| Résumé : |
(Auteur) In this work, the scope of the present day gravity data acquisition, processing and interpretational technique in using weak signal gravity anomalies, to study mass distributions, has been tested. To carry out this test, measurements were taken in Darmstadt and Bremen. The study in these areas, that encompasses the different stages of gravity data measurement, analysis and modelling procedures, has led to the development of some new methods and algorithms. The survey in Darmstadt was aimed at recovering the shape, location and density of two Rain Water Reservoirs, from measured gravity data. On the other hand the survey in Bremen was aimed at computing the vertical component of the gravity gradient within the "Drop Tower Bremen".
At first, the implementation of a spectral analysis technique in choosing a sampling interval that avoided aliasing effects was discussed and demonstrated, using the data in the Darmstadt survey area. The simulation of data, for this analysis, was partially carried out using an algorithm that can compute the gravitational attraction of cylindrically shaped bodies. This algorithm, which has reduced the well-known problem in computing the gravitational attraction of vertical cylinders on points that lie outside the axis of symmetry, is developed based on an existing theoretical relationship. The effectiveness of this algorithm was also tested in modelling the effect of the different steal and concrete structures in the "Drop Tower Bremen". Besides this, the need to apply a modified terrain correction procedure for data measured at a certain height above the surface of the ground has been shown.
One of the major findings of this study, is the development of a trend analysis method that makes use of the polynomial fitting technique. This technique is advantageous in its capability to suppress the role of the residual in determining the trend surface. Especially its capability to incorporate a priori information about the trend surface enables the modelling procedure to focus on the trend surface only. In addition, the method estimates the densities for the Bouguer and terrain reductions in an environment where all other effects, other than the elevation dependant term, are modelled using different polynomial terms. In conjunction with this, the role of anomalous gradient effect in gravity data interpretation has been discussed and its effect in the Darmstadt survey area has been analysed.
The other major finding of the present study is the development of a three-dimensional gravity inversion technique. This method makes use of diagonal weighting matrices to reduce the need for large computer memory and to save large computational time. Besides this, the use of a priori information together with a regularization method that depends on the misfit between the measured and model data iteratively, leads to the desired stable solution. Based on this three-dimensional gravity inversion method, a software which can be used for different applications has been developed. The test made on simulated data and the field gravity data, in the Darmstadt survey area, showed the effectiveness of the method both in error-free and error contaminated data. The test made on error-free data, simulated by simple shape three-dimensional models, particularly showed the capability of the method in recovering the causative body using a minimal amount of a priori information. The inversion of measured gravity data taken from the Darmstadt survey area showed not only the effectiveness of the inversion method but also the ability of the present day gravity data interpretation technique in analysing weak signal gravity anomalies.
The comparison of the upward continued gravity data and that measured at different heights of the "Drop Tower Bremen" confirmed the validity of Newton's inverse square law, within the accuracy of the survey. Moreover, through the use of this comparison between the measured and analytically continued gravity data, the vertical component of the gravity gradient has been computed with high accuracy. In general, the study in both the Darmstadt and Bremen survey areas showed that the gravity method, if effectively used, can serve as a tool in locating mass distributions, even if these masses generate only weak signal gravity anomalies. |
Modelling and Inversion of High Precision Gravity Data (1997)
