THE UPWARD CONTINUATION OF ANOMALIES IN

Downloaded 03/12/15 to 169.230.243.252. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

TOTAL MAGNETIC INTENSITY FIELDS*

ROLAND G. HENDERSONt AND ISIDORE ZIETZt

ABSTRACT
In airborne magnetometry consideration must be given to the necessity of multi-level observa-
tions,~The problem of computing tntal intensity anomalies from data observed~on !ower !evels is
investigated in the case of contours exhibiting both two- and three-dimensional features. It is found
that even fairly complex anomalies can be extended, with results differing but slightly from observa-
tions at the higher level. Maps for comparing computed and observed anomalies are presented. The
mathematical basis is given together with numerical formulas and procedures for affecting the com-
putations.

INTRODUCTION

The advent of the airborne magnetometer, which makes it possible to obtain

magnetic data on one or more flight levels in a relatively short time has revived
discussions on the value of multi-level observations. Most geophysicists agree
that it is theoretically possible to continue fields in regions devoid of sources;
however, many question its practicality when applied to observational data be-
cause of the inaccuracy in or the scarcity of the latter, or because of the difficul-
ties inherent in representing it mathematically. Heretofore investigations of the
field-extension problem necessarily were confined to distributions computed from
simple geometries. Recently aeromagnetic surveys of several areas have pro-
vided data on two or more levels above the ground, making possible experimental
checks of computations to higher levels.
It is the purpose of this paper to investigate the means, accuracy and limita-
tions of the analytical continuation process in application to aeromagnetic data.
Anomalies representing extremes in gradient and irregularity were selected for
extension. The results are presented in the form of contour maps which can be
readily compared with contour maps of observed data on corresponding levels.
Both three-dimensional and two-dimensional fields are treated.

MATHEMATICAL DEVELOPMENT

The flux-gate type magnetometer used in aerial surveys measures AT, the
component of the total-intensity anomaly in the direction of the total field. This
study is confined to AT anomalies usually encountered in prospecting, i.e., those
covering an area sufficiently small to permit considerations of both a plane earth
and an invariant normal field !?; and those with sufficiently small amplitude to
permit the adoption of the direction of ?as that of A?. Under these conditions
AT may be regarded as a potential as is shown below.
* Presented By Title at the St. Louis Meeting of the Society, March 16, 1949. Manuscript re-
ceived by the Editor April 12,1949. Published by permission of the Director, U. S. Geological Survey.
t Geophysicist, U. S. Geological Survey.

517
 518 ROLAND G. HENDERSON AND ISIDORE ZIETZ

Suppose AV is the magnetic potential anomaly produced by the disturbing

geological formation. At all points of free space AV’satisfies the Laplace equation
Downloaded 03/12/15 to 169.230.243.252. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

V2(AV) = o (I)
where

The component of the total-intensity anomaly in the direction of the earth’s

normal field is given by

d(AV)
AT= -~ (2)
at
+
whereTis the unit vector in the direction of T. Operating on each side of equation
(2) with Vr there results

d(AV)
V*(AT) = - V* - = - $ V2(AV) = o.
at
The interchange of the order of differentiation is permissible since AV and its
derivatives are continuous. Thus it is established that AT satisfies the Laplace
equation and is, therefore, amenable to the methods of potential theory.
Suppose we have an aeromagnetic map of AT (x, y,) on S1 (Fig. I), a surface
of constant elevation. The problem of continuing AT (x, y) into the space above
Sr to some point P (x, y, z) is a matter of solving equation (3) subject to certain
boundary conditions, viz., thesolution BT (2, y, a) reduces toknown -values on
Sl and vanishes at great distances above it. This is a classic problem of mathe-
matical physics known as the first boundary value problem (or the Dirichlet

FIG. I. Coordinatesystem and observational planes.

 UPWARD CONTINUATION OF ANOMALIES IN MAGNETIC INTENSITY FIELDS 519

problem) for a plane. The solution may be obtained from Green’s theorem with
Downloaded 03/12/15 to 169.230.243.252. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

the aid of a Green’s function of the first kind: or following the development given
by Byerly2 the solution may be built up from particular solutions of equation (3)
with the aid of the Fourier integral representation of AT (x, y) over plane &.
Since the solution is unique, either approach leads to a solution of the form
c9 00 zAT(x’, y’)dx’dy’
AT(x, y, z) = f. (4)
2a ss-00 __oo [82 + (x’ - x)” + (y’ - y)2]3/2 .

This solution is valid for z>o. The prime variables are the running coordinates
of the plane Sr. There is no loss in generality if the origin is shifted so that the z-
axis contains the point P(x, y, z). Converting to cylindrical coordinates which are
found to be more convenient for numerical evaluation, equation (4) becomes
00 2r zAT(Y, cw)rdrda!
AT(z) = f (5)
23rss0 0 (12 + z2)3’2

where x’= r cos CX,y’=~ sin (Y. Equation (5) permits the computation of AT at
any point of the positive polar axis. To compute the AT anomaly over some plane
SZ at a fixed height, h, above Sr the constant value z= h is inserted in equation (5)
and the latter evaluated at a sufficient number of points of S1 to permit the con-
touring of AT on &.
The evaluation of the integrals of equation (5) involves a very simple proce-
dure. Noting that
zr
AT(r) = -!_
2* 0 S
AT@, a)da

where A? (r) is the average value of AT on a circle of radius Y, equation (5) be-
comes the simple integral

- hrAT(r)dr
AT(h) =
S 0 (12 + h2)3’2

or

AT(h) =
S AWh+,
0
hPw -- (7)

where
hr
4(~., h) =
(y2 + h2)3’2 -

To facilitate the determination of AT (r), a square grid is superimposed over the

anomaly and the values AT (x, y, o) are interpolated at each grid corner. A sheet
’ Webster,A. G., Partial diJerential equetions of mathematical physics, 2nd edition,pp.222-225
z Byerly, W. W., Fourier Series and Sphem’cal Harmonics.
 520 ROLAND G. HENDERSON AND ISIDORE ZIETZ

of transparent paper is placed over the grid and with some corner as center, a
Downloaded 03/12/15 to 169.230.243.252. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

family of n concentric circles passing through as many grid corners as possible

is-described. Each~corner pas& U +4 LI rough- is indicated by a dot on it3 respectike
circle. For each circle a coefficient

f#a(Y<,
h) = hyr
(Yi2 + lzy2

FIG. z. Total intensity aeromagnetic map relative to arbitrary datum, Elk City, Oklahoma.
 UPWARD CONTINUATION OF ANOMALIES IN MAGNETIC INTENSITY FIELDS 521
Downloaded 03/12/15 to 169.230.243.252. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

FIG. 3. Total intensity aeromagnetic map relative to arbitrary datum, Elk City, Oklahoma.
(Flight level approximately 6,000 feet above ground and sea level.)

is determined where ri is the radius of the i-th circle. The circle of radius r,,,
= (h/A) on which 4(r;, h) is a maximum should be included in the family. The
products KF(ri)+(ri, k) may be integrated by plotting and using a planimeter or
by summing the areas of elemental trapezoids, keeping in mind the fact that the
intervals along the v-axis are not evenly spaced. Using a trapezoidal rule, the in-
tegral in equation (6) may be replaced by the following approximation,
 522 ROLAND G. HENDERSON AND ISIDORE ZIETZ
Downloaded 03/12/15 to 169.230.243.252. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

with the understanding that YO=O. In view of equation (8) it is possible to deter-
mine a set of coefficients, K(ri, h) = #$(ri, h) (yi+l-~i-l) to use with the first

FIG. 4. Total intensity aeromagneticmap relative to arbitrary datum computedat 7,935feet

from aeromagneticdata observedat 2,935feet, Elk City, Oklahoma. (Contour
interval IO gammas-elevations abovesea level.)
 UPWARD CONTINUATION OF ANOMALIES IN MAGNETIC INTENSITY FIELDS 523
Downloaded 03/12/15 to 169.230.243.252. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

FIG. 5. Total intensity aeromagnetic map relative to arbitrary datum, Elk City, Oklahoma.
(Flight level 15,935 feet above sea level-approximately 14,ooo feet above ground.)

(n-1) values AT(rJ and a coefficient K(r,, h)=~~(~,,, la)(r,--,,_J to use with
the nth. The working formula is then
n-l
AT(L) B CP(ri)K(ri, h) + AT(r,)K(r,, h). (9)
i-1

Theoretically, the integration is to be carried out over the entire plane; prac-
 524 ROLAND G. HENDERSON AND ISIDORE ZIETZ

I I I I I I I I I I I I I I I
Downloaded 03/12/15 to 169.230.243.252. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

7 : .I 1 4 Ill.,

FIG. 6. Total intensity aeromagnetic map relative to arbitrary datum computed to 15,935 feet
from aeromagnetic data observed at 2,935 feet, Elk City, Oklahoma. (Contour
interval IO gammas-elevations above sea level.)

tically.it is not necessary, for, as r approaches some large value, the integrand in
equation (7) is either negligibly small due to +(r, h) being small or can be made
so with impunity by arbitrarily smoothing to zero. When anomalies are complex
and/or are of high gradient, the smoothing of the integrand to zero for large val-
 UPWARD CONTINUATION OF ANOMALIES IN MAGNETIC INTENSITY FIELDS 525
Downloaded 03/12/15 to 169.230.243.252. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

FIG. 7. Grid used in extension of Anomaly, Elk City, Oklahoma.

ues of r assumes no mean importance. It is necessary, therefore, to add to equation

(9) a smoothing term

where r, is the radius of the outermost circle and AT(r,_I)+(r,_I, k) > hT(r&(m, k).
The radius I, can best be determined by plotting several representative inte-
 96 ROLAND G. HENDERSON AND ISIDORE ZIETZ

grands and selecting the average Y for which the curves begin to level off. In most
casesI* is so large that accurate determinations of ~T(Y,,) are impossible, but this
Downloaded 03/12/15 to 169.230.243.252. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

is not serious since 4(m) is very small. In case ~T(Y,-&$(~,,--~, h) SAT(r,)+(m, h)

the smoothing should be done entirely by graphing.

CONTINUATION OF OBSERVED THREE-DIMENSIONAL

ANOMALIES

As an illustration of the method of analytical continuation consider the aero-

magnetic survey in the vicinity of Elk City, Oklahoma. The results on planes at
2,935, 7,935, and 15,935 foot elevations above sea level are shown respectively
in Figures 2,3 and 5. A r-mile grid was laid down over the anomaly on the first
level and AT was interpolated for the grid corners (Fig. 7). A portion of the
data used in computing to the second level (It = 0.947 mile) is presented in Table
I. A set of circles (Fig. 8) of radii shown in column I, Table I were used. In column
3 are given the coefficients K(ri, h) and in the remaining columns are found the
averages mT(rJ reduced by 200. At the bottom of the last-mentioned columns is
found the sum of the products of the items in that column with those of column
3. To each result there has been added a constant of 167 gammas to bring the
computed values up to the datum of the observations on this level. A contour
map of the computed values of AT is presented in Figure 4. Comparing the com-
puted anomaly with the observed anomaly (Fig. 3), the contours agree generally
within a half-contour interval, representing an average error of less than 5 per-
cent. The agreement could be improved in several ways, e.g., by using more
circles, by increasing the number of points on each circle, or by employing a
correction term to smooth the integrands. A few points were recomputed in-
corporating these refinements, with the result that the average error was less
than 2 percent. Furthermore, it is noted that the area which can be effectively

TABLE I
COEFFICIENTS AND dT(r) VALUES (FOR SAAWLE POINTS) USED IN COMPUTING TOTAL MAGNETIC
INTENSITY FIELD TO 7,935 FEET FROM DATA AT 2,935 FEET ABOVE SEA LEVEL. AVERAGES
HAVE BEEN REDUCED BY zoo-y. ANOMALY IN VICINITY OF ELK CITY, OKLAHOMA
- Z-Z= =
I 2 coki-
cients a42 a96 a94 a68 bw hs hc bu ha
Radiusri +(ri) K(ri; h)
--~ -- --- --- _-- -_ -- .- _-
O.pO o. 3860 0.1292 157 147 169 155 171 160 170 178 165 170
0.670 0.4059 0.101~ 157 147 169 I.55 171 160 170 177 165 17”
I .ooo 0.3624 0.1348 157 148 168 1.54 171 160 170 177 165 170
I .4=4 0.2715 0.1678 157 148 167 151 171 161 170 176 165 170
2.236 0.1478 0.1292 155 I44 65
‘ =47 167 156 168 174 164 169
3.162 0.0832 0.0835 54
‘ 141 160 141 163 153 166 172 162 167
6-3 0.0489 0.0449 :p :3j ij4$ i3i 1~55 r48 =sy fi4 160 164
5.000 0.0359 0.0285 146 129 149 124 149 142 154 162 156 162
5.831 0.0268 0.0670 141 122 143 116 142 137 148 165 152 156
10.000 0.0093 0.0194 92 94 99 80 99 56 101 TO8 119 126
-- -- -- ---- _-- -_ --- --- _-
Computed anomaly+ 1677 306 294 314 298 317 307
-
316
II
- ..23 313
-
318
 UPWARD CONTINUATION OF ANOMALIES IN MAGNETIC INTENSITY FIELDS 527

1
Downloaded 03/12/15 to 169.230.243.252. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

FIG. 8. Diagram for computingaveragevalues of AT on circlesof radius * (C‘ and C,

usedonly in computingto 7,935feet and 15,935feet respectively.)

FIG. Q. Total intensity aeromagneticmap relative to arbitrary datum, Magnum, Oklahoma. (Flight
level 1,935feet above sealevel-approximately 3oo feet above ground.)
 528 ROLAND G. HENDERSON AND ISIDORE ZIETZ

mapped on the second level by computation has its linear dimensions limited by
Downloaded 03/12/15 to 169.230.243.252. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

the diameter of the largest circle. Since reducing the largest circle means sacrific-
ing accuracy, for a given set of data the area which can be covered bears an in-
verse relation to the accuracy of computation.
As a further test of the method the data on the first level were used to com-
pute to the third level (h= 2.462 miles). The results, presented in Figure 6, com-
pare favorably with the observed data in Figure 5.

FOG.IO. Total intensity aeromagnetic map relative to arbitrary datum, Magnum, Oklahoma.
(Observed at 2,935 feet above sea level.)

The general character of the Elk City anomaly is such that the successin ex-
tending it is not beyond expectation. The aeromagnetic survey in the vicinity
of Mangum, Oklahoma, on planes 1,925, 2,935, and 7,935 feet above sea level
affords a more exacting test of the method in view of the complexity, steep gradi-
ents and high amplitudes of the anomalies disclosed. (See Figs. 9, IO, 12.) Indeed
with amplitudes of the order of 2,000 gammas it might well be inquired if the
invariancy of the direction of the earth’s normal field is still a tenable assump-
tion. To some extent this question can be answered empirically by proceeding
as before and checking computed against observed results. The g-mile grid used
in effecting these computations was superimposed in such a manner as to coin-
 UPWARD CONTINUATION OF ANOMALIES IN MAGNETIC INTENSITY FIELDS 529
Downloaded 03/12/15 to 169.230.243.252. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

FIG. II. Total intensity aeromagneticmap relative to arbitrary datum computedat 2,935-foot
level from aeromagneticdata at 1,935 feet, Magnum, Oklahoma.

FIG. 12. Total intensity aeromagneticmap relative to arbitrary datum, Magnum, Oklahoma.
(Observedat 7,935 feet above sealevel.)

tide as nearly as possible with the flight lines. The smoothing term, S, was of
non-negligible proportions and was applied routinely to all values.
In the extrapolation from the I,o35-foot level to 2,935 feet, a set of ten circles
including one of radius r = 0.134 mile for 4(r) maxand an outermost one of radius
Downloaded 03/12/15 to 169.230.243.252. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
 UPWARD CONTINUATION OF ANOMALIES IN MAGNETIC INTENSITY FIELDS 531

r=4 miles gave sufficient accuracy and coverage. A contour map of the com-
Downloaded 03/12/15 to 169.230.243.252. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

puted values is shown in Figure II for comparison with those outlined in the cen-
tral region of Figure IO. Except for a small area in the northwest corner of the
former, the two are in excellent accord, the greatest discrepancy of any point
being a contour interval.

FIG. 15. Comparison between observed and computed profiles for the total magnetic intensity
map of Benson Mines. (5,400feet). Analytical continuation by three-dimensional method from data
at 2,400feet.

The continuation from the 2,935-foot level to 7,935 feet was achieved with
similar success, as shown by a comparison between the computed values (Fig.
13) and the observed values (outlined area Fig. 12). A set of eight circles was
used in this case. An attempt was made to improve the approximation to the
surface integral by increasing the number of points on each circle as well as in-
creasing the number of circles. The resulting values were insignificantly different
from the original ones.
As another illustration, consider the aeromagnetic survey over the Benson
Mines magnetite deposit, St. Lawrence County, N. Y. The results on the 2>400-
and 5,4w-foot levels are shown in Figure 14. Since the north-south dimensions
of the area limit the diameter of the outermost circle of the set used in computing,
only values at points along an east-west profile A-A’ could be extended. A com-
parison between the observed and computed profiles is presented in Figure 15.
 532 ROLAND G. HENDERSON AND ISIDORE ZIETZ

TEE TWO-DIMENSIONAL CASE

Downloaded 03/12/15 to 169.230.243.252. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Anomalies whose contours exhibit a tendency towards parallelism to the ex-

tent that transverse profiles are essentially alike may be regarded as two-dimen-
sional. An example of an idealized two-dimensional anomaly is one produced by
a line of poles. Since many geologic structures have well-defined strikes, e.g., the
anomaly shown in Figure 14 near the 75th meridian, magnetic anomalies of this
character are frequently encountered.
For the two-dimensional case the Laplace equation becomes
a2(AT) a2(AT)
V2(AT) = a22 + a22 = o (10)

assuming that AT does not depend upon the y coordinate, i.e., that a single pro-
file paralleling the x-axis embodies the essential features of the anomaly AT.
A solution to equation (IO) suitable for continuing values of AT into the space
above the plane of observations can be obtained at once by integrating equation
(4) with respect to y’ and setting y= o. The equation becomes
m
zAT( x’)dx’
AT(x, z) = -?
r S-_22
+(x'
- x)'+ (11)

Equation (II) is valid as long as z>o. To use equation (I I) z is replaced by h, as

before, and the integral evaluated for each point on the x-axis whose value is to be
extended. To avoid integration over the entire x-axis, AT is arbitrarily smoothed
to zero outside a g xs b, the essential range of the anomaly. Simplifications are in-
troduced by taking AT at evenly spaced intervals. The integral may be evaluated
by methods suggested above, i.e., by means of a planimeter, or by means of Simp-
son’s or the trapezoidal rule. If the latter is used it is convenient to tabulate
AT(x/), i=o, I, 2 , - . . , n, and to determine a set of coefficients
h(xi’+1 - d-1)
Iqri’, 2, h) = T- i = I, 2, 3, * * * , (n - I),
27r pa2 + (zil - x)“] ’

which can be used repeatedly. The working equation is then

n-1

AT(s, h) = c AT(zi’)K(xi’, x, h). (12)

i-l

As an illustration of the use of equation (12) consider the AT anomaly pro-

duced by a line of poles striking N. 44’ 06’ W, (mag) in a total-intensity field of
inclination 7z” 05’ and located two units below the plane of observations. From
the equation3
2m(z cot I sin /3 - d) sin I
AT =
x2 + a2
a Seethe authors’“Analysis of Total Magnetic-Intensity Anomalies Produced by Point and Line
Sources,” Geophysics,Vol. 13, no. 3 (July 1948).
 UPWARD CONTINUATION OF ANOMALIES IN MAGNETIC INTENSITY FIELDS 533

where 13is the angle between the strike and magnetic north, brais pole strength
Downloaded 03/12/15 to 169.230.243.252. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

per unit length, and Z is the angle of inclination, the first level anomaly A.T(x, o)
(column 2 of Table 2) was computed by setting d= 2 and the second level anom-
aly AL\T(x,4) (column 3 of Table 2) used in checking the accuracy of the exten-
sion to 4 units above the first level by setting d = 6. The coefficients, K(x.! , x, h)
for x=o are given in Table 3. In computing any other point in the range, it is
clear that the same set of coefficients may be used, however they would be shifted
with respect to the AT(z(). It should be observed that AT(x, o) has been arbi-
trarily smoothed to zero at x= - 14 and x= +9. The values of AT(x, 4) obtained

TABLEZ TABLE 3
ANOMALIES PRODUCJSDBY LINE OF POLES 2 AND 6 UNITS BELOW COEFFICIENTS TO BE USED
PLANES z=o AND z=4 RESPECTIVELY: AND ANOMALY AT WITHCOLUMN 2 OF TABLEZ
CONTINUED ~dz=4 FROYVAL~ES ONZ=O

x A%, 0) to;922 A%, 4) K(d) h)

for d=z
.- -
-14 0.0490 0.0751 0 0.07958
-13 0.0542 0.0828 I 0.07490
-12 0.0604 0.0920 o .06366
-11 O.O681 0.1027 f 0.05093
-10 0.0778 0.1154 4 0.03979
0.0901 0.1305 0.124 0.03105
zg o. 1063 5 0.02449
0.1484 0.143
0.1284 o. 1696 o. 165 0.01959
1; 0.191 5 0.01592
0.1594 0.1943
-5 0.2050 0.2223 0.219 9 0.01313
0.2759 0.2525 0.250 IO 0.01097
r; 0.3916 0.2823 0.2’80 II 0.00929
-2 0.5827 o .3068 0.305 12 0.00796
-1 o .8468 0.3202 0.319 13 o .00688
0 0.9515 0.3172 0.316 14 0.006o1
I 0.6756 0.2970 0.296 15 0.00528
2 0.3687 o. 2640 0.264 16 o .00468
3 0.1940 0.2252 0.225 17 0.00417
4 0.1047 0.1866 0.187 18 0.00374
0.0574 0.1521 0.153 19 0.00338
2 0.0309 0.1223 0.124 20 0.00306
0.0153 0.0991 21 0.00279
; 0.0056 0.0799 22 0.00255
9 0.0006 0.0646 23 0.00234
-

by analytical continuation are given in column 4 of Table 2. The results indicate

that the computation can be effected with surprising accuracy except in the
neighborhood of the end points where the effects of smoothing become prominent.
The accuracy was only slightly improved by reducing the interval by one-half.
The Benson Mines anomaly, Figure 14, suggests that a profile taken along
B-B’, Figure 16, on the z,4oo-foot level might admit of a two-dimensional
analytical continuation. The results of the computation to the 5,4oo-foot level are
shown in the upper half of Figure 16, where the observed (solid line) and the
computed (dashed line) have been plotted using the same axis. The accuracy
seems to be of the same order as that resulting from the three-dimensional treat-
 534 ROLAND C. HENDERSON AND ISIDORE Z1ET.Z

ment above, cf. Figure IS. The three-dimensional approach has the advantage of
Downloaded 03/12/15 to 169.230.243.252. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

yielding greater coverage, and the disadvantage of requiring more computational

effort.

FIG. 16. Profiles for total magnetic intensity map of Benson Mines.
A. Comparison between observed and computed profiles at 5,400 foot level using two dimensional
analysis.
B. Observed profile at 2,400 foot level.

CONCLUSION

The use of surface integrals affords a general continuation process and per-
mits the detailed mapping, on higher levels, of fairly complex fields. It is signifi-
cant that the process is in no way concerned with the geologic formations disturb-
ing the field. In general the successof field extensions depends upon the number
of points used in computing as well as the area1 extent and complexity of the
anomaly.
i3n the basis of the relative successmet with in analytically continuing values
of AT on one plane into the space above the plane, there is reason to believe that
little is to be gained by conducting multi-level aeromagnetic surveys. There is
inherent in the first-level data the essential information of the higher. level. In-
deed, if the latter is required, within the limits indicated, one computer can
provide it at a much lower cost than that involved in field operations.