [PDF]
TI: Earth-model discrimination via gravimetric terrestrial spectroscopy
AU: * Omerbashich, M
EM: omerbasic@unb.ca
AF: Department of Geodesy and Geomatics Engineering, University of New Brunswick, PO Box 4400, Fredericton, NB E3B 5A3 Canada
AU: VanˇŠek, P
EM: vanicek@unb.ca
AF: Department of Geodesy and Geomatics Engineering, University of New Brunswick, PO Box 4400, Fredericton, NB E3B 5A3 Canada
AB: We devised a method for Earth-model discrimination, of potentially general interest to geophysics. Based on Jeffreys's rule of thumb, it is claimed that the last decade's mean-diurnal free oscillation magnitudes (MDOM) of the Earth's gravity field linearly correlate with seismic energies and seismic magnitudes of 381 earthquakes larger than M = 6.3 found in the record. Here the oscillation of the gravity field is caused by the seismic waves generated by these medium-to-large earthquakes. Oscillation magnitudes are taken at the discrete values predicted by the individual model's long eigen-periods; we take all modes between 12 min - 2 hrs. Using the correlation coefficients computed from variance-spectra between the MDOM and M $>$ 6.3 global seismicity, which in some cases reach up to 0.97, we subsequently propose a general method for Earth-model discrimination: if a model yields high correlations of the MDOM with Earth's seismicity on the day of the large earthquake, that model is to be considered successful. Various criteria for model discrimination could then be derived from this observation. Gaps are normally present in the record of temporal gravity variations obtained by a seismically excited gravimeter. We thus start the data processing for our technique by applying a non-equidistant Gaussian filter with 4sigma-width, to the last decade's one second step record collected by the Canadian superconducting gravimeter (SG) that operates at Cantley, Quebec. Thus the original one-second data are transformed to records with steps 8 and 32 sec. The non-equidistancy feature of our filter means that one-second gaps in the record are properly accounted for and gaps larger than the filtering steps set at 8 and 32 sec remain in the series. We then perform the least squares spectral analysis (LSSA) of so filtered a record, without any further preprocessing of the data so as to satisfy the rigour requirement. The LSSA seems suitable for our purpose due to its ability to handle gaps in data, unlike other spectral analyses. We obtain power- and variance-spectra of the record (including noise), and using three different geophysical models we look into the free oscillation of the Earth at the periods predicted by these models; we compute respectively the mean-diurnal and mean-weekly gravity oscillation magnitudes (MWOM). For all three models the correlations are higher when seismic energies are used rather than seismic magnitudes, as well as when variance-spectra are used rather than power-spectra. As deduced from the variance-spectra (but not the power spectra) of gravity, the correlation is maximum for eigen-periods of around 821 sec. When a depth-of-earthquake separation is performed, the best correlation between the MDOM series and the deep (d $>$ 400 km) earthquakes shows a curious delay of three days, for all the models. For this we have no physical explanation. Finally, the series of MDOM values of the Earth's gravity field appears periodic over the studied decade: a synodic semi-monthly and solar semi-annual periods show up as the only two significant periods from the one-day to ten-years period interval. Various researchers in the area of earthquake prediction and "tidal triggering" have in the past pointed their finger at the same periods. We know where these periods come from but we have no physical explanation as to why they should trigger the characteristic oscillations.
UR: http://einstein.gge.unb.ca
DE: 8122 Dynamics, gravity and tectonics
DE: 8123 Dynamics, seismotectonics
DE: 8149 Planetary tectonics (5475)
DE: 8150 Plate boundary--general (3040)
DE: 8194 Instruments and techniques
SC: Geodesy [G]
MN: 2003 Fall Meeting