Estimation of gravity tilt response to atmospheric phenomena at the Fredericton tiltmetric station using a least squares response method

Description

Variations in apparent direction of gravity, with respect to the tiliting bedrock, at the Fredericton tilitmetric station have been observed, although somewhat intermittently, since 1974. Gaps in these observations account for over half of their overall time span. These tilt observations contain, among others, variations due to atmospheric pressure and surface temperature induced tilts. This work is concerned with modelling the response of the observed tilt, both for purposes of noise reduction and for understanding the phenomena themselves, to the atmospheric pressure and surface temperature variations, which are also observed at this station. As a preliminary evaluation of the collected tilt data, tidal analyses were performed for the purpose of estimating M[subscript 2] diminishing factors and testing ocean loading models which were supplied by Beaumont [1980]. This analysis also provided a means for testing the general performance of the Fredericton station and revealed a backlash effect of the recording apparatus on the recorded tilt. Being confronted with the problem of estimating the response of an observed phenomenon to other observed phenomena, for the case of gappy and noise data, directed this research estimation method capable of handling such data. An evaluation of the existing techniques of cross spectral analysis and time domain convolution (see Appendix III) shows that these techniques are inadequate for our purposes. This least squares response method, developed in Chapter 4, has its basis in least squares spectral analysis [Vanicek, 1971] which is reviewed also in Chapter 4. A statistical test of significance of peaks in the least squares spectrum was also derived during this work. Application of the least squares response method to the Fredericton station data is discussed and summarized in Chapter 5. The results of this analysis indicates that more study is required in interpreting the response estimates in physical terms. However, apparently valid results for tilt response to surface temperature and atmospheric pressure are generated by the least squares response method from the very noisy data of the Fredericton station. In particular it is shown that atmospheric pressure induced tilts have magnitudes in excess of tidal tilts. Recommendations regarding future work at the Fredericton station are made.