4.

3W-31+3, 7 VIGS.
GROPHYSICS, VOL. XXII, NO. 2 (APRII„ 1957), PP.

AEROM AGNETIC
A •NEW METHOD FOR INTERPRETATION
PSEUDOGRAVI.NIETRIC ANOMALIES'

V. BAR ANOVf

• ABsTRAcr
The purpose of this paper is to describe a method of interpretation based
result on atransformation
of this transformation is
of the total magnetic intensay anomálies loto simpler ar.ontalius,
the climination of the disiort ion due to the obliquity uf the normal 1,•,Ignetic field, so that the resulting
ag11,111,,I bo,ties.
ill be located 140 the vertical of the disturbing m mn.onelie notential V ar.d
HM/Mil lieS
. The starting pon oC littz Clitutywhich can be written:
Newtonian potencial 1.1—relation
3. graPU fcr • V
with: —4
5 ...1 magnetization _
f .. 66 .7 . 10-9 o. CGS
a .. assumed density of magnetized bodies
relation may be considered as a partial ditlerential equat ion. The boundary condition con-
This . known at ttacit point P et the datum plane. As
sists la the measured values of the total ficld T(P)we can determine ibis function everywhere abov e
:T(P) is theplane,
derivative. of a harmonic function V,
the datum Solving that the partial ditIerential equation, we find the Newtonian potential U
and its vertical g, g`, g" • • • .
rínally, wederivati
obtainves
these quantities as functionals of the measured magulle artomaly T(P).
Por ínstance, one of the formulas is
. . . r
g(M) ... —
('5 II(M, P)T(P)dSp
• .
is the kernei of the tránsformaticn allowin g the direct computation of g(Af)—an .1
where nig, P)
anomalous field which will be called "pseudo-gravimetric
l the anornaly."
usual properties of a gravimetric anornaly.
Of course, the pseudo-gravimetric anomaly has ah
The 'field of g(M) prescnts no distortion and the interpretation becomes as eaay as that of a Bouguer
T
ano mal).
.maly perform the actual routine calculations of chis transformation, we start with the values of
T. s as for the usual computation of the vertical deriva.tiveS. The
taken un a rectangular or trigonal grid i calculations,
use of punched-card equipment speeds up considerably the e

I. INTRODUCTION

In applied geophysics,' the interpretation of results assumes two different

aspects, corresponding to two successive pilases. The first phase is a detailed
analysis of the data obtained by each of the methods applied. The second is the
syntheis of both geophysical and geological data. This second stage most of ten
concerns the'geologist: the geophysicist only points out the possibilities and dis-
cards the inconsistent assumptions, but the first preliminary phase is the geo-
physicist's concern only.
In gravimetry, the main problem of the preliminary step is the elimination
* Paper presented at the .A.amual Meeting of the Socicty of Exploration Geophysicists, March
1955.
Manuscript received by the Editor December 2r,
toss, New York, N. Y.
Compagnie Genérale de Geophysiquc, Paris, France.
360 V. ISA

of the regional anomaly. A1odern m•thods, comp,1111 ion of the vertical deriva-
tives r:■
f the field, are just ífiecl only because oí the r. !int, in magnetic
wo•k„the preliminary pitase hecomes VCry iMportani„
As a nyt te r of fact, the results of a magnetic survey are more difficult to inter-
pret than those of a gravimetric survey. While, in gravimetry there is a simple
connection between the residual anomalies and the tectonic features, in rnag-
netism the anomalies do not Cake place.at the apex of the disturhing bodies. The
magnetic picture of the tectonic occurrences undergoes a (istortion, clue to the
inclination of the magnetizing vector.
The di is further increased in the case of an aerornagnetic survey. Then,
insteaci of 1.he vercicai eoinponent of the 11C,(1, tne vanations ot the total magnetic
field are recorded, and this again increases the distortion of the map: the effects of
the obliquity of the recorded field are added to those of the inclination of the mag-
netization.
An additional difliculty crises at the final stage of the interpretation, when
one tries to check the gcological assumptions and to compare them, for instance,
with a simple typical case. In gravimetry, the computation of th¿ influence is
quite immediate; in magnetism, the computation is much more complicated and
a greater number of variable factors must be taken finto account, such as the in-
clination of the normal field, the orientation of the structure with. respect to the
magnetic mericlia:a, and so on.
These considerations show that, when faced with a problem ¿f magnetic inter-
pretation, it is advantageous, to begin with, to transform the isanomal map, in
.Order to suppress the distortion, and to make the mai) as clear as a map of the
Bougucr anomalies.
The purpose of this paper is to show that such a transformation is possible.
It is not more difficult than the computation of vertical derivátives, and can be
carried cut as a routine j'oh. The transformation is based en the fact that, in most
practical rases, the magnetic potential V is connected to a gravimetric potential
U. \Ve shall first establish this relation.
II. FUNDAMENTALS OF THE METITOD

The gravimetric potential is a Newtonian potential„ the general expression of

which is
fifí a — dv

wheref is the gravitational constant and 0- the density, varying, in general, from
point to point. The expression of the magnetic potential can be written as

V iff 1 .tirad
('
where i is the magnetization vector, which may be variable too. The only assump-
-›
tion that we shall make is lo suppose that the direction of 3 is the sante in all Mag-

Awa
 361
RPR ETA T ION OF A ERall A GN ril C 31 A PS
I NT 1,

llulii,ed mases. But. NVC are not obliged to asume that tris direction is the same
3 that of the normal terrestrial magnetic fiel! . ->
31, Computing the scalar product of the magnet iza3 ion 3 by the gradient 6f the
!ti a vimetric potential, we get the following relation:

, gra.d U -- fo-V.
i.
This formula seems to be known since Poisson's time. Let us recall that the
founder of applied gravimetry, Eütvos, used Poi3=3-son's theorem 'Lo deduce two
omaly, and the
famous relations bet•een the three componen ts of themagneticen
,5Ion ".::1,113a-3,7% 1.3311- now. the 1,vos frmulae are only of
va i ltes recor ded ,,,, a 1.0 two potenti a ls may b e U se d
historical interest. Nowever, the relation between theransfo rmation of the exp eri-
lo deduce more practica' fornaulae, allowing direct t
mental magnetic anomalies into simpler anomalies, which we propuse to cal'
"pseudo-gravimetric" anomalies. _ ci.
.o pol_k_no...w . .the_true..density
There are two reasons for this name: first we 1 ...U. The un-
Thcoridly, oply the pagnetized masses contribute to the. potential :are. not..the
the anomalies deduced..from U.
magnetized rocks do not show'. Thus.
trate zravimetric anomalies: theyare still magnetic anomalies, but cornputed on
...... . . . —_—__.
llie J.a113_221kil
___. -. _
that the magnetization vector is vertical. To be perfectly citar,
- —__._.___________—___
_-
we cal. them "pseudo-gravimetric anomalies." ouguer
U are . as simple as B
The essential fact is that anomalies deduce(' from and do
anomalies. They are located on the vertical, of the magnetized
not depend on the inclination of the normal field, nor on the direction of the
magnetization. Thu.S the interprétation'and all the computations become very
simple,as ese factors no longer need to be taken in account.
th
e, we sha.'" select a conventional clensity, by writing:
There or

v
4 the unit vector representing the direction of 3, we obtain:
Then, cálling
-+ dU (2-r)
v grad U = =V
dv
derivative of the pseudo-
which menos that the magnetic potential is the obligue ati
gravimetric potential, in the direction of the magnetizon.
III. TRANSFORMATION OF TITE VERTICAL MAONETIC COMPONEN?
Let us now take the upward vertical as the Oz axis, and dilTerentiate the re-
lation (2-r) with respect to z.
dg (3-1 )
=Z
dv
of the magnetic anomaly is the derivative of the
i.e., the vertical component Z
'7471)
 367 V. BA RANO V

pseudo-gravimetric anornaly, that we shall call g, as in gravimetry, as there is

no risk of confusion.
Before dealing with our main problem—the transformation of the anomalies
of the total force--let us consider the equa.tion (3-r), which we propuse to inte-
gra te. •
I.et us consider (Fig. r) a straight-line (D), with its origin at and extending
towards infinitv, eitlrer in the direction of the vector ú or in the opposite direction,

./(
1»

FIG. I

but in such a manner that (D) will not cross zones where magnetized masses may
have a chance to be found, More specifically, let us consider the northern hemi-
sphere. The unit vector 77, then, will, in most cases, be directed downwards. We
• sha!l take for .(D) the half-line with direction opposite to v It will be in totality
aboye the earth's surface. Let us call Q any given point of this line.
To obtain the function g(AI) from (3-1), we need only to integrate Gris ex-
pression along the line (D). We get:

—g(Af) f Z(Q)ds
o
where s is the length of the variable sgement 2/Q. This is an inverse operation
with respect to the derivation in the direction of the vector r. Of course, we do
not know the value of the anomaly Z(Q) on line (D). \Ve only know the values
of. Z measured on the plane (ir).
 C 1.1.4 P
' A ER0111:1
INTERPRETA TION horizont,a,
Q aboye
Z at every 1' is a point o,
s possi1) eto compute f
However, it the well known upward continuati,,n orm
ula •

Mane (r), using is given by the expression:

the plane, 2,*(Q) a rC (p),1 P
(Q) Q17
21r
is a surface element, the integrat ion c(..Gnprising of
theintegrations,.
whole arca o',we•
where dS p Using this value óf Z.(Q) and changing tbe order
plane (Ir).

ohtain: ít
ds
¿iz'■
QP/

Let us put in eviclence the function

K(M, 1') o i)cs ()P

is. the, kernel of the linear transformation

This (3-3)
'g(3.1)
21r J
r K(M P)Z(P)dSp

Z to the
allowing dirima passage from the anomaly of the verticalaboye the plane (7r).
component
any giVen point
g(M) at . is the derivative, with respect
pseudo-gravimetric ancanaly K
The formula (3-2) suggests that the kernel P) and. which will be
to z, of an harmonic function that we:shall denote by fachf,
useful hereafter. Thus we shall write (3-4)
aic(M, l'yaz
— K(Af, 1)
Iioweve, we cannot invert the two operations—integration and derivation with
respect to z—because the integral denling with the reciprocal of the distance.
J/0 fi is divergent. To remove this difficulty, we shall substituta to this function

the following clifterence:

s so
is an adequate positive constant. If now we take the clerivative with
respectl
where so to this variable, the additional terne, independent from z, disappears,

while the integral

ds
o s so PQ
364 - r tANOy

becoutes alsolutely conv ,ent

o(
and the differentialion under the integral sign is justified.
' final int cual ion is an clementary one. .Ve easily obtain
u Log (r 77) Log aso

where - -- ,11- /) is the tector from M te P, and

= + n2 I

\Ve chose. th? of coorilhiates Oxyz on the plane (ir), so thai noint M, of
elevation z, is 'en the vertical axis. The coordinates of point
1' are t.: ri o.
Last, to get K, we need only te take the deriva tive of u with respect te z. \Ve
have:
- I z ur
K(111, = 44 (3-6)
r vr

where u is the cosine of the acute angle between the vector -v4and the axis Oz.
\Ve can easily check that n(M, P) are two functions the ',aplacían of which
is zero (except on 1)),
The formitlite (3-3) and (3-6) completely.solve the probiem of the transforma •
tion of the anomalies of the vertical component into pseudo-gravirnetric anom-
alies.
TRANSFORMATION OF THE TOTAL FORCE
We Shall now consider the anomalies of the total magnetic force T.
--"I the unit 'Vector in the direction of the normal terrestrialfield.
Let us call y
The anomaly 7' is the derivative, with the inverse sign, of the magnetic potential
V, in the direction
dV
T
dy
Thus, taking in account the fundamental relation ( 2-1) we have
d2u
dydv

Let us
now compare this relation with the (3-r) formula. It can be noticed
Thus, wíthout repeating the
that, Itere, g is replaced by —1111/dy and .2" by T.
whole reasoning, we can use the (3-3) forninla and write
dt.1
--= 27r f f (M, P)T(P)dSp.
dy
 G.A'ETIC ,51 I'
ET TION OP .1

the usual ir y ltiri i sl

Alter difTerentia ting with.respeet tu z, we get,
13'1
fig
....... (. . . . . . -- ........ T(P)dSp.

dy ,I, - -Ir
of course, we must integrate again álong the half-line the direction
obtain- g,
of which is, now given byTfins .• of v. We > lall gel then
» ./Thi\
g(111) = \dy/
o

Let us use the expression of the derivative in the preceding formula,

' after sub-
1 the orden of the integrations, we
stituting the variable point Q for M. If we joyel-
obtain: ¿no. P)
f T(P)d.Si, -----
g(A1), .1 o
211-
Wc shá put K(Q, 1')
(Is (4-2 )
11(111, 1') =
clx

= z j(* f P)T(P),ISe. (4- 3)

27r
This latter formula maíces it possible to pass from the anomaly of the total
force to the pseudo gravimetric anomaly. The kernel of the tra.nsformation
is given by (4-2) where the integral is extended along the haif-line of
11(M; 1')
direction while K depends on v.
It is known Gua in practico (Me can very of ten assume that the direction of
the mágnetization of the underground is that of the normal terrestrial field. 'Tus,
y and tj'iare idcnticai. Our aim was to show
we can assume that the unit vectors 7 ,
that this simple assumption is not indispensable to carry out the transformation
Ifowever, hereafter, we shall put 7=77, te simplify the theory and to short en the
comptitations.
us put
Before . computing the kernel 11, let
av(M, j›)
= (4-4)
11(A1,
az
is the derivative with respect to z of a
for the formula (4-2) suggests that 11
But, before integrating with respect to s, we
function that we call v(M, P).
P), so that the kernel may
must Subtract the principal part (s+so) from K(31,
be integrated. \Ve get
v(M, 1') = rK(.11, P) iltt(211, 1'). (4-5)

The anotnaly is considered as positive when it increases the recorded value.

1
 V. BAf lNO t'

expre,ic.lon uf the kernel )7(.3t, e)

can be obtained in differentiating with
resti.T_t lo z as i:uicated in (4-4). Ifowever, we shall not need this general expres- -.
si,a, as the only thíng we are concerned with is the anomaly in plane z=o. We
thus consider v.rhat become the kernels K.and 1/ at the limit, for z =o.
It \vill be conveliient to choose the axis of coordinates so that the Ox axis will
be directed towards the magnetic Norih, and to use polar coordinates, putting
E = p cos , p sin w.
Then, if we call 1 the inclinado') of the normal fieidy we shall have 1.i=sin I. For
practical purposes, we shall wr.ite X= cos I. We d'en obtaint

r 7.-7= z2 ;i2; i hp cos

and therefore

u = Logr-Vp2 z2 pz -1- Xp cos w j

and

K
+ Z2 + z
P2 VP2 -i-
- 7, 2 + Xp cos
For z=o, diese t o functions become
..r .
u = Log p Log(14- X cos CV)
(4-6)
and
1.1
K= — (4-7)
p + X cos
The kernel K is the product of the factor rip by the function of w abone

Li (w) =
+ X cos w (4-8)

The factor I/ p is iiiiinite for p= o, tiros K presents a singular point at the origin,
However, as dS=p• tlp• dw, this factor disappears under the sign sum (3-3) and
,the integral

g(o) = 27r J f Z(p, 0.›)14(co)iip • tho (4-9)

giving the value of the pseudo-gravimetric anomaly at the origin of the coordi-
untes includes no singular element.
Let us now consider the integral (4-3). Differentiating (4-5) with respect to
and taking in account (3-4), we get
z (9,u
11 =--- Al r K.
r az, ( 4 10)
 367
NTERPRET ATION .OP A ER0:11 A GNE'll C M.1 PS

uted . To
The first term is null for z=o, the last one has been already comp rmonic.
%Ve S1101.11d .remember that u is
ha
compute the second derivative of
Therefore,
a2u I x(x cos co)
a211 a au
— Op p U • p ( 1 + X cos 0))'
Dz' Op

is oblained inimediately from (4-6). At last, we have:

which
( 4-11)
11(p, o)) x ---5)2(co)
p

with
(X' — p2) + cos (4-I'2)
111(w) .._,
(r + X cos co) 2

and hence
(4-13)
f T(p, c.))9.2(w)(1ptioa.
g(o) =
2r
This integral enables the computation of the pseudo-gravimetric anomaly at 'f.the
origin of the coordinates: it is functional of the anomaly of the total force

V. DEVELOPED FORMULAE OP THE XERNELS

For the actual computation of the'Mtegrals, we shall be obliged to develop
122 into Fourier's series.
the functions K, H, L and-
Let us start from the well known series
— a'
2 y (_a)b cos kw
-I- 2a cos -I- a' k:1

and write
r z bz I
with n
•a
X.
The developrnent of (3-6) will be easily found:
2 ao r zy°
K (— (--- ) os kw.
r r

Then, as
a
(rK) p,K
az
ji
3o8 B.4 R ANO V
and

az
(rK) E - n) 9e COS kco
r PI
we shall get

1.4
11 .-- — -- -- - --E (-77) 4 ------- (k + 1.4) cos kw.
r r k.„./ ( 5- 2 )
p
Tí we make z= o and r=p in (5-1) and (5-2), we shall have
0(1

L I (co)= 1 + 2 E (-77) k cos kco

k.. t (s-3)
and

22(W) — 2 E ( --7/) k (k cos (5-4)

These series are absolutely convergent as, in general, n <1. The case of the mag-
netic equator, where i= r, must be considered separately.

VI. COMPUTATIOn OP THE VERTICAL DER I VATIVES

Our next problem is the computation of the vertical derivatives g', g", • • ,
directly from the recordeel anomaly Z or T. It is well known that diese derivatives
are more convenient for the interpretation, as they more clearly point out the
various anomalies, the separating power of the clerivatives being greater. We shall
seo, further, that they are easier to compute, as the convergence of the integrals
appearing in the computation is more rapicl.
Let us first consicler the meaning of the vertical pseudo-gravimetric gradient
g'(x, y) (the accent indicates the derivative with respect to z). The fundamental
relation (3-r) shows that the vertical cornponent Z is the obligue derivative of
Let us assume that the geological structure, cause of the anomalies, is in the region
of the magnetic pole. Then the vector iris vertical; the derivative with respect to
is identical to ag/c9z. On the other hand, al the pole, the component Z is identical
to T, and, thus
1
›, 51
ag T(poieb
aZ

l'hus the pseudo-gravimetric vertical gradient is nothing else than the 7_7:_..a.,gtet_ic_ •
anomaly reduced lo the pole.
1,et us suppose that bite whole geological anomaly is carried over to the crea 7 .4
of the magnetic pole, or, in other words, let us replace the actual magnetic 'z
which presents some inclination, by a vertical one. The results of measurements
of the total field, performed under such conditions, would precisely give the
pseudo-gravimetrie vertical gradient.

14
 3.6 9
INTERPRETATION OF AER0.11.4 GNETIC :VI APS

When cornputing the vertical derivatives, une must keep in mirad that the
kernels are no longer continuous at the ()right of the coordinates, This singular
point, thus, should be isolated by a small circle of radios e, to separa te the plane
y in two regions: the inside and the outside of this circle. Instead of (3-3), we shall
write:

g I,(z) ff
27r P>o

/,(z) being the integral extended to the area p<e. The integral (4-3) 'will be
similarly handled.
As to the field Z (or T) in the small circle, it is harmonic, thus analytic, and,
accordingly, can be developed into series such as
Z A+ Cn +•••

or, in polar coordinates:

Z = A + (13 cos C sin cks)p 0(p2).

We shall use ibis expression and that (5-r), of K, to compute I,(z). Integrating
first with respect to we get
2r
KZcica = 7113 ---- 0(p3 )
27r o r r

Integrating then, with respect to p, from o to

e2
z
It(z) = f f KZprIp-(10)=- A(r z) — nB(— zr + 2)
2
2r P>e

wliere
r = \/(z2

1f we put, how, z= o, the value of I,(z) is A. e+ 0(e2). It tends toward zero with E.
13ut, computing the derivatives, for z = o, we have
1.'(0) = — A -I- nBE + 0(0).

and
1-
A
I ." (0) — 27113 0(e).

Thus:

= — A -I-- ff K'Zpcip• dco

7r .
370 V. N A 1?..,1 NO11

and

t1 2 ,ri B
K "Z PdP • dee,
p>

The accents still indica.te the derivatives with respect to -z. \Ve have the
to differentiate under the integral sign, after eliminating the singular eleme:,t,
To compute the deriva tive K', we only need to take the second derivative oil u,
which we have already computed. We find:

= = .1: 2(w) (6-t"t

Pa

with

L2(6)) = f = 277 COS •–• 4112 COS 24) + 6113 COS 3CO — (6-2)
(1 + X COS w)'

The same rnethod, applied to K, gives

1
"= Ls(co)
Pa

with
L$(0)) = x + 6/72 cos 2W cos 3(.4.1 + • • •.

It is remarkabie that the (6-2) series has no constant term. We can thus write the

final formula as foliows:

g' = Z'(o) r f z ( p, dp do.) (6-5)

211"

and we know that the integral comprises no singular element.

As for the (6-4) series, the missing term is that in cos W. Therefore

2r
— L3(CO) COS Wthx) = O.
2T

Let us now consider a constant anomaly, equal to A abd compute the

integral for the whole area outside of the small circle. We have:

dp
f Ls(co) — deo =
2-7r >
2

This value compensates the identical, but of opposite sign, term in r(o). Thes
the singular element is completely eliminated and we can put o, removing the

circle which we excluded at first. We finally obtain the foliowing formula:

 I NT ERPRET A TION Oh A ERO . 1G C Al APS 371

/ Z
g 277
[Z(p, .7(0)113(w) rlw (6-a)
Ox 2

for the second pseudo-gravirnetric derivative.

The same method will permit to compute the de/ ívatives.g', g", • • as fu nc-
tionals of T. It is not necessary to repeat the reasoning; it suflices to observe that,
when passing from K to II, the first terco of the Fourier's series is multiplied by
—ji, the second orle by —(r +,2). l'hus the constant A must be replaced by —,L2A
and 13 by — (x-1-11).B. The final formulae are as fol;ows:
dp
g' = p,T(o) f f T(p, ce)9,3(w) — <fu.> (6- 7)
27r

and

g 2A k
\ux
ag 271'
f tPj u, - 7 (, )124(.d) 2 /07, oN
k s0
p
with

S/ 3(w) = 2 E ( —n)kk(k 4-1.4) cos lew (6-9)

k-1
and
o
94(W) = µ --• 2 (_„),(k2 x)(k cos kw. (6-ro)
k-1 .
The transformations may be just as easily obtaincd for perrnitting one to fonn
the derivative directly from Z or T. We shall not elaborate, in Order to save
space.
VII. PRACTICAL FORMULAE FOR THE COMPUTATION OF THE PSEUDO-GRAVIMETRIC
ANOMALIES—INTEGRATION WITII P.ESPECT TO co
Expressions such as (4-13), (6-7), (6-8) are the theoretical formulae giving
g, g" as functionals of the magnetic anomaly T. Oí course, it is necessary to
find some simple, accurate and rapid means for the carrying out the practical
computation of these expressions.
- A very-large humber of methods can be imagined. The most convenient seems
to be the use of a "canonical grid:" It is the method commonly used for the
computation of the vertical derivatives. Its main advantage is that it does not
- entail numerous readings on the map, and lends itself easily to computation,
through use of punched card mochines. This-is a well known method for geophysi-
cists, so we do not need to go into details.
The usual square grid, or a trigonal grid—such as that illustrated in Figure 2-
may be used equally. The method is the same, whichever be the data to be com-
. puted: g, gt or g". We shall therefore consider only one of these formulae. For
 372 V. BARANOV

FIG. 2

instance, let us consider the expression (6-7), for the reduction of aeromagnetic
anomalies to the pole. It first contains a constant term —,aT(o) which needs no
computation (it is read directly on the map) and a double integral:
dp
I — T(p,w)123(w) — da)
271-

covering the whole plane of measurements.

\Ve shall first integrate with respect to w, putting
2r
T(p) = f T(p, u))23(co)du). (7-1)
27r o

Then, we shall integrate with respect to p

00 dp
I= T(p) — • (7-2)
fo
this is not an absolute requirement, It would be possible, too, to
integrate first with respect to p, this would require some precautions. In fact, it
is the only possible method in the case of an irregular grid, for instance a rectangu-
lar or star-shaped one, which is sómetimes more advantageous. But, in the
case of a trigonal grid (Fig. 2) considered in our example, it is more convenient
to integrate first with respect to w.
The function 523(w) appearing in (7-i) and acting as a weight function can be
developed into Fourier's series (6-9). T(p, u)) is an experimental function known

A(1)
 INTERPRETATION OP AEROMAGNETIC MAPS
373

The six azimuths are equally

at six canonical points on the circle of fixed radius p.
spaced; their directions are detennined by the angles

with v=-- o, 1, 2 3, 4, 5.
9

T ,,(v) It is obvi-
We shall represent these values read on the grid by the notation
a
ously a periodic function; it would thus be natural to develop it, too, into
trigonometrical series. However, we propose to take a slightly different approach.
Let us put:

T(p, w) = T p(o)lfr(co) T p(1)51/ (co — 1-)

3
57r)
T P(2)111(41 " • • ± T P(5» (°)
3

«co) is an influence function of the point of zero azimuth, such azimuth being
directed toward the magnetic North. For w = o, its value is it is zero in any other
point. These conditions are satisfied for an infinite number of functions. To make
T, the
a choice, we shall introduce a "simplicity condition." Like the anonaly
influence function must present as few higher harmonics as possible. This condi

FIG. 3
-74 y, BA IZA NO

is in accordance with the filtering qualhies uf the canonical grid.

We hall thus put:

Vi(co) - I- Ces w)(2 C — 1) ( 2 cos w

(7-3)
1 1 1 , 1
--- - — COS 1,/ + -- COS w + -- COS 3w
6 3 3 6
because this function is equal to zero for cos w=-- —I, +.1i and —a, while for
w ,,--, o, it has the val-de 4,(o)—, t. This is the simPlest influence function satisfying
to the foresaid conditions indicatecl aboye. Figure 3 shows the pattern of this
function. I
As to the other azimuths, we shail compute:
I I yr i 21-ir I
,-. --±— cos -- cos w + — cos — cos 2W 4- -- cos vir cos 3w
(C° 3 6 .3 3 3 3 6
+ the odd tcrms in sin w, sin 2w, sin 3w.

(we do not need to express these odd ternas, as the function f is an even one),
The Fourier's coefticients of these functions are tabulated hereafter:

Y 1 COS U) cos 2 (O cos 3 w

o I I I-
1 1 ic —/ ~. 1
8
2 i « k `t I
3 1 —$ i —i
4 i i
6 «'' :J 1

5 k i —1 —i

We can now develop the given function T(p, w) into a Fourier's series on the
circle of radius p. We obtain the following expréssion

T(P, w)

[TP(0) + TP(1) + TP(2 ) + TP(3) + + TP(5)]

• — Tp (o) + T p ( r ) — Tp(2) 2 TP(3) --- TP(4) + TP(5)1 cos

6

• -1—.[2Tp(o) — ./.'p(r) T P(2) + 2 TP (3) — TP(4). T P(5)] cos 2«,

6

+ [Tp(0) Tp(I) Tp(2) — TP(3) -I- TP(4) Tp(5)] cos 3(.0

6
• odd tercos
 INTER1'.RE7 , (TIILN'

, ear v.)11en integrating with

(it is useless.to write these odd terms, since they disap!
respect to
To coMpute the integral (7-t), we have to inicate the prn(iuct oí two lune-
Vonrier's serius (5-9) and (7-4).
and 113, given by their developments of
tions T . .f the weight f netion as foliows:
To be more general, let us write the developmen (
271. eus 2y2
1 - 9(() 4

For the funclion 11a, the values uf the coefilcints are:

'Yo = o, 71' 11( 1.

72 1.z)
'Ys 37N3 * 11)

For the integration, we only need lo sum up ihe products of the coefficients of
We obtain an expression which can be conveniently rea.rranged
(7-4) by the 7s.
according, to the canonical values of the anoinaly:
(7- 5)
T(p) ao Tp(o) ai Tp(i) ± • « • -17 as Tp(5)

in order lo show the numerical coefficients by which the readings from the grid
must -be multiplied. The expressions of these coeflicients ivhich depend no the
inclination of the- normal field are the ilollowing:

— t'Yo 'r• 2111 2Y2 'Ys)

ao
6.

al 6

a2 «4

x
as — ('Yo -I- 2"Y1 -1- 272 — 'Ys)
6
2, 3, • • •,
may be easily computed for the values p = 1,
T 1msthe function T(p)
corresponcling to the radii of the circles passing througli the vertices of the grid
illustrated on Figure 2.
VIII. INTEGRATION WITH RESPECT TO p

We must now compute the integral

dp (S- r)
= f T(p)
o
376 A RA NOV.

wherc• the function equal to zero at the oriyin, l known at points of abscissae
p.-- 1, 2, 3, • • • . Numerous niethokls are available to effect the approxImate (load-
raturt.°.s. \Ve sha!l outline here a inethod while perhaps net the most
accui ate, is very widely known, and the casiest describe. The axis Op wili be
divided irito partial intervals:

(o, -- (7, 3) -- (3, 5) and (5, ce).

We shall adrnit that, in the first intuya] (o, I), the function -7(p)/ p may be repre-
sented by a parabola

T(P)
A Bp + C p2

passing through the points p= 1, 2 anca 3. We shall have:

A + L.B+.1c. jr-irdp
2 3 o P
A -1- 13 + C '1 .

A + 213 + 4.C•= — 1-12

A + 3B + 9C = T,
3
where n=T(k) Eliminating A, F3, and C from these four equations, we get:
. .
T(p) 3 ', ,
dP 2 "-T- 1 3. (8-2)
O" P 4 6
For the intervals 3) and (3, 5), we sha]] apply Simpson's rule, which
causes us tu put:
3 dp II I
j 11(P) 3 T2 + -
3 T1 + —
=— Ta (8-3)

fs T(p)
P 9
Ta -2- T4 —I--
3 75
Tó. (8-4)

Finally, the last interval (5, co) corresponds.to the "tail" of the curve T(p) , where
tlie function can be approximately ex-pressed•üsing tlie following formula:
A
P 133

In fact, the average weighted T a.nomaly decreases like the square or the cubo
of the distance. To compute the constants A and 13 we'Shall take lato considera-

749
4
 INTERPRETATION O AEROMAGNETIC MIPS 377

tion the fact that the curve passes through the points p=.1
p.=5. Wc shall

then get thrce equations:

A B dp
+ Y()) —
50 375 A

A B
16 64
A B
I b
25 125

wlience, eliminating 4. and E:

n 04 tin 22 7
T4 —6 TB. (8-5)
jT(P) —
p '

Summing the expressions (8-2), (8-3), (8-4) and (8-5), we get

7 37 —
r dp 7 1 - 7
78 — Y4 -- T5.
jo 7.17(19) -3- TI — 72
30

To check this formula, we chose a theoretical function

a2p
(8-6)
= ( p2 a 2)Z/2

The exact value of the integral is equal to unity.

The approximate values, computed with the (8-6) formula are shown in the
table below:
1 Error
a •
9.2%
2.3o/o
r.0006,
2
1.0227
3
0 9950 —0.5%
4

is very sharp at the origin.

For a small value of the constant a, the curve '
With increasing a, the error rapidly decreases.
The integration. may be made more accurate, through separate consiaeration
of an additional partial interval (5, 7).
In any case, the result of the integration with respect to p is represented by a
linear expression
dp
7k
Pk ( 3-7)
f . 7r(P)
o (k)

where the 13k are the niinterical coefficients computed once for all. This formula
shbuld be combined with (7-5) to give the final result:
3 73 V. BARANOV

I = E cr,131Tk(7). (8-8)
k

we must not forget to add the constant term — pTo appearing in (6-7).
Figure 4 shows the values of the coefficients at their respective locations.
The process just described is well suited for dips ./>30° of the normal mag-
netic field. For smaller dips, near the magnetic equator, the computation requires
more tare. In all cases, the computation practically involves the reading of the
values of the fie]d T on a canonical grid, followed by a cumulative multiplication

410
e e
11)
O OOO O 111) 41> NORD

e
11>
•
FM. 4

of the readings by numerical coefficients computed once for all, and which de-
pends only upon the mean average geomagnetic latitude of the area under con-
sideration.
Of course, this computation is time consuming, if one uses ordinary comput-
ers'. However, it can he considerahly speeded up with the use oí electronic com-
puters operating with punled cards. The \ aiti( z, uf the 1ic:(1, icad . , n a point uf
the canonical grid, and the coordinates of this point (obligue if the grid is tri-
gonal) are plotted on "mark sensing" cards (one card for each point), transformed
into normal cards by the reproducen.
l'r? A first list is then given by the tabulator, a list that is compared to the read-
ings. This operation is designed to simultaneously check the readings, the plot-
tings and the punching. The computation itself is carried out in the computer.
 379
INTERPRETATION OF A EROMAGNEfIC MAPS

Finally, the cards pass again through the tabulator, whia prepares the final
list of the values of the transformed field.
grid with the numerical coefficients for the
Figure 5 gives an example of a
in an area where the average dip is
reduction to the pole of the field T
-
FORMULA AND COEFFICIENTS TO COMPUTE THE PSEUDO
-GRAVIMETRIC VERTICAL GRADIENT
dp
1 dw
T (p,w) D- 3 (U))
—g.= -sin I. T_ + u 227t

3 cos I + cosi (2 +cosz I) cos

= sin I 3 M.N
( 1 +cosi. cosw)
083

0011 :04 401

,
1 / 1W+::

.1mozgeri
.4111

1
.00,8

4
0%

00
0040wk
* 41
role.■
WOsseigagl
IrsiN
4014earbrogo%
FiG.5
h, shown on
,1 ', y cplIere huried at a depth
Figure 6 shows the anu.naly cau,e are otte e
-the map. On the left the contours of the total field anomaly ver plal gradientthis
- tic
right side, the field of the computed pseudo-gravimetric
represented. The computed curves are almost perfect circles, and their common
center on the apea aboye the sphere. The two externa1 curves are very slightly
affected by the unavoidable errors of the numerical computation.
Figure 7 shows the effect of a vertical prismatic body, of rectangular cross

Aen
 V. T,' A NO V

Fields due to a sphere

M.M.
Total magnetic íntentity I . 60? Patudo- gravirnetrie gradient

17( 29

//

FIG. 6

section. The computed psuedo-gravimetric field, on the right of the figure, has
a svmmetric appearance, whereas the total field anomaly is shifted with respect to
the magnetized mass.
Thc reduction of the field T to the pole seems to be a very advantageous trans-
f,•i Llr the
i) the qualitative interpreta tion of the transformed map is easier and more
immediate, as the anomalies are not shifted as a result of the obliquity of
the normal field;
2) the comparison of the results of a survey with a geological map is much
easier. One does not need to take into consideration the shifting of the
magnetic anomalies;
3) the transformed map can be compared to a gravimetric map, much better

leo;
 1
INTERPRETATION OP ARROMAGNETIC MAPS
381

flan the original acromagnetic map, especially if this latter too is trans-
formed into a map of the first vertical derivatives. intleed, the two quanti-
ties, then, are of the same nature;
4) the pseudo-gravimetric anomalies do not depend upon the inclination of
the normal field, nor on the orientation of the tectonic structure. This
simplifies the quantitative interpretation and the evaluation of depths.
The number of typical cases, such as blocks, faults and so on, is consider-

FIELDS DUE TO A PRISM OF INFINITE

LENGTH WITH VERTICAL SIDES

M N.

TOTAL MAGNETIC INTENSITY PSEUDO-GRAVIMETRIC GRADIENT

I= 60?

FIG. 7
 38 7J,f ls`rtNO11

ably reduced. Once computed, a single typical a.nnmaly can be used for all
dlps and orieblit-ttiOns;
5) 1 he coMputation of the theoretteal pseudo-gravimetric anomalies is sim-
ph,r (han that of the magnctic anomalíes. For instante, let us consider a
vertical Wad:. In magnetism, the vertical walls of the compartment give
vise toa very complex anomaly, which is eliminated if the structure is
loca ted a t the pule. It is easy tu draw a number of simple nomographs with
wh telt the cototsUtation of the pseudo-gravimetric anomalies becomes very
rapid.
The approximat e computation of integrals suela as (6-7) and (6-8) has made
some pr(-) ■
!,ress since 1955, so that a better mcthod now exists than the one de-
scribcd ¡II t he paper. In particular, for the deteimination of the numerical co-
efficients uf the transformation, it is no longer necessary to use a Fourier develop-
ment of the kernels.
ACKNOWLEDGMENTS

The theoretical Nv o r k and the numerical computations recorded in this paper

were carried out on account of, and in the Research Labora tory of, the Compagnie
Générale de Geophysique.
• The author wishes lo express his sincere appreciation to bis General Manager
Mr. Leon Migaux, for allowing hitn to publish the aboye results, and lo Mr. E. G.
Leonardon, for bis hclp in putting Ibis paper in irs final form.

DISCUSSION OF V. I3ARANOV'S PAPER, "A NEW METHOD OF

INTERPRETATION OF AEROMAGNETIC MAPS, 1 PSEUDO-
GRAVIMETRIC ANOMALIES"

JAMES AFFLECKt

Baranov?s pseudo-giavimetric anomaly method is based on well-known

relationships and his development appears to be sound. Applications of bis meth-.
ods sheuld be of very appreciable use in evaluating aeromagnetie surveys. The
comMents which follow shoulcl not be interpreted as criticisms. They are intencled
to .emphasize the conditions and limaza:10ns under which Baranov's •techniques
-might be utilized.
Conditions which trust: be recognized are these:.
1. The direction of magnetization is not necessarily in the present direction
of the earth'S field. This is-recognized by the author.
2. The direction of magnetization may be very much different for adjacent
anornalies.
.* Receivéd by the Editor October x3, 1956.
j Cutí Research and Deve.,,pment Company, Pittsburgh, Pennsylvania.

ti I
 1 NTERPRETATION 01? ,I.EROMAGN ETIC 1 1A PS 383

Magnetizations may be non-uniform within a rock mass. Graciational

rnagnetizations scem to be common.
When the three items aboye are considered, it is obvious that the pseudo-
. gravity anomaly peaks will not necessarily be coincident , with the apea or center
of the magnetized mass. Lack of uniformity is undoubtedly more common in
magnetizations than in densities.
The author does not imply . that his methods can .be used quantitatively in
conjunction with observed gravity over any area. It is likely that such quantita-
tive comparisons will be inferred, and they play be of appreciáble value in sorne
instances. But here again the same clang,ers are present. The pseudo-gravity
anomalies certainly represent the effects of the magnetic rocks only. observed
{,yavity anomalies renresent the cornbined effects of all rocks, magnetic and non-
magnetic. A subtraction of the pseudo-gravity anomalies from the true gravity
anomalies might then yield the effects of the non-magnetic rocks only. This would
be true if the assumptions held rigorously and if the conditions listed in Items
and 3 aboye were not applicable. If they are a.pplicable, d'en the difference map
might be misleading. The effects of angularity and non-uniformity of magnetiza-
tions might be mistaken for sedimentary structure.
The author deals with the pseudo-gravity vertical gradient. 11k definition
might be reworded thus: Tite pseudo-gravity vertical gradicnt is the equivalent •
of what the vertical magnetic intensity would be if the rock mass viere magnetized.
vertically. If considered strictly in this sense, then it would be useful in eliminat-
ing the angularity effect of non-vertical magnetization and non-vertical positton
of the sensing element of the airborne inagnetometer. Tite dírection of the ele-
ment of the magnetometer is probably sufficiently well known, but the direction
of magnetization cannot be generally assumed, sottihat the limitations must be
considered here as in the case of pseudo-gravity. •