Comparison of some commonly used regional ...

23 downloads 0 Views 438KB Size Report
(Nettleton, 1954; Fuller, 1967) and, in that case, the upward continuation distance is an arbitrary parameter. A more rigorous approach is proposed by Jacobsen ...
Downloaded 04/26/16 to 132.156.2.2. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Comparison of some commonly used regional residual separation techniques Pierre Keating*, Nicolas Pinet and Mark Pilkington, Geological Survey of Canada Summary Regional-residual separation is essential in gravity and magnetic data interpretation and a variety of techniques can be used. This is generally done by estimating a regional field which is later subtracted from the measured field to obtain a residual field. Here we compare some simple and easy to use techniques. Graphical techniques allow including geological information in the estimation. Low pass and non-linear filtering, and upward continuation can also be used. These techniques are tested using gravity and magnetic data from the Gaspé Peninsula in eastern Canada. We show that for gravity data graphical techniques and nonlinear filtering give similar results in our study area. Upward continuation and non-linear filtering give equivalent results, however upward continuation is easier to use and implement. For magnetic data good results are obtained using low-pass and non-linear filtering. In practice one should always use two different techniques to determine the regional field since a good agreement between their results can be an indication of an acceptable regional field.

Pinet et al. (2006) in a sedimentary basin. Simple upward continuation can also be used to obtain the regional field (Nettleton, 1954; Fuller, 1967) and, in that case, the upward continuation distance is an arbitrary parameter. A more rigorous approach is proposed by Jacobsen (1987) who assumes that the field is due to an ensemble of mutually uncorrelated thin sources. He shows that the field from a slab located z1 and z2 is simply the difference between the field upward continued to 2z1 and 2z2. Upward continuation can then be used as a standard filter for potential field separation. Although the assumption of non correlation between the layers is often difficult to meet, the resulting maps remain very useful for anomaly detection and pattern discrimination (Jacobsen, 1987). Low pass filtering can also be used to define the regional field. However, since the spectrum of most geological features is broadband the spectra of features located at different depths overlap and that their spectra cannot be separated completely (Telford et al., 1990). Keating and Pinet (2011) have shown how nonlinear filters can also be used to determine a regional field anomaly that closely resembles the one obtained from graphical techniques.

Introduction Application to gravity data The separation of gravity and magnetic field anomalies into their respective regional and residual components is necessary for their quantitative interpretation. Various approaches to regional-residual separation exist and there is no unique solution; different residual fields will therefore result in different interpretations. We discuss some of the techniques in use and compare the results of their application on real data. All separation techniques make some hypothesis about the source distribution: a common assumption is that the observed field is the sum of the effects of shallow and deep sources. However, it is possible for the so-called shallow sources to extend to great depth and this makes the separation difficult. Here we restrict ourselves to techniques that are easy to implement. Results from synthetic models are presented as well as results from a case study in eastern Canada and compared to regional gravity and magnetic anomalies obtained by these techniques. Methodologies

The techniques are tested on gravity and magnetic data from the Gaspé Peninsula area in Eastern Canada. Pinet et al. (2006) used a graphical technique to determine the regional gravity field from 24 profiles. The basic assumption is that the regional component of the gravity field should not include the effect of outcropping or shallow sources. The Bouguer gravity anomaly is shown in Figure 1. The depth extent of the Sept-Iles layered intrusion as interpreted from gravity data is 7 km and the depth extent of the Shickshock Group is 5 km according to seismic data. Obviously, the regional should not include the effect of these sources. The gravity anomaly associated with the Sept-Iles layered intrusion has diameter of about 75 km, while the northsouth extent of the Shickshock anomaly is about 25 km and its length about 150 km. The regional obtained from graphical techniques presented in Figure 2 is characterized by its general decrease to the north-west and a large area of nearconstant field in its southeastern part.

The oldest technique is graphical separation. In this case, the interpreter uses a series of profiles along which the separation is manually performed. This process allows taking into account surface data and other information such as the known or estimated depth of some sources. Typical examples of the use of graphical separation are given by Gupta and Ramani (1980) in a Precambrian terrain, and

Upward continuation as a separation filter is first tested by using a continuation height of 14 km which should remove the gravity effect of bodies that have depth extent less than 7 km, the maximum depth extent of the Sept-Iles layered intrusive and the Shickshock Group. The resulting regional nevertheless contains anomalies that are directly associated to these two bodies. A series of increasing continuation

©The Society of Exploration Geophysicists and the Chinese Geophysical Society GEM Beijing 2011: International Workshop on Gravity, Electrical & Magnetic Methods and Their Applications Beijing, China. October 10-13, 2011.

Downloaded 04/26/16 to 132.156.2.2. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Comparison of regional residual separation techniques

heights are tested and the effect of the Sept-Iles and Shickshocs anomalies is finally removed at a continuation height of 80 km which, in theory, should remove the gravity effect of sources located between the surface and a depth of 40 km.

southeast. Moreover, amplitudes of both the graphical and non-linear filtered regional anomaly maps are approximately the same. Application to magnetic data We now compare the regional anomalies obtained by low pass filtering, upward continuation and non-linear filtering. The study area is located within the central part of the region covered by the gravity data. The residual total magnetic field obtained after removing the IGRF is characterized by short-wavelength anomalies superimposed on a wide magnetic depression (see Figure 3). The magnetic anomalies located in the southern part of the map are associated with the volcanics of the Mont Alexandre syncline while the wide and high amplitude anomaly in the north-west corner of the map corresponds to the Lemieux intrusive suite.

Figure 1: Bouguer anomaly in the Gaspé Peninsula region

Figure 3: Residual total magnetic field over the central part of the Gaspé Peninsula

Figure 2: Regional anomaly obtained by the use of graphical techniques. Adapted from Pinet et al., 2006 To obtain a regional using non-linear filtering, an 80 km long non-linear filter is first successively applied along the columns and rows of the grid. The resulting regional does not show any residual anomalies that may be caused by near-surface features. Similarly to the regional obtained by the graphical technique, the non-linear regional decreases to the north-west and exhibits a wide maximum to the

Given the apparent width of the narrow anomalies that we wish to remove, we use a 24 km low pass Butterworth filter to estimate the regional. The resulting regional, shown in Figure 4, is smooth, but the effect of the Mont Alexandre volcanics (a broad high) is still clearly visible making this regional unacceptable. A continuation height of 12 km is also used to compute a regional field since the maximum depth extent of the Mont Alexandre syncline is 7.5 km, but the depth extent of Lemieux intrusive suite (Pinet et al., 2010) is unknown. This regional, shown in Figure 5, is smooth and does not show any effect due to short wavelength anomalies. However, we note that the amplitude

©The Society of Exploration Geophysicists and the Chinese Geophysical Society GEM Beijing 2011: International Workshop on Gravity, Electrical & Magnetic Methods and Their Applications Beijing, China. October 10-13, 2011.

Downloaded 04/26/16 to 132.156.2.2. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Comparison of regional residual separation techniques

range is smaller than in Figure 4. To test non-linear filtering we use a 4.5 km long non-linear filter since the apparent width of the magnetic anomalies caused by the Mont Alexandre syncline volcanics is about 4 km at the most. The calculated regional (not shown here) still shows some response from the hinge zone of the syncline. All short wavelength responses in the north-western part of the map have been removed and only the high-amplitude long wavelength component due to the Lemieux intrusive suite intrusive is present.

A better evaluation of these results is obtained by comparing profiles (Figure 6) of the measured residual total magnetic field and calculated regional fields. The regional fields obtained from non-linear and low-pass filtering are nearly identical except a relatively minor (~20 nT) discrepancy over the narrow high-amplitude magnetic anomaly due to the Mont Alexandre volcanics. There the regional field calculated from the non-linear filter is slightly less affected by this sharp anomaly than the regional obtained from low-pass filtering. Over the Lemieux intrusive the regional fields estimated by non-linear and low-pass filtering are similar but we cannot determine if they are correct since too little is known on the structure of this intrusive at depth. This is a good example for the importance of geological constraints when determining a regional field.

Figure 6: Measured magnetic data and estimated regional fields along profile A-B indicated in Figure 4 Figure 4: Regional field from 24 km low-pass filtering

The regional estimated from the upward continued magnetic field at a height of 12 km does not correspond with the regional that would be drawn by an experienced interpreter. In particular, its amplitude is too high over most of the profile, and too low over its northern part. Also, over the south portion of the Lemieux intrusive, we believe that none of the tested techniques provide an adequate regional background. At this geomagnetic latitude we expect the regional field to be at about the same level as the base of the three small amplitude magnetic anomalies located there, however it is much higher. Over this section, the lengths of the low pass and non-linear filters used to estimate the regional are two long. On the other hand, selecting shorter filter lengths would result in poor results over the Mont Alexandre volcanics. Conclusions

Figure 5: Regional field from upward continuation to a 12 km height

The good correspondence between regional gravity anomalies determined by a graphical technique and non-linear filtering is not surprising as non-linear filtering somewhat mimics what one does graphically when removing local anomalies. The only geological a priori information that can be taken into account in the non-linear filtering is the

©The Society of Exploration Geophysicists and the Chinese Geophysical Society GEM Beijing 2011: International Workshop on Gravity, Electrical & Magnetic Methods and Their Applications Beijing, China. October 10-13, 2011.

Downloaded 04/26/16 to 132.156.2.2. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Comparison of regional residual separation techniques

apparent width of the anomalies to be removed. For gravity data, the use of upward continuation as a separation filter was not as successful as expected. By trial and error we found that the regional obtained from an upward continuation of the gravity field to a height of 80 km shows some correspondence to the regional gravity anomalies determined from graphical or non-linear filtering, however, in theory this is equivalent to removing all sources between the surface and a depth of 40 km. This seems excessive as the thickness of the crust is about 40 km in this region (Marillier and Verhoef, 1989). Upward continuation makes the hypothesis that sources are uncorrelated thin horizontal plates located at different depths. This is not the case here, as some of the outcropping sources such as the Sept-Iles and Shickshocks anomalies which extend laterally for several kilometers have more or less subvertical sides and are thus vertically correlated. Upward continuation can nevertheless be used as an ad hoc separation filter if one neglects its physical significance. For magnetic data, non-linear and low pass filtering perform equally well while upward continuation does not provide an acceptable regional field in the example shown here. We were unable to determine a continuation height that provides an acceptable regional magnetic field as it was the case in the gravity example. Interestingly, for the magnetic data set used in this study, we could easily select the parameters of the low pass filter that gives good results. This is because the magnetic anomalies due to the Mont Alexandre volcanics have a well defined width which is of a much shorter wavelength than the other anomalies present on the map.

Keating, P. and Pinet, N, 2011, Use of non-linear filtering for the regional-residual separation of potential field data: Journal of Applied Geophysics, 73, 315-322. Marillier, F., and Verhoef, J., 1989, Crustal thickness under the Gulf of St. Lawrence, northern Appalachians, from gravity and deep seismic data. Canadian Journal of Earth Sciences, 26, 1517-1532. Nettleton, L. L., 1954, Regionals residuals and structures: Geophysics, 19, no.1, 1-22. Pinet, N., Keating, P., Brouillette, P., and Dion D.-J., 2006, Production of a residual gravity anomaly map for Gaspésie (northern Appalachian Mountains), Quebec, by a graphical method: Geological Survey of Canada, Current Research 2006-D1, 8p. Pinet, N., Keating, P., Lavoie, D., and Brouillette, P., 2010, Forward potential-field modeling of the Appalachian orogen in the Gaspé Peninsula (Québec, Canada): Implications for the extent of rift magmatism and the geometry of the Tacadian orogenic wedge: American Journal of Science, 310, 89-110. Telford, W.M., Geldart, L.P., Sheriff, R.E., and Keys, D.A., 1990, Applied geophysics, Cambridge University Press, Cambridge.

We note that for both for gravity and magnetic data that at least two different techniques give similar results. We conclude that in practice one should always use two different techniques to determine the regional field since a good agreement between their results can be an indication of an acceptable regional field. It is also likely that different results would be obtained in other geological contexts. References Fuller, B. D., 1967, Two-dimensional frequency analysis and design of grid operators: Mining geophysics, V-II, 658708. Gupta, V. K., and N. Ramani, 1980, Some aspects of regional residual separation of gravity anomalies in a preCambrian terrain: Geophysics, 45, no.9, 1412-1426. Jacobsen, B. H., 1987, A case for upward continuation as a standard separation filter for potential field maps: Geophysics, 52, no.8, 1138-1148.

©The Society of Exploration Geophysicists and the Chinese Geophysical Society GEM Beijing 2011: International Workshop on Gravity, Electrical & Magnetic Methods and Their Applications Beijing, China. October 10-13, 2011.