Professional Documents
Culture Documents
Advanced Processing and Interpretation
Advanced Processing and Interpretation
Prepared by
GETECH
This short document is intended to provide background theory and methodology of the uses of
gravity and magnetic data in exploration.
Section 1 discusses the interpretation process itself, outlining the importance of qualitative
interpretation and the complementary roles that gravity and magnetic data offer.
Section 2 provides examples of the various types of enhancements (or transforms) applied to
gravity and magnetic data to highlight particular characteristics or features to aid qualitative
interpretation.
Section 3 describes additional advanced methods of quantitative processing in support of
interpretation that can be applied to gravity and magnetic data, including 3D gravity inversion,
depth to source estimation and 2D modelling.
advanced_processing_and_interpretation.doc
page 1
1.
As with all geophysical interpretation, the analysis of gravity and magnetic data has two distinct
aspects: qualitative and quantitative.
The qualitative process is largely map-based and dominates the early stages of a study. The
resultant preliminary structural element map is the cornerstone of the interpretation. Qualitative
interpretation involves recognition of:
the nature of discrete anomalous bodies including intrusions, faults and lenticular intrasedimentary bodies - often aided by reference to characteristic magnetic response charts
and perhaps performing simple test models
structural styles
The most important element in this preliminary qualitative stage surprisingly is not the
interpretation of anomalous bodies themselves (that follows later) but rather the network of
discontinuities e.g. lines of truncation and strike-slip faults that serve to compartmentalise and
delimit discrete anomalies that at first sight may appear as a confused pattern of unravellable
anomalies. Strike-slip faults/shear zones, small and large-scale, are commonplace particularly
within intra-continental situations where crust is old, bearing witness to countless fault
reactivations. They provide the principal means by which major structures are truncated and
crustal stress is decoupled (fully or partially) from one crustal block to another.
The quantitative process. Putting lines on maps during the qualitative process is the start of
quantitative phase. Refinement of these locations begins with the determination of z i.e. depth
values. For example, depth estimates to tops of anomalous magnetic bodies are generated by a
number of means including: slope measurement methods, analytic methods such as Euler and
Werner. Gravity and magnetic modelling (ideally seismically controlled) including forward and
inversion approaches contribute significantly to location in x, y and z. Accurate results of all these
rely upon sensible qualitative recognition of body types.
advanced_processing_and_interpretation.doc
page 2
Interpretation of magnetic data is theoretically more complex than the corresponding gravity
data due to:
the dipolar nature of the magnetic field, in contrast with the simpler monopolar gravity
field
the latitude/longitude dependent nature of the induced magnetic response for a given
body due to the variability of the geomagnetic field over the Earths surface
However, in practice it is often simpler than that of gravity due to the smaller number of
contributory sources. Often, though not always, there is just one source - the magnetic
crystalline basement.
geologic section.
In the case of intrasedimentary bodies, the dipolar nature of the magnetic response is particularly
diagnostic of the disposition (e.g. dip) of the source. It is for this reason that it is important for the
interpreter to be familiar with a wide range of induced magnetic responses produced by simple
geological bodies at the geomagnetic field inclination for the region. Seeking mutual consistency
advanced_processing_and_interpretation.doc
page 3
Characteristic geomagnetically induced magnetic responses for regions close to the geomagnetic
equator.
Modelling of potential field data is an important aspect of the interpretation, and is often
performed using a bottom-up / outside-in / magnetics-first approach. This ensures that deep
magnetic basement sources, which impact regionally on the study area, are understood first,
before attention is focused on the detail within the area of interest. The interpreter should always
be aware of the potential confusion generated by overprinting of similar wavelength responses
caused by: (i) deep crustal features, (ii) laterally distal crustal features, and, (iii) broad centrally
located shallow crustal features. Resolving this confusion is invariably achieved by seeking
consistency between the modelled gravity and magnetic data, while adhering to sensible
geological principles and experience. The following expands on this process.
A magnetics first approach recognises that the sedimentary section often possesses little
significant magnetic susceptibility. The major proportion of magnetic signal is generated at
crystalline (igneous or metamorphic) basement level. This is useful, because unlike gravity where
the entire section contributes to the observed field, all but the shortest wavelength magnetic
responses can be ascribed to the underlying basement. If shallow intra-sedimentary magnetic
sources do exist, these are usually of short wavelength and sufficiently discrete to be recognised
for what they are. The modelling of the magnetic data is particularly important for extending
interpretation below the effective level of seismic penetration. Once the magnetic data have
advanced_processing_and_interpretation.doc
page 4
Interpretation of magnetic anomalies close to the magnetic equator is complicated for several
reasons:
Magnetic anomalies are generated when the flux density cuts the boundary of a structure. If the
structure strikes parallel with the field then in Equatorial areas the flux stays within the structure
and no anomaly is generated.
Induced magnetic response of a 2D rifted basin striking W-E and N-S at or near the geomagnetic
equator. The sediments are assumed to have low susceptibility and the basement high susceptibility.
Small arrows show the induced magnetisation vector directions.
advanced_processing_and_interpretation.doc
page 5
Since faults and many structures have irregular shapes, albeit in regional form they may be 2D,
then parts of the structure will be magnetically imaged where the flux cuts the structural interface
generating dipole shape anomalies. Thus N-S striking structures may be identified by a string of
pearls i.e. line of magnetic dipole anomalies. The Analytic Signal is the best derivative to recover
the N-S contacts in equatorial regions as is shown by the diagram below.
advanced_processing_and_interpretation.doc
page 6
2.
The area used in this series of images is a region of Poland traversed (NW-SE) by the TeisseyreTornquist Zone which divides the shallower crystalline basement of the East European platform
to the NE, from the deeper West European platform to the SW. The thick Palaeozoic and
Mesozoic sedimentary cover of central Poland has undergone significant deformation (folding
and faulting) during the Caledonian, Variscan and Alpine orogenic phases. This has generated a
set of clear magnetic and gravity responses from basement and the sedimentary section that
allow similarities and differences to be clearly observed in the images generated.
The gravity images are on the left hand side of the page and the magnetic images are on the
right.
All the techniques described in this section were generated using GETECHs own GETgrid
software package. The software utilises FFT and spatial domain operators and has a host of
additional features (e.g. boolean logic, vector overlays, grid arithmetic).
advanced_processing_and_interpretation.doc
page 7
This technique transforms induced magnetic responses to those that would arise were the
sources placed at the magnetic pole (vertical field). This simplifies the interpretation because for
sub-vertical prisms or sub-vertical contacts (including faults), it transforms their asymmetric
responses to simpler symmetric and anti-symmetric forms. The symmetric highs are directly
centred on the body, while the maximum gradient of the anti-symmetric dipolar anomalies
coincides exactly with the body edge. Pole reduction is difficult at low magnetic latitudes, since
N-S bodies have no detectable induced magnetic anomaly at zero geomagnetic inclination. Pole
reduction is not a valid technique where there are appreciable remanence effects.
Pseudo-Gravity and Pseudo-Magnetic Fields
A magnetic grid may be transformed into a grid of pseudo-gravity. The process requires pole
reduction, but adds a further procedure which converts the essentially dipolar nature of a
magnetic field to its equivalent monopolar form. The result, with suitable scaling, is comparable
with the gravity map. It shows the gravity map that would have been observed if density were
proportional to magnetisation (or susceptibility). Comparison of gravity and pseudo-gravity maps
can reveal a good deal about the local geology. Where anomalies coincide, the source of the
gravity and magnetic disturbances is likely to be the same geological structure. (see Automatic
Lineament Tracing). Similarly, a gravity grid can be transformed into a pseudo-magnetic grid,
although this is a less common practice.
advanced_processing_and_interpretation.doc
page 8
Filtering is a way of separating signals of different wavelength to isolate and hence enhance
anomalous features with a certain wavelength. A rule of thumb is that the wavelength of an
anomaly divided by three or four is approximately equal to the depth at which the body
producing the anomaly is buried. Thus filtering can be used to enhance anomalies produced by
features in a given depth range.
Traditional filtering can be either low pass (Regional) or high pass (Residual). Thus the technique
is sometimes referred to as Regional-Residual Separation. Bandpass filtering isolates wavelengths
between user-defined upper and lower cut-off limits.
advanced_processing_and_interpretation.doc
page 9
advanced_processing_and_interpretation.doc
page 10
VDR =
This enhancement sharpens up anomalies over bodies and tends to reduce anomaly complexity,
allowing a clearer imaging of the causative structures. The transformation can be noisy since it
will amplify short wavelength noise. In our example it clearly delineates areas of different data
resolution in the magnetic grid.
advanced_processing_and_interpretation.doc
page 11
HDR =
+
x y
2
This enhancement is also designed to look at fault and contact features. Maxima in the mapped
enhancement indicate source edges.
derivative enhancements above. It usually produces a more exact location for faults than the first
vertical derivative, but for magnetic data it must be used in conjunction with the other
transformations e.g. reduction to pole (RTP) or pseudo-gravity. Specific directional horizontal
derivatives can also be generated to highlight features with known strikes. This technique can be
applied to pseudo-depth slices to image structure at different depths.
advanced_processing_and_interpretation.doc
page 12
The second vertical derivative serves much the same purpose as residual filtering in gravity and
magnetic maps, in that it emphasises the expressions of local features, and removes the effects of
large anomalies or regional influences. The principal usefulness of this enhancement is that the
zero value for gravity data in particular closely follows sub-vertical edges of intrabasement blocks,
or the edges of suprabasement disturbances or faults. As with other derivative displays, it is
particularly helpful in the processing stage where it can be used to highlight line noise or mislevelling.
+
AS =
+
x y z
The analytic signal, although often more discontinuous than the simple horizontal gradient, has
the property that it generates a maximum directly over discrete bodies as well as their edges. The
width of a maximum, or ridge, is an indicator of depth of the contact, as long as the signal arising
from a single contact can be resolved. This transformation is often useful at low magnetic
latitudes because of the inherent problems with RTP, (at such low latitudes).
advanced_processing_and_interpretation.doc
page 13
The automatic lineament detection algorithm requires the data to have been processed (or
transformed) such that the edge of a causative body is located beneath a maximum in the grid.
Several transforms satisfy this requirement e.g. horizontal derivative of gravity (or of pseudogravity, for magnetic data) and also analytic signal. The results help to quantify the different
gravity and magnetic responses of structures located in the shallow and deep sedimentary
sections and in the basement.
A significance factor N, ranging in value from 0 to 4, is assigned to each grid cell depending on
the relation to its neighbours. N=1 might represent a point on a spur, N=2 and N=3 a point on a
ridge and N=4 a point on a peak. The values of N are colour coded and displayed as a grid. These
lineament grids can then be displayed on top of any other grid.
advanced_processing_and_interpretation.doc
page 14
Aside from the data transformations applied to grids it is often beneficial to display grids
themselves in a variety of ways. This ensures that the maximum amount of information contained
in the transforms can be utilised in the interpretation phase. The following three grids of gravity
data show the same data displayed in grey-scale shaded relief, colour shaded relief and in a dipazimuth display. Vector data (station locations, flight lines, coastlines etc.) can be added as an
overlay. The dip-azimuth display highlights slope changes in all directions and is therefore useful
for picking out multiple trends in the data simultaneously.
advanced_processing_and_interpretation.doc
page 15
VDR
TDR = tan 1
THDR
The Tilt derivative (TDR) is similar to the local phase, but uses the absolute value of the horizontal
derivative in the denominator
Due to the nature of the arctan trigonometric function, all amplitudes are restricted to values
between + /2 and - /2 (+90and -90) regardless of the amplitudes of VDR or THDR. This fact
makes this relationship function like an Automatic Gain Control (AGC) filter and tends to equalise
the amplitude output of TMI anomalies across a grid or along a profile.
The Tilt derivatives vary markedly with inclination but for inclinations of 0and 90, its zero
crossing is located close to the edges of the model structures.
The Total Horizontal derivative of the TDR is independent of inclination, similar to the Analytic
Signal, but is sharper, generating better defined maxima centred over the body edges, which
persist to narrower features before coalescing into a single peak.
advanced_processing_and_interpretation.doc
page 16
3.
The enhancement techniques described in Section 2 generally help to estimate the 2D spatial
location of structures and their edges but do not generally provide estimates of the depth. The
exceptions are the pseudo-depth slicing and the analytic signal. This section describes semiautomated methods of depth estimation (3D Euler, Werner and SPI) and forward modelling of 2D
and 3D data. These are routinely used to model sub-surface structures constrained by seismic and
well data.
Euler deconvolution
GETECH has developed several in-house 3D anomaly interpretation packages for application to
total magnetic intensity (TMI) data, which employ Euler's homogeneity equation to identify
location, depth and nature of any sources present (Reid et al., 1990):
(x x 0 ) T + (y y 0 ) T + (z z 0 ) T = N(B T )
x
where:
(x0, y0, z0): the position of a source whose total field T is detected at any point (x,y,z)
B: the background value of the total field
N: the degree of homogeneity, interpreted physically as the attenuation rate with distance,
and geophysically as a structural index (SI):
Pipe
1 (Z)
Horizontal cylinder 1 (X or Y)
Dyke
2 (Z and X or Y)
Sill
2 (X and Y)
Contact
3 (X, Y and Z)
NA
advanced_processing_and_interpretation.doc
page 17
(a) Model topography, (b) forward modelled TMI field and (c) Euler solutions
advanced_processing_and_interpretation.doc
page 18
the model was created by taking a real topography dataset for an area with numerous exposed fault scarps of
varying size and orientation and then scaling these data to provide a buried topography analogue for the
faulted basement surface of a sedimentary basin.
advanced_processing_and_interpretation.doc
page 19
Buried topography test. Plot of 2D constrained Euler solution depth versus model depth at the same x,y
location for homogeneous susceptibility basement model (from Williams, 2004).
Buried topography test. (a) Manually defined polygons (shown in grey) to isolate clusters of 2D
constrained Euler solutions (blue dots) for analysis of averaged source parameters. (b) Plot of mean
solution depths, plotted at the mean solution x,y location, of the solution clusters defined in (a), with
contours showing the basement depth in the same colour scale (contour interval 200 m). (c) Mean 2D
constrained Euler solution depth versus model depth (from Williams, 2004)
advanced_processing_and_interpretation.doc
page 20
advanced_processing_and_interpretation.doc
page 21
advanced_processing_and_interpretation.doc
page 22
(a) synthetic TMI field and (b) Local wavenumber depth estimates
advanced_processing_and_interpretation.doc
page 23
2D Profile Modelling
GETECH uses the GM-SYS modelling software from Northwest Geophysical Associates, Inc. (NGA).
It is an interactive forward modelling program which calculates the gravity and magnetic
response from a user defined hypothetical geologic model. Any differences between the model
response and the observed gravity and/or magnetic field are reduced by refining the model
structure or properties (e.g. density or susceptibility of model components).
It should be noted that gravity and magnetic models are non unique, i.e. many earth models can
produce the same gravity and/or magnetic response, and similarly, several geological lithologies
may be interpreted from a given model blocks density and susceptibility properties. It is
therefore important to use as many independent sources of information as possible to help
constrain the model, e.g. seismic structural horizons and density logs from wells located near the
profile. Such control may be included in the GM-SYS model as image backgrounds (e.g. depth
converted seismic lines) or as symbols (e.g. wells with lithology tops annotated with depth).
advanced_processing_and_interpretation.doc
page 24
contributed to by the entire geologic section. To accurately model the upper crustal, residual
components requires accurate definition of the regional, lower crustal density variations, such as
Moho relief. In some cases, the positive regional gravity response from extended crust, giving rise
to an elevated Moho, can be relatively well constrained from the gravity profile itself. An example
is provided overleaf where the gravity profile shows negative perturbations (due to the basin
sediments) from a regional, long wavelength gravity high. Alternatively, two shorter wavelength
highs may be observed on either side of the basinal gravity low from which the regional may also
be interpolated:
advanced_processing_and_interpretation.doc
page 25
The difficulty arises when the residual gravity lows due to the sedimentary fill almost cancel the
gravity high due to crustal thinning and an elevated Moho (i.e. the basin is isostatically
compensated), or if ramp flat detachment geometry prevails and the regional gravity high is not
laterally coincident with the basinal gravity low. In these cases as much additional information as
possible is used to constrain the model, such as well data or simultaneous magnetic modelling.
Werner Deconvolution
Werner deconvolution is a profile-based interactive technique used to analyse the depth to and
horizontal position of magnetic source bodies, and the related parameters of dip and
susceptibility. It is a rigorous, iterative, two-dimensional inversion technique that takes into
account interference from adjoining anomalies. Analysis of the total magnetic intensity data
yields these parameters for thin, sheet-like bodies such as dikes, sills, intruded fault zones, and
basement plates of minor relief compared to the source-sensor separation distance. Applied to
the horizontal gradient data Werner Deconvolution yields similar parameters for geologic
interface features such as dipping contacts, edges of prismatic bodies, major faults, and slope
changes of the basement surface.
advanced_processing_and_interpretation.doc
page 26
References
BLAKELEY, R.J. AND SIMPSON, R.W, 1986. Approximating edges of source bodies from magnetic or
gravity anomalies. Geophysics, v51, No 7, pp 14941498.
FAIRHEAD, J.D., WILLIAMS, S.E. AND FLANAGAN, G., 2004. Testing Magnetic Local Wavenumber
Depth Estimation Methods using a Complex 3D Test Model. SEG Annual Meeting,
Denver, Extended Abstract.
MUSHAYANDEBVU, M.F., LESUR, V., REID, A.B. AND FAIRHEAD, J.D., 2004. Grid Euler deconvolution
with constraints for 2D structures. Geophysics, v69, pp 489-496
NABIGHIAN, M.N. AND HANSEN, R.O., 2001. Unification of Euler and Werner deconvolution in three
dimensions via the generalized Hilbert transform. Geophysics, v66, No 6, pp 1805-1810.
PHILLIPS, J.D., HANSEN, R.O., AND BLAKELY, R.J., 2006. The Use of Curvature in Potential-Field
Interpretation. ASEG2006, expanded abstracts.
REID, A.B., ALLSOP, J.M., GRANSER, H., MILLET, A.J., AND SOMERTON, I.W., 1990. Magnetic
interpretation in three dimensions using Euler deconvolution. Geophysics v55 pp 80-91.
SALEM, A., WILLIAMS, S., FAIRHEAD, J.D., RAVAT, D., AND SMITH, R., in press. Tilt-Depth method: A
simple depth estimation method using first order magnetic derivatives. Submitted to
The Leading Edge.
SPECTOR, A. AND GRANT, F.S., 1970. Statistical Models for Interpreting Aeromagnetic data.
Geophysics, v35, No 2, pp 293-302.
THURSTON, J.B., AND SMITH, R.S., 1997. Automatic conversion of magnetic data to depth, dip, and
susceptibility contrast using the SPI (TM) method. Geophysics, v62, No 3, pp 807 -813.
WILLIAMS, S.E., 2004. Extended Euler deconvolution and interpretation of potential field data from
BoHai Bay, China. PhD Thesis (unpublished), University of Leeds.
advanced_processing_and_interpretation.doc
page 27