Survey Geofisika: GRAVITY SURVEY FOR BASEMENT CONFIGURATION

figure 1 Bouguer Anomaly

Gravity observations at the earth’s surface includes anomalies that result from the density of subsurface structures. A small body will cause an anomaly with a short wavelength while a large body will cause an anomaly with a longer wavelength. Similarly deeper structures will generate longer wavelength anomalies than shallower structures. In order to isolate these changes, a wavelength filter and spectral analysis were designed around the specified areas of interest. These tools assisted in highlighting the geological structures to be analyzed.

Filtering is a way of separating signals of different wavelengths to isolate and hence enhance anomalous features within a given bandwidth. A good rule-of-thumbs is that the wavelength of an anomaly divided by three or four is approximately equal to the depth at which body producing the anomaly is located. Using this filtering technique can it is possible to enhance anomalies produced by features in a given depth range. This filtering technique is sometimes referred to as Regional-Residual Separation.

In order to extract meaningful information from the gravity data we applied an optimum Wiener filter using Geosoft® program (Geosoft Inc., 1994). The purpose of this filter is to reduce the effect of anomalous source up to a certain depth leaving a smoother anomaly associated with deeper sources. The results of the Wiener filter which rejects source up to 250, 500, 750 and 1000 meters depth are presented in Figure 16, 17, 18 and 19 respectively.

figure 2 Regional Anomaly1

figure 3 Residual Anomaly 1

The quantity to be determined in gravity exploration is local lateral variation in density. Generally density is not measured in situ, although it can be measured by borehole logging tools or estimated from seismic velocities. More often density measurements are made in the laboratory using small outcrop or drill-hole samples. Laboratory results, however, rarely give the true bulk density because the samples may be weathered, fragmented, dehydrated, or altered in the process of being collected. Consequently, for field specific situations, density measurements are seldom carried out.

A reasonably satisfactory method of estimating near-surface density uses a gravity profile over topography that is not co relatable with density variation (Nettleton, 1976).

Density values are applied to gravity measurements while reducing these measurements to create Bouguer Anomaly profiles. In addition the effects of the variation of the terrain around the measurement site have to be corrected for. The effect of the terrain correction on the gravity data is a function of the density.

A set of profiles of Bouguer Anomaly values using different density values can be generated. Typically the profile that is least affected by the terrain is the one with the best density match.

figure 4 Nettleton Method

A more quantitative method than the graphic method used during the Nettleton density derivation, described above, is the Parasnis method. The Parasnis method uses a mathematical least-square algorithm to determine the density value that best fits the test data set.

Linear regression (least squares) method

Assumes no correlation between topography and subsurface density (i.e., anomalies are randomly distributed with respect to elevation)
Therefore correlation between topography and gravity (g) will be due to Bouguer slab
Plot Free Air Anomaly (Dg _fa₎ against elevation (h)
Fit line through points

slope will approximate 2pGr; solve for r (Bouguer density)

figure 5 Regional Anomaly2

figure 6 Residual Anomaly 2

As an additional processing, we also tried a preliminary inversion of the gravity data to obtain an insight into 3-D density distribution of the area. The inversion technique is a modified version of 3-D magnetic inversion presented by Yudistira and Grandis (2001) which follows a technique proposed by Fedi and Rapolla (1999). A model-smoothing factor was used in the resolution of the matrix inversion by truncated singular value decomposition (SVD) method (Press et al., 1987).

The band-pass filtered gravity data (with wavelength of 2 – 10 km) were re-sampled into a 500 ´ 500 meters grid spacing. The subsurface is represented by a grid of rectangular prism with 500 ´ 500 ´ 100 meter in dimension which covers the whole survey area up to 3000 meters depth. This under-sampling of the data and model was necessary to reduce the computation time of the 3-D inversion. Therefore, we prefer looking for rather large scale feature by inverting the gravity data filtered for 2 – 10 km wavelength.

The inversion was performed without any “a priori” constrain and the obtained model of density distribution (or contrast density relative to 2.0 gram/cm³) at three depth slices (500, 1000 and 2500 meters) is presented in Figure 23. These figures of the model are given without scale or coordinate for simplicity, but they cover the whole survey area as other figures or maps.

The models from 500, 1000 and 2500 meters depth slices represent superficial, intermediate and deep density distribution respectively. We can observe that the model is merely another type of representation of the gravity data, i.e. 1000 meters depth slice is nearly identical to 1 – 5 km band-pass filtered gravity data while 2500 meters depth slice reproduces 2 – 10 km band-pass filtered gravity data. Inversions of other filtered gravity data gave similar results, i.e. the model tends to mimic the data such that they are related each other by a simple transformation. This fact is mainly due to the lack of constrain in the inversion procedure such that the inversion only reproduces the data with no or little additional information.