See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/379736467

Using Euler Deconvolution as Part of a Mineral Exploration Project

Article in Minerals · April 2024

DOI: 10.3390/min14040393

CITATIONS READS

0 41

1 author:

Gordon Robert John Cooper

University of the Witwatersrand
251 PUBLICATIONS 4,002 CITATIONS

SEE PROFILE

All content following this page was uploaded by Gordon Robert John Cooper on 12 April 2024.

The user has requested enhancement of the downloaded file.

 minerals
Article
Using Euler Deconvolution as Part of a Mineral
Exploration Project
G. R. J. Cooper

School of Geosciences, University of the Witwatersrand, Johannesburg 2050, South Africa;

[email protected]; Tel.: +27-11-7176-608; Fax: +27-11-7176-579

Abstract: Mineral exploration projects can make considerable use of a variety of geophysical tech-
niques and datasets, including magnetic and gravity data. The interpretation of large quantities
of data can be very time consuming, so semi-automatic interpretation techniques are often used
to provide initial estimates of the parameters (primarily the location and depth) of the sources of
anomalies. Euler deconvolution is a commonly used interpretation method for potential fields which
has a number of advantages over many other techniques, such as working in the presence of remanent
magnetisation, and not being restricted to a particular model such as a contact. A second-order
version of Euler’s equation is introduced here, which is much less affected by trends in the data than
the standard method and additionally produces depth parabolas, which simplify the interpretation of
results. The method was applied to aeromagnetic data from a mineral exploration project in Southern
Africa and provided plausible results.

Keywords: gravity; magnetics; exploration

1. Introduction
Mineral exploration projects can make considerable use of a variety of geophysical
techniques, including magnetic and gravity data. The interpretation of large quantities of
geophysical data can be very time consuming, so semi-automatic interpretation techniques
are often used to provide initial estimates of the parameters (primarily the location and
Citation: Cooper, G.R.J. Using Euler
depth) of the sources of anomalies There are many semi-automatic interpretation tech-
Deconvolution as Part of a Mineral
niques available, such as analytic-signal-amplitude-based methods [1–4], Werner deconvo-
Exploration Project. Minerals 2024, 14, lution [5], source-distance approaches [6,7], and Euler deconvolution. Euler deconvolution
393. https://doi.org/10.3390/ is based on Euler’s homogenous function theorem, which is widely used throughout math-
min14040393 ematics and physics. It states that for homogenous functions (i.e., those in which all terms
are raised to the same power) of degree N, then [8–11]
Academic Editor: Stanisław Mazur

Received: 8 March 2024 f (t·x,t·z) = tN f (x,z) (1)

Revised: 5 April 2024
Accepted: 6 April 2024
Differentiating Equation (1) with respect to t, and then setting t = 1, provides
Published: 10 April 2024
∂f ∂f
x +z = −N· f (2)
∂x ∂z
Note that Equation (2) basically states that function f can be written in terms of
Copyright: © 2024 by the author.
combinations of its first order derivatives. When applied to potential field data f, then x
Licensee MDPI, Basel, Switzerland.
and z are interpreted as the distances (∆x and ∆z) from the current point to the source in the
This article is an open access article
x and z planes. The degree of homogeneity N (which is also termed the structural index,
distributed under the terms and
or SI) is the rate of decay of the amplitude of the field with distance from the source, e.g.,
conditions of the Creative Commons
N = 1 for the magnetic response of a dyke. Equation (2) is usually solved for ∆x and ∆z
Attribution (CC BY) license (https://
using a moving window of data points, with N being specified. The solutions are effectively
creativecommons.org/licenses/by/
4.0/).
averaged over the window size, so while larger window sizes can reduce the effect of

Minerals 2024, 14, 393. https://doi.org/10.3390/min14040393 https://www.mdpi.com/journal/minerals

Minerals 2024, 14, 393 2 of 8

random noise, they can result in the smearing of the solution’s horizontal location and can
also make the results more sensitive to interference from adjacent anomalies. Sometimes
a background field term B is subtracted from the field f, but Euler’s equation is only for
single sources, which implies that (if necessary) this has already been done.
Euler deconvolution can also be applied to the Hilbert transform H(f ) of the data [12,13]

∂f ∂f
∆x − ∆z = −N·H( f ) (3)
∂z ∂x

2. Materials and Methods

Derivatives of potential fields are also homogenous functions, so Euler deconvolution
is frequently applied to them [14,15]

∂2 f ∂2 f ∂f
∆x + ∆z = −( N + 1) (4)
∂x2 ∂x∂z ∂x

∂2 f ∂2 f ∂f
∆x + ∆z 2 = −( N + 1) (5)
∂x∂z ∂z ∂z
Applying Euler deconvolution to higher order derivatives of the field can reduce the
sensitivity of the method to regional field and interference issues. Euler deconvolution can
also be applied to combinations of the derivatives, such as the analytic signal amplitude [16]
or the Tilt angle [17].
Substituting Equations (4) and (5) into Equation (2) provides the second-order
Euler equation
∂2 f ∂2 f ∂2 f
∆x2 2 + 2∆x ·∆z + ∆z2 2 = N ( N + 1) f (6)
∂x ∂z∂x ∂z
Higher order equations can be generated in a similar manner. Using Laplace’s equation, then
  ∂2 f ∂2 f
∆x2 − ∆z2 + 2∆x∆z = N ( N + 1) f (7)
∂x2 ∂z∂x

Equation (7) is solved for (∆x2 − ∆z2 ) and (2∆x·∆z) using a moving window of data
points in the usual manner, then ∆x and ∆z are obtained by solving a quadratic equation,
i.e., let a = (∆x2 − ∆z2 ) and b = (2∆x·∆z), then
 1/2 1/2  1/2 
1 2 2 2 2
∆x = √ a +b −a a +b +a (8)
2b
 1/2 1/2
1  2
∆z = √ a + b2 −a (9)
2
Similarly substituting Equations (4) and (5) into Equation (3) provides the Hilbert
transform of Equation (7);
  ∂2 f ∂2 f
∆x2 − ∆z2 − 2∆x∆z 2 = N ( N + 1) H ( f ) (10)
∂z∂x ∂x
Note that the second-order equations do not have any first order derivative terms, which
makes them less sensitive to the presence of linear regional fields than Equations (2) and (3).
√ √
Additionally, as the depth of the source ∆z = −a when ∆x = 0 (Equation (7)), plotting −a
yields a ‘depth parabola’, which is a useful check on the validity of the Euler solutions.
Equations (4)–(10) use second-order derivatives of the field, which naturally can make
them sensitive to noise. If this is an issue, then either regularised derivatives [18] or upward
continuation of the field can be used.
 Equations (4)–(10) use second-order derivatives of the field, which naturally can
make them sensitive to noise. If this is an issue, then either regularised derivatives [18] o
upward continuation of the field can be used.

Minerals 2024, 14, 393 3. Results 3 of 8

Euler deconvolution is applied to the magnetic anomaly from an isolated dyk

(Figure 1), to which a small linear trend has been added. The solutions from standard
3. Results
Euler deconvolution (Equation (2), Figure 1b are widely dispersed around the top of th
Euler deconvolution is applied to the magnetic anomaly from an isolated dyke
dyke, while
(Figure 1),those from
to which the second-order
a small linear trend hasEuler
beendeconvolution (Equations
added. The solutions (7) and (10)) ar
from standard
much more
Euler tightly grouped
deconvolution (Equation(particularly
(2), Figure 1b in
arethis
widelycase when the
dispersed Hilbert
around transform
the top of the wa
dyke,
used). while
Euler those from theissecond-order
deconvolution well knownEuler deconvolution
for its production(Equations (7) and (10))
of large numbers of spuriou
are much
solutions more tightly
scattered grouped (particularly
throughout in thisand
the subsurface, casethis
whenis the
theHilbert transform
case even withwas
this simpl
used). Euler deconvolution is well known for its production of large numbers of spurious
synthetic model. However, plotting the depth parabola provided by Δz = √−a helps to
solutions scattered throughout the subsurface, and this is the case even with this simple
identify the model.
valid solutions, as it should have a maximum depth √
synthetic However, plotting the depth parabola provided by ∆zextent
= −aover helpsthe
to source
These depth
identify theparabolas are clearly
valid solutions, located
as it should haveover the tops
a maximum of the
depth dykes
extent over (Figure 1c,d) and
the source.
Euler solutions
These that areare
depth parabolas farclearly
from located
them canoverbe
thedisregarded.
tops of the dykes (Figure 1c,d) and Euler
solutions that are far from them can be disregarded.

Figure 1. (a) Magnetic anomaly from the dyke shown in (b–d) below. A linear trend (dotted line) is
added to the data. (b) Solutions from Euler deconvolution (Equation (2)) are shown as blue + symbols.
A SI of 1 and window size of 11 points is used. (c) Solutions from second-order Euler deconvolution
(Equation (7)) are shown as blue + symbols. A SI of 1 and window size of 11 points is used. The
depth parabola is overlain as a black line. (d) Solutions from second-order Euler deconvolution
(Equation (10)) are shown as blue + symbols. A SI of 1 and window size of 11 points is used. The
depth parabola is overlain as a black line.

A second, more complicated example application is shown in Figure 2. In this case,

there are now several dykes with different dips and depths, and whose anomalies interfere.
 deconvolution (Equation (10)) are shown as blue + symbols. A SI of 1 and window size of 11 point
is used. The depth parabola is overlain as a black line.

A second, more complicated example application is shown in Figure 2. In this case

Minerals 2024, 14, 393 4 of 8
there are now several dykes with different dips and depths, and whose anomalie
interfere. All of the methods successfully locate the shallow dyke on the left of the profile
TheAlldeeper dykes are
of the methods more difficult
successfully targets,
locate the shallowhowever,
dyke on theandleftwhile clusters
of the profile. Theofdeeper
solutions ar
present
dykesnear them,
are more the targets,
difficult clusters are often
however, rather
and while dispersed,
clusters and are
of solutions other solutions
present near exis
them, the
between theclusters
dykes.areIf often
only rather dispersed,
the solution and other
clusters thatsolutions
are near exist
to between
the depththe parabolas
dykes. ar
If only the solution clusters that are near to the depth parabolas are considered
considered as being valid, then all the dykes have solutions from the second-order Eule as being
valid, then all the dykes have solutions from the second-order Euler deconvolution lying
deconvolution lying near to their upper surface (the depth parabola associated with th
near to their upper surface (the depth parabola associated with the very shallow dyke is
veryhard
shallow dyke
to see due to is
thehard to seeEuler
coincident due solutions).
to the coincident
Similarly,Euler
depth solutions).
parabolas that Similarly,
are not depth
parabolas
associatedthat areclusters
with not associated with clusters
of Euler solutions of Euler solutions can be ignored.
can be ignored.

Figure 2. (a)
Figure Magnetic
2. (a) Magneticanomaly fromthe
anomaly from thedyke
dyke model
model shown
shown in (b-d)
in (b-d) below.below. (b) Solutions
(b) Solutions from Eule
from Euler
deconvolution
deconvolution (Equation
(Equation(2)) areshown
(2)) are shownasas blue
blue + symbols.
+ symbols. A SIA of SI of 1window
1 and and window size
size of 11 of 11 point
points
is used.
is used. (c) (c) Solutionsfrom
Solutions from second-order
second-order Euler
Eulerdeconvolution (Equation
deconvolution (7)) are(7))
(Equation shown
are as blue +as blue
shown
symbols. A SI of 1 and window size of 11 points is used. The depth parabola is overlain as a black line.
(d) Solutions from second-order Euler deconvolution (Equation (10)) are shown as blue + symbols. A
SI of 1 and window size of 11 points is used. The depth parabola is overlain as a black line.
Minerals 2024, 14, 393 5 of 8

Minerals 2024, 14, x FOR PEER REVIEW 5 of 8

4. Application to a Mineral Exploration Project in Southern Africa

Figure 3a shows an aeromagnetic data set from Southern Africa that was used as part
symbols. A SI of 1 and window size of 11 points is used. The depth parabola is overlain as a black
of a mineral exploration project. Unfortunately, confidentiality issues preclude most of the
line. (d) Solutions from second-order Euler deconvolution (Equation (10)) are shown as blue +
details being provided here. The flight line direction of the survey was north–south, the
symbols. A SI of 1 and window size of 11 points is used. The depth parabola is overlain as a black
line
line. spacing was 1 km, the grid interval was 0.25 km, and the flight height was 100 m. This
dataset was interpreted by Cooper using a new Contact-Depth method [6], which only
works for the magnetic
4. Application to a Mineralanomalies from contacts
Exploration Project and that requires
in Southern pole reduced data. The
Africa
depths to the contacts are obtained at the location of the zero values of the first vertical
Figure 3a shows an aeromagnetic data set from Southern Africa that was used as part
derivative of the data. The results are shown in Figure 3b. Two profiles were abstracted
of a mineral exploration project. Unfortunately, confidentiality issues preclude most of the
from this aeromagnetic dataset and their locations are marked in Figure 3a. The first profile
details being provided here. The flight line direction of the survey was north–south, the
(marked ‘A’) is shown in Figure 3c, and the results of applying Euler deconvolution to it
line spacing was 1 km, the grid interval was 0.25 km, and the flight height was 100 m. This
are shown in Figure 3d–f. The Contact-Depth results provided depths of 1 km or less at
dataset
the edges wasof interpreted
the main anomaly by Cooper using
in the centrea new
of theContact-Depth
profile, and allmethod
the Euler[6],results
which were
only
compatible with this. The depth parabolas of the second-order Euler method made The
works for the magnetic anomalies from contacts and that requires pole reduced data. the
depths to the contacts
interpretation are obtained
of the Euler solutionsateasier,the location of thethose
particularly zero associated
values of the from first vertical
the small
derivative of
amplitude the data.atThe
anomalies the results
edges of aretheshown in In
profile. Figure 3b. Two
addition, profiles
varying were abstracted
the structural index
from this
slightly aeromagnetic
allowed dataset
the stability of theandEulertheir locations
solutions areassessed.
to be marked The in Figure
red bars3a.associated
The first
profile
with (marked
each solution‘A’)show
is shown in Figurethat
the location 3c, the
andEuler
the results of applying
solutions would move Euler todeconvolution
if a larger SI
to it are shown in Figure 3d–f. The Contact-Depth results provided
was used, while the blue bars show the location that will result if the SI were depths of 1reduced.
km or lessIf
at the edges of the main anomaly in the centre of the profile, and all
these error bars are large (for example, see the shallow solutions in Figure 3f, then thosethe Euler results were
compatible
solutions arewith
deemedthis. unreliable.
The depth parabolas of the second-order Euler method made the
interpretation
A second of the Euler
profile (markedsolutions
‘B’ on easier,
Figure particularly
3a is shown those associated
in Figure from the small
4. The solutions from
amplitude anomalies at the edges of the profile. In addition, varying
standard Euler deconvolution (Equation (2), Figure 4a do not really show well defined the structural index
slightly allowed
clusters, but the the stability
depths of the Eulercompared
are reasonable solutions to be assessed.
with The red bars
the Contact-Depth associated
results. The
solutions from the second-order Euler methods are slightly better clustered, but athe
with each solution show the location that the Euler solutions would move to if larger SI
depth
was used, help
parabolas whiletothe blue bars
identify showsolution
probable the location that(e.g.,
clusters will at
result if theof
locations SI15,
were
25, reduced.
and 30 km If
these
on error4c.
Figure barsTheareaddition
large (forofexample,
error bars seetothe
theshallow solutions
plots again helpsintoFigure 3f, then
identify those
unreliable
solutions are
solutions, suchdeemed unreliable.
as the shallow solutions in Figure 4d.

Figure 3. Cont.
Minerals 2024, 14, x FOR PEER REVIEW 6 of 8
Minerals 2024, 14, 393 6 of 8

Figure
Figure (a)(a)
3. 3. Aeromagnetic
Aeromagnetic data
data from
from Southern
Southern Africa.
Africa. TheThe flight
flight line
line direction
direction is is north–south,
north–south, thethe
line spacing is 1 km, the grid interval is 0.25 km, and the flight height is 100 m.
line spacing is 1 km, the grid interval is 0.25 km, and the flight height is 100 m. The profiles marked The profiles marked
A A(left)
(left)and
and B (right)
(right)are
areshown
shown in Figures 3c and
in Figures 3c 4a,
andrespectively. (b) Depths
4a, respectively. (b) to vertically
Depths magnetized
to vertically
magnetized vertically
vertically dipping dipping
contacts contactsusing
obtained obtained using the
the method method[6],
of Cooper of from
Cooper [6], the
which from which
Figure the
is taken.
Figure is taken. (c) Aeromagnetic data profile ‘A’ from Figure 3a. (d) Solutions
(c) Aeromagnetic data profile ‘A’ from Figure 3a. (d) Solutions from Euler deconvolution (Equation (2)) from Euler
deconvolution
are shown as(Equation (2)) areAshown
blue + symbols. as blue
SI of 0.25 and+window
symbols. A SI
size ofof 0.25 and
7 points waswindow
used. (e) size of 7 points
Solutions from
wassecond-order Euler deconvolution (Equation (7)) are shown as black dots with blue andasred
used. (e) Solutions from second-order Euler deconvolution (Equation (7)) are shown black
error
dots with blue and red error bars overlain. A SI of 0.25 and window size of 7 points is used. The
bars overlain. A SI of 0.25 and window size of 7 points is used. The depth parabola is overlain as a
depth parabola is overlain as a black line. (f) Solutions from second-order Euler deconvolution
black line. (f) Solutions from second-order Euler deconvolution (Equation (10)) are shown as black
(equation 10) are shown as black dots with blue and red error bars overlain. A SI of 0.25 and window
dots
size of 7with blue
points is and
used.red
Theerror
depthbars overlain.
parabola A SI of 0.25
is overlain as a and
blackwindow
line. size of 7 points is used. The
depth parabola is overlain as a black line.
 clusters, but the depths are reasonable compared with the Contact-Depth results. Th
solutions from the second-order Euler methods are slightly better clustered, but the dept
parabolas help to identify probable solution clusters (e.g., at locations of 15, 25, and 30 km
Minerals 2024, 14, 393 on Figure 4c. The addition of error bars to the plots again helps to identify 7 ofunreliabl
8
solutions, such as the shallow solutions in Figure 4d.

Figure 4. (a) Aeromagnetic data profile ‘B’ from Figure 3a. (b) Solutions from Euler deconvolution
Figure 4. (a) Aeromagnetic data profile ‘B’ from Figure 3a. (b) Solutions from Euler deconvolutio
Equation (2)) are shown as blue + symbols. A SI of 0.25 and window size of 7 points is used.
(equation 2) are shown as blue + symbols. A SI of 0.25 and window size of 7 points is used. (
(c) Solutions from second-order Euler deconvolution (Equation (7)) are shown as black dots with blue
Solutions from second-order Euler deconvolution (Equation (7)) are shown as black dots with blu
and red error bars overlain. A SI of 0.25 and window size of 7 points is used. The depth parabola
and red error bars overlain. A SI of 0.25 and window size of 7 points is used. The depth parabola
is overlain as a black line. (d) Solutions from second-order Euler deconvolution (Equation (10)) are
overlain as a black line. (d) Solutions from second-order Euler deconvolution (Equation (10)) ar
shown as black dots with blue and red error bars overlain. A SI of 0.25 and window size of 7 points is
shown as black dots with blue and red error bars overlain. A SI of 0.25 and window size of 7 poin
used. The depth parabola is overlain as a black line.
is used. The depth parabola is overlain as a black line.
5. Conclusions
5. Conclusions
Mineral exploration projects have many different facets, and the use of modern data
processing techniques can greatly facilitate the process. Second-order Euler deconvolution
Mineral exploration projects have many different facets, and the use of modern dat
has been introduced and applied to both synthetic and real aeromagnetic datasets. In addi-
processing
tion to beingtechniques
less sensitive tocan greatly
regional facilitate
trends in thestandard
the data than process.
Euler Second-order
deconvolution Eule
deconvolution
(because it doeshasnot
been
use introduced
the first orderand appliedoftothe
derivatives both synthetic
data), and real depth
it also generates aeromagnet
datasets.
parabolas, which further aid in the interpretation of the results. Adding error bars to thestandar
In addition to being less sensitive to regional trends in the data than
Euler
Euler solutions shows
deconvolution their sensitivity
(because it does to theuse
not choice
the of structural
first index and assists
order derivatives of theindata),
the it als
assessment of their reliability.
generates depth parabolas, which further aid in the interpretation of the results. Addin
error bars to the Euler solutions shows their sensitivity to the choice of structural inde
Funding: This research received no external funding.
and assists in the assessment of their reliability.
Data Availability Statement: The data used in this project are confidential and not generally available.
Conflicts
Funding: of Interest:
This research The authors declare
received no conflicts
no external funding.of interest.
Minerals 2024, 14, 393 8 of 8

References
1. Fedi, M. DEXP: A fast method to determine the depth and the structural index of potential fields sources. Geophysics 2007, 72,
I1–I11. [CrossRef]
2. Hsu, S.; Sibuet, J.; Shyu, C. High-resolution detection of geologic boundaries from potential-field anomalies: An enhanced
analytic signal technique. Geophysics 1996, 61, 373–386. [CrossRef]
3. Hsu, S.; Coppensz, D.; Shyu, C. Depth to magnetic source using the generalized analytic signal. Geophysics 1998, 63, 1947–1957.
[CrossRef]
4. Keating, P.; Sailhac, P. Use of the analytic signal to identify magnetic anomalies due to kimberlite pipes. Geophysics 2004, 69,
180–190. [CrossRef]
5. Hartman, R.R.; Teskey, D.J.; Friedberg, J.L. A system for rapid aeromagnetic interpretation. Geophysics 1971, 36, 891–918.
[CrossRef]
6. Cooper, G.R.J. The automatic determination of the location and depth of contacts and dykes from aeromagnetic data. Pure Appl.
Geophys. 2014, 171, 2417–2423. [CrossRef]
7. Cooper, G.R.J. A generalized source-distance semi-automatic interpretation method for potential field data. Geophys. Prospect.
2023, 71, 713–721. [CrossRef]
8. Huang, L.; Zhang, H.; Sekelani, S.; Wu, Z. An improved Tilt-Euler deconvolution and its application on a Fe-polymetallic deposit.
Ore Geol. Rev. 2019, 114, 103114. [CrossRef]
9. Huang, L.; Zhang, H.L.; Li, C.-F.; Feng, J. Ratio-Euler deconvolution and its applications. Geophys. Prospect. 2022, 70, 1016–1032.
[CrossRef]
10. Reid, A.B.; Allsop, J.M.; Granser, H.; Millett, A.T.; Somerton, I.W. Magnetic interpretation in three dimensions using Euler
deconvolution. Geophysics 1990, 55, 80–91. [CrossRef]
11. Thompson, D.T. Euldph: A new technique for making computer assisted depth estimates from magnetic data. Geophysics 1982, 47,
31–37. [CrossRef]
12. Mushayandebvu, M.F.; van Drielz, P.; Reid, A.B.; Fairhead, J.D. Magnetic source parameters of two-dimensional structures using
extended Euler deconvolution. Geophysics 2001, 66, 814–823. [CrossRef]
13. Philips, J.D. Two-step processing for 3D magnetic source locations and structural indices using extended Euler or analytic signal
methods. In Proceedings of the SEG International Exposition and 72nd Annual Meeting, Salt Lake City, UT, USA, 6–22 October
2002; pp. 1–4.
14. Cooper, G.R.J. Euler deconvolution applied to potential field gradients. Explor. Geophys. 2004, 35, 165–170. [CrossRef]
15. Marson, I.; Klingele, E.E. Advantages of using the vertical gradient of gravity for 3-D interpretation. Geophysics 1993, 58, 1588–1595.
[CrossRef]
16. Salem, A.; Ravat, D. A combined analytic signal and Euler method (AN-EUL) for automatic interpretation of magnetic data.
Geophysics 2003, 68, 1952–1961. [CrossRef]
17. Salem, A.; Williams, S.; Fairhead, D.; Smith, R.; Ravat, D. Interpretation of magnetic data using tilt-angle derivatives. Geophysics
2008, 73, L1–L10. [CrossRef]
18. Pasteka, R.; Richter, F.P.; Karcol, R.; Brazda, K.; Hajach, M. Regularized derivatives of potential fields and their role in semi-
automated interpretation methods. Geophys. Prospect. 2009, 57, 507–516. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.

View publication stats