Gravity surveying: a brief introduction

Hugh C. Pumphrey

September 24, 2014

Abstract
This is an introduction to gravity surveying for absolute beginners. Students
should read it before the Edinburgh/Paris Sud/M unster geophysics field course. It
explains what gravity measurements can tell us about the subsurface, how a gravity
survey is done and how the data are processed. It does not explain the details of
using a gravity meter: these will be explained by the teaching staff on the field course.

1 Introduction
Everything is gravitationally attracted to everything else. And the gravitational attrac-
tion of an object is proportional to its mass. So if the rocks below you at a given place are
denser, then the gravity there will be slightly larger. The changes in gravity from place
to place are small: gravity, g, at the Earths surface is about 9.81 ms2 , but the local
variations are a tiny fraction of this; often we are measuring differences of 106 ms2 . If
you can measure how g changes from place to place, you can learn something about how
the density of the rocks below you varies: see Figure 1.

2 Measuring gravity
It is possible to build an instrument that measures g directly. Such an instrument is
called an absolute gravity meter and is large, unwieldy and expensive. For field surveys it
is more usual to use a relative gravity meter. These are cheaper, smaller and more robust.
But they do not measure the absolute value of g. They can only measure the differences
in g between one place and another. Relative gravity meters are essentially a mass hung
on a spring: if you go to somewhere where gravity is a bit larger, the spring stretches a bit
more. The extra stretch is tiny: to measure it, we pull on the spring with a micrometer
screw to restore the mass to the original position. Levers are used to make the system
more sensitive and the whole mechanism is enclosed in a temperature-controlled box to
prevent changes in temperature from affecting the results.

3 Measuring altitude
Although g is affected by differences in the rock density below the ground it is affected
a lot more by the height of the ground surface. This means that in order to see the
small differences in g caused by the density changes, you have to measure the height
differences for each of your gravity measurements. For small-scale surveys (1 km long,
say) the heights need to be measured to an accuracy of 1 cm. GPS can not provide
this level of accuracy: the altitudes must be surveyed in the traditional manner, using a
theodolite and a levelling stave. The process is shown in Figure 2.

1
 Bouguer anomaly / Grav units

10
Entered
Fitted OK

5
Fitted (conv fail)
Data

0
5
15

2000 1000 0 1000 2000 3000

Distance / m
500

Entered (+)
Entered ()
Fitted (+)
0

Fitted ()
Depth / m

1000

2000 1000 0 1000 2000 3000

Distance / m

Figure 1: Gravity surveying: the basic idea. The lower panel shows a cross-section
through the ground. The circles represent denser (right) and less dense (left) regions.
The upper panel shows the gravity that might be measured at the surface.

4 Reducing the data

The number which is read from the gravity meter is simply the number of turns of
the micrometer screw from the end position. To convert this into a useful value for
interpretation it is necessary to apply several corrections to it.

4.1 Taking differences

As the instrument is only a relative instrument, we subtract the reading for the first
measurement point from all of the other measurements.

4.2 Applying the calibration constant

Each gravity meter has its own calibration table which converts micrometer dial turns
into physical units. The traditional unit for gravity is the milligal (mgal: 105 ms2 ); the
modern tendency is to work in gravity units (gu: 106 ms2 ). The calibration constant
varies slightly with number of dial turns. Figure 3 shows an example.

4.3 Applying the drift correction

If left at the same place, the measurement recorded by a relative gravimeter tends to
change slowly (or drift) due to fluctuations in temperature etc. We correct for this my
making a measurement at the same location at the beginning and end of the day. By
recording the time of every measurement and assuming that the drift in the instrument

2
 b1 f2

f1
b0
S2 (height H2 )
S1 (height H1 )

H1 H0 H2 H0

S0 (height H0 )

Figure 2: Use of a theodolite (black) and levelling staves (grey) to measure the altitude of
stations S1 and S2 with respect to the base station S0 . The theodolite is a telescope that
looks exactly horizontally. The staves are giant rulers, marked in cm. The backwards
and forwards readings bi and fi are read by looking the staves through the theodolite.
The heights are given by H1 = H0 + b0 f1 , H2 = H0 + b0 f1 + b1 f2 etc.

is linear in time we can correct our measurements to account for the drift, as shown in
Figure 4

4.4 Applying the height correction

If a gravity meter is moved upwards (away from the Earth) then the value of g which it
measures will decrease. Measured values must be corrected for this effect; this is called
the free air correction. The usual reason that a gravity meter is at a different altitude is
that the ground is higher. The correction for the gravity of this extra mass of ground is
called the Bouguer plate correction. To make it, a value for the typical density of rock
must be assumed. As the free air and Bouguer plate corrections are both proportional
to height they are often combined into a single height correction. Gravity values with
the height correction applied are called Bouguer anomalies. For a typical rock density
of 2.67 tonnes / m3 , the height correction is about 1.97 gu (0.197 mgal) for every m of
height.

4.5 Applying the latitude correction

Gravity varies with latitude according to the normal gravity formula:

gn = ge (1 + 1 sin2 + 2 sin2 2)

For a small scale survey we often approximate this formula by a linear or quadratic
fit using either latitudes or grid Northings. At the latitude of Derbyshire (53 N) the
latitude correction is approximately 7.834 gu (0.7834 mgal) for each km north of the base
station. At the latitude of Bermuthshain, Germany (50.47 N) it is 7.999 gu (0.7999 mgal)

3
Figure 3: Section of the calibration table for a Lacoste-Romberg gravity meter. The
important number is the one labelled FACTOR FOR INTERVAL we multiply the
counter reading by this to get a value in physical units. (We would also multiply by 10
to get the answer in gu rather than mgal.)

Figure 4: Applying a correction for in-

strumental drift to relative gravity mea-
20

surements. Each point is supposed to be

Measured Gravity / gu

a measurement at a different location.
10

But the first and last points were mea-


sured at the same place, so gravity there

should not change. The measured drift
0

will include the effect of the gravity tides

(daily amplitude about 1 gu) as well as
10

Measured

Corrected
the instrument drift. Serious people cor-
rect for the tide separately but we will
0 1 2 3 4 5
lump it in with the drift correction.
Time / hours

for each km north of the base station. At Montalivet-les-Bains, France it is 8.1397 gu

(0.81397 mgal) for each km north of the base station.

4.6 Applying the terrain correction

This correction accounts for the fact that a nearby mountain pulls upwards on the gravity
meter, reducing the measured value. And a nearby valley fails to pull downwards on the
meter, again reducing its value. This is by far the most tiresome correction to apply as
we need to consider all of the terrain surrounding the measurement point. If the terrain
is sufficiently smooth then this correction is often not carried out.

5 Modelling the results

Once you have reduced your measurements to Bouguer anomalies, they must be inter-
preted. The usual way to do this is to construct a model of the sub-surface density
variations which reproduces the measured gravity. Figure 1 shows an example of this.
It is important to understand how under-constrained the problem is: there are many
possible arrangements of density which could match a set of measurements.