Content uploaded by George Tuckwell
Author content
All content in this area was uploaded by George Tuckwell on Oct 06, 2015
Content may be subject to copyright.
doi:10.1144/1470-9236/07-224
2008; v. 41; p. 371-380 Quarterly Journal of Engineering Geology and Hydrogeology
G. Tuckwell, T. Grossey, S. Owen and P. Stearns
The use of microgravity to detect small distributed voids and low-density ground
Quarterly Journal of Engineering Geology and Hydrogeology
service
Email alerting
to receive free email alerts when new articles cite this article click here
request
Permission
to seek permission to re-use all or part of this article click here
Subscribe
Lyell Collection
to subscribe to Quarterly Journal of Engineering Geology and Hydrogeology or theclick here
Notes
Downloaded by
George William Tuckwell on 18 August 2008
© 2008 Geological Society of London
The use of microgravity to detect small distributed voids
and low-density ground
G. Tuckwell, T. Grossey, S. Owen & P. Stearns
STATS Limited, Porterswood House, Porters Wood, St. Albans AL3 6PQ, UK
(e-mail: george.tuckwell@stats.co.uk)
M
icrogravity is established as a technique for
the detection of natural and man-made
cavities. However, past published examples
have concentrated on substantial natural
cavities or mine workings. Two cases are described for
which the voids present are relatively small and discon-
tinuous, and where much of the affected ground is
characterized by low-density ground rather than large
open voids. Here we use the term ‘low-density ground’ to
encompass ground that has been disturbed by the col-
lapse or partial collapse of material into a void such that
the affected ground shows an anomalously low density in
comparison with the surrounding unaffected ground.
Case study 1 was undertaken with Mouchel Parkman on
behalf of Hertfordshire County Council. A doline had
opened up within a school playground. The collapse was
expected to be related to natural voids within the chalk at
7–10 m depth. The low-density areas identified in the
microgravity survey, in addition to some control locations,
were subsequently proved by dynamic probing. The
investigation accurately identified voided and poorly
consolidated ground, and provided the basis for an
assessment of the risks of imminent and potential future
collapse. Case study 2 was undertaken with Laing
O’Rourke, who were principal contractor for the construc-
tion of the A590 bypass, Cumbria. Exploratory boreholes
confirmed the presence of loose material in the glacial till
overburden associated with suspected voids within the
limestone bedrock. The microgravity survey identified
areas of low subsurface density, which were subse-
quently targeted by boreholes. A detailed ground model
constructed from borehole data allowed a synthetic
gravity map to be calculated for comparison with the
measured gravity. Poorly consolidated ground, depth to
bedrock and surface topographic effects could be iso-
lated and a clear interpretation of subsurface produced.
The microgravity technique is shown to be an effective
tool for the investigation of poorly compacted ground and
small shallow voids in complex and disturbed ground
conditions. When followed by targeted intrusive investi-
gation such surveys can yield a great deal of information
that would not otherwise be available to the engineer.
Different subsurface materials have different bulk densi-
ties. Microgravity surveys seek to detect areas of con-
trasting or anomalous density by collecting surface
measurements of the Earth’s gravitational field. There
are three common origins of local lows in microgravity
data: subsurface voids, variations (usually deepening) in
the depth to bedrock, and low-density ground in the
shallow subsurface. Here we use the term ‘low-density
ground’ to encompass ground that has been disturbed
by the collapse or partial collapse of material into a void
such that the affected ground shows an anomalously low
density in comparison with the surrounding unaffected
ground.
A gravity meter is a highly sensitive instrument that
measures the acceleration due to gravity. When posi-
tioned above a dense material it records the acceleration
(g) as a relative high (a positive gravity anomaly). When
positioned above a low-density feature (e.g. an air-filled
cavity) a relative gravity low (or negative gravity
anomaly) is recorded. Gravity anomalies arising from
natural or man-made subsurface features such as voids
and cavities are superimposed on much larger variations
related to height, latitude and regional geological varia-
tions. To isolate the subtle signal of interest, careful data
acquisition and processing are required.
A number of examples of successful microgravity
surveys have been described in the past over large
features including natural cavities (e.g Patterson et al.
1995; Cooper 1998; Styles et al. 2005) and mine work-
ings (e.g. Bishop et al. 1997; Emsley & Bishop 1997).
The technique has also been used to monitor the pro-
gressive collapse of abandoned salt mines through time
(Branston & Styles 2003). In addition to the discrete
void itself, it is common for the cavity to be associated
with a ‘halo’ of disturbed ground, typically characterized
by enhanced porosity or fracturing. This low-density
ground may act to effectively enlarge the target visible to
a geophysical technique.
The interpretation of microgravity data begins with
the identification of local negative anomalies from the
processed residual Bouguer anomaly map. More sophis-
ticated analysis can be performed to calculate the depth
to causative bodies by Euler deconvolution (e.g.
Durrheim & Cooper 1998; Cuss & Styles 1999) or by
iterative matching of predicted gravity signals from 3D
models to the measured data (Styles et al. 2005).
Gravity has been proven effective for large voids at
depth. Other geophysical techniques such as ground
penetrating radar, EM ground conductivity, resistivity
and seismic methods are capable of locating voids in the
shallow subsurface. These shallow techniques are not
appropriate in all site conditions. For example, if surface
Quarterly Journal of Engineering Geology and Hydrogeology, 41, 371–380 1470-9236/08 $15.00 2008 Geological Society of London
types vary considerably (e.g. from concrete to tarmac
to landscaped grass) this may manifest in the EM
conductivity data as variations caused by changing
thickness of made ground and different drainage con-
ditions associated with different cover materials, which
may mask more subtle features associated with low-
density ground. Where the ground materials are electri-
cally conductive the depth from which reliable ground
penetrating radar reflections can be obtained may be
severely reduced. Where there are areas of hardstanding
it may not be possible to drill holes for electrodes to
undertake resistivity surveys. Of the seismic methods,
the reflection technique is best suited to the detection
of voids; however, these surveys can be slow and expen-
sive to undertake. Noisy site conditions and difficult
access to place geophones can make such surveys
impractical.
Therefore there is a need for a geophysical technique
that can detect small voids in the relatively shallow
subsurface, to bridge the gap between reliable near-
surface investigations using other geophysical tech-
niques and the proven ability of microgravity as a
technique for the detection of large deep features. Fur-
thermore, it is desirable for the technique to be able to
support other geophysical methods or to operate in
situations where the other geophysical techniques may
give data that are unreliable or difficult to interpret.
Here we present data and interpretations from two
projects in which microgravity measurements were used
to direct subsequent intrusive investigations. In both
cases the voids present are relatively small and discon-
tinuous, and much of the affected area is characterized
by low-density ground rather than large open voids. The
combination of geophysical and intrusive data allows a
detailed assessment of the interpreted gravity datasets,
and an assessment of their efficacy in detecting relatively
small near-surface voids.
Case study 1
Introduction
This project was undertaken with Mouchel Parkman on
behalf of Hertfordshire County Council. The trigger
event for the investigation was the appearance of a
doline in one corner of a school playground. The hole
had the characteristics of a small collapsed solution
feature, and was c. 0.6 m in diameter and 1.3 m deep
(Fig. 1).
The site is within an area known to be affected by
natural voids within the chalk. The published 1:50 000
scale geological map of the area (Sheet 256, ‘North
London’) indicates that the site is underlain by the
Lambeth Group of the Palaeogene overlying White
Chalk of the Cretaceous. In this area the chalk is
expected to lie at a depth of 7–8 m. Previous investiga-
tions at the site established that the ground profile
comprises metastable granular (sandy) material overly-
ing chalk that lies at a depth of 7–8 m below ground
surface. A layer of cohesive material has been observed
between the sandy material and the chalk. As a result of
erosion and dissolution, the interface between the chalk
and overlying material is likely to be irregular. Based on
the published geological map and the results of unpub-
lished investigations near the site, the water table is
expected to be at a depth of 20 m below surface. No
direct measurements of water table were made at the site
as part of this project.
There is a considerable range of natural solution
feature geometries possible in the shallow subsurface. A
detailed discussion of formation mechanisms is beyond
the scope of this study. An overview of possible solution
feature geometries is shown in Figure 2. Of particular
relevance to this study is that it is possible that no void
is present and the solution feature manifests as a volume
of low-density ground. It is also possible that a void may
lie within the chalk offset laterally relative to the associ-
ated near-surface low-density ground so as not to lie
directly beneath it.
The school grounds are relatively flat, and are pre-
dominantly tarmac hardstanding, with some areas of
landscaping, artificial soft play surfaces and grass.
Access was restricted in some areas by trees, and by play
equipment.
A geophysical survey was commissioned to determine
the possible presence of voids within the chalk bedrock
elsewhere on the site that may have the potential to
cause future subsidence.
Microgravity survey
The survey was designed to allow for near-complete
coverage of the whole site with the microgravity tech-
nique. The gravity survey was conducted on a 4 m by
4 m grid with three independent readings collected
Fig. 1. Photograph of the doline feature that opened up in one
corner of the grounds of the school.
G. TUCKWELL ET AL.372
at each station to reduce error. Because of access
restrictions the original collapse feature lay outside the
survey area.
Data recorded on site include accurate position and
height measurements for each recorded station. A
theodolite total station was used to survey each station
location with a repeatability of <5 mm.
To account for time-variant effects on gravity read-
ings such as Earth tides and instrument drift, a base
station reading was recorded every hour during the
survey so that the appropriate corrections could be
made during the data processing stage. A Scintrex CG5
instrument was used to collect the data. The instrument
has a quoted accuracy of 1 µGal. The survey was
undertaken in normal working hours, but in the school
holidays, so there was little environmental noise from
human activity. Weather conditions were calm and dry,
so there was also little disturbance from wind or tree
roots. Repeat readings taken as part of the survey
provided repeatability of less than 5 µGal.
The raw data were corrected for Earth tides and
instrument drift on the basis of linear variations between
base station readings, resulting in an expression of the
raw gravity data with temporal variations removed. The
data were corrected for the free air and Bouguer effects
to remove the gravitational effects of elevation and
ground material in order to produce a Bouguer residual
anomaly map (Fig. 3). There was no significant topo-
graphic gradient across the site, and therefore no correc-
tive calculation was required beyond that accounted for
by the Bouguer correction. Gravity observations were
made adjacent to the school building. Analysis of these
data points in the context of the complete dataset
indicated that the mass of the single-storey school build-
ing did not have an effect on the data of sufficient
magnitude to obfuscate an interpretation of the ground
conditions.
Six clear negative anomalies can be identified in the
data based on their relative low values compared with
their immediate surroundings. Anomaly 1 is the largest
in plan view extent. Anomalies 2 and 3 are greatest in
magnitude. Anomalies 1–3 appear to be elongated in the
same orientation, and as such may reflect natural align-
ments, perhaps weaknesses, within the chalk. Anomaly 4
is relatively small in plan view extent in comparison
with others highlighted. Anomaly 5 is relatively low
amplitude. Anomaly 6 is relatively high amplitude in
comparison with the ‘background’ values immediately
around it.
The depth and size of the features causing each
anomaly can be estimated by assuming that they take
the form of simple geometric shapes. As a first approxi-
mation it is useful to assume spherical features. This
assumption allows a quick and simple calculation of the
depth of the feature using the standard technique of
measuring the half-width of the anomaly profile. In this
case calculations suggest that each of the six highlighted
anomalies is likely to be located in the upper 5 m below
ground level. This places each anomaly above the ex-
pected depth to bedrock of c. 7–8 m below ground level.
In turn, this suggests that the highlighted anomalies
represent areas of loose fill or unconsolidated material in
the shallow subsurface above the chalk. It is expected
that these features have been caused by voids within the
chalk bedrock causing low-density ground above.
Dynamic probe results
The dynamic probing test is used to determine the
resistance of soils in situ to the intermittent penetration
of a cone, driven dynamically in a standard manner and
in accordance with BS 1377 (British Standards Institu-
tion 1990). The cone was driven by a weight of 50 kg
falling 500 mm, each fall being termed a blow. The data
Fig. 2. Schematic diagram to illustrate the range of natural solution features that may be present in Chalk. (1) Swallow hole where
stream water permeates down through pre-existing solution pipe below surface hollow. (2) Swallow hole where stream water
permeates down through solution-widened joints and bedding. (3) Solution cavity formed within chalk by solution widening of
joints and bedding as ground water permeates downwards to reach the water table. (4) Solution cavity formed by vadose or phreatic
dissolution, whose roof has become destabilized causing collapse of cover into cavity, forming a collapse doline at the ground
surface. (5) Solution pipe remnants, downslope from cover deposit margin, not infilled with disturbed cover deposits. (6) Solution
pipe formed below thin cover deposits, shallow zone above pipe mouth being disturbed by solution subsidence. (7) Solution pipe
present beneath cover deposit, where upward migration of metastable ground conditions has resulted in ground subsidence forming
a subsidence doline at the ground surface. (8) Solution pipe beneath cover deposit, where downward solution subsidence movement
of the cover deposits has ceased because of the presence of a resistant layer, whereas material below this has continued to subside,
creating a metastable cavity. Material above the cavity is undisturbed. (9) Solution pipe formed below thick cover deposits, zone
above pipe mouth being disturbed by solution subsidence. After Edmonds (1987).
MICROGRAVITY FOR CAVITY DETECTION 373
recorded are the number of blows required for the cone
to penetrate 100 mm into the ground. At each 100 mm
interval this blow count is recorded as the N
100
number.
The locations of the dynamic probe investigations
were chosen so that a comparison could be made
between areas expected to contain low-density ground
and those areas that are expected to be unaffected.
Within Anomaly 3 access with the probing rig was
restricted because of the presence of climbing frames in
the playground. There was no access for dynamic prob-
ing within Anomaly 4 because of a climbing frame and a
tree.
A summary of the dynamic probe results is presented
in Table 1. A blow count of two or less for the
penetration of 100 mm of the dynamic probe into the
ground indicates low-density ground. Interpretation of
dynamic probe results is subjective, there being no
widely recognized published correlation between blow
counts per 100 mm and material properties. Dynamic
probing in the undisturbed granular material generally
reveals relatively low blow counts in the upper metre or
so, typically less than N
100
= 5, rising rapidly to N
100
values of 5–10 through competent material. At the
interface with the chalk, blow counts drop to typically
N
100
= 2–3. Low-density ground is generally recognized
by its weak resistance to the probe, with blow counts of
N
100
%2.
Control locations The control probes away from areas of
negative microgravity anomaly showed a mixed pattern.
DP6 refused at shallow depth on an obstruction, DP7
and DP8 showed the expected ‘normal’ profile of weak
Fig. 3. Microgravity residual Bouguer anomaly plot for the survey area in case study 1. Identified negative anomalies are highlighted
with dashed lines. Dynamic probe locations are indicated and logs are associated with each location. All dynamic probe logs are
plotted at the same scale.
G. TUCKWELL ET AL.374
soils to around 1.5 m depth with dense material beneath.
DP7 penetrated to the presumed top of the chalk at
around 8 m below ground level (mbgl), whereas DP8
refused at around 6 m in dense soil. At the DP11
location initial blow counts were low, typically
N
100
= 0–1, to a depth of 1.5 mbgl, after which blow
counts remained suppressed at N
100
= 3–6 to a depth
of around 5 mbgl, where more competent soils were
encountered. The top of the chalk is inferred to be at
about 7.2 mbgl.
Anomaly 1 Three dynamic probes (DP1, DP2 and DP3)
were carried out at the Anomaly 1 location. DP1 shows
a ‘normal’ ground profile with no evidence of weak
ground and is probably located just outside the
anomaly. The top of the chalk is inferred to occur at
around 8 mbgl. DP2, however, shows very weak to weak
ground to a depth of around 4 mbgl, and below this
depth N
100
values remain depressed in relation to typical
values, with blow counts of typically three or four blows
per 100 mm. The boundary between the chalk and the
overlying soil is not clear. The results from DP3 are
similar, although blow counts are generally a little
higher and the interface with the chalk is evident at
around 8 mbgl.
The dynamic probing has confirmed that Anomaly 1
is an area of weak ground. However, N
100
values are for
the most part above two and there is no evidence of
large voids at depth. It was concluded that there was no
immediate hazard to site users at the Anomaly 1 loca-
tion; however, as with all such features in the area, the
soils within the anomaly have already been weakened
and further loss of strength and associated settlement
may occur in the future, particularly if excessive water
ingress occurs as a result of poor site drainage, leaking
services, or a period of unusually wet weather. The soils
may also be susceptible to collapse under the influence of
additional load from building foundations, etc.
Anomaly 2 Anomaly 2 was investigated by DP4. The
results from DP4 indicate fairly competent ground,
although a little weaker than normal, to a depth of
c. 3.5 m, where blow counts drop to N
100
= 0 or 1. Only
two blows were required to drive the probe from 3.6 m
to 5.0 m, indicating very weak or voided ground at this
location. N
100
values remain at zero or one to a depth of
around 8 m, where the top of the chalk is inferred. N
100
values then rise to N = 6 or 7 to the full depth of the
probe.
The ground beneath Anomaly 2 appears to be much
weaker than that beneath Anomaly 1, with a probable
1.4 m void at a depth of 3.6 mbgl and very weak ground
beneath to a depth of around 8 m.
Anomaly 3 Anomaly 3 was investigated with dynamic
probe DP5. The N
100
values were initially low at N =0
to 2, with a small void or very weak zone at 1.5 mbgl.
Below this depth, N
100
values increase gradually to
attain a ‘normal’ profile by about 2.8 mbgl. The top of
the chalk is inferred to be at about 7.6 mbgl.
Based on the results of this single dynamic probe,
Anomaly 3 would appear to represent an area where the
low-density ground is present to a slightly greater than
normal depth below ground level, with no evidence of
very weak ground or voids beneath. However, past
experience has shown that within the areas of low-
density ground, soil strength can vary markedly over
very small horizontal and vertical distances, and weaker
ground may exist within the Anomaly 3 location that
has not be revealed by the single probe. It should also be
noted that access restrictions prevented the investigation
of the centre of the anomaly. It is possible that more
extensive volumes of poor ground are present but have
not been encountered by the single dynamic probe
location.
Anomaly 5 Dynamic probes DP9 and DP10 investigated
Anomaly 5. Neither probe show any unusual areas of
low-density ground, therefore the Anomaly 5 location is
not considered to be a cause for concern. DP10 suggests
that the top of the chalk is c. 8.2 mbgl.
Anomaly 6 Dynamic probe DP12 was used to investigate
Anomaly 6. Initially, DP12 N
100
blow counts were
slightly low, but roughly consistent with the nearby
DP11, formed away from any anomalous microgravity
data. However, at a depth of c. 4.6 m, N
100
values drop
dramatically with N
100
= 0 or 1 to the full depth of the
probe at 9 mbgl. From 6.9 to 9 mbgl, only four blows
were required to drive the cone 2.1 m. No increase in
resistance was evident at 9 mbgl and it is likely that the
low-density or voided ground extends to greater depth.
It is also noted that this is the only anomaly investigated
where poor ground would appear to extend into the
chalk.
Table 1. Summary of dynamic probe results, case study 1
Dynamic probe
number
Zone of low-density ground
(N
100
= 0 to <2) (mbgl)
DP1 1.0–1.7 and 8.1–9.5
DP2 0.7–3.0
DP3 0.5–1.5 and 7.9–8.9
DP4 1.3–1.9 and 3.6–8.0
DP5 0.0–2.0
DP6 n.a.
DP7 0.0–1.4
DP8 0.4–1.7
DP9 0.6–1.3
DP10 0.0–1.1
DP11 0.0–2.2 and 7.0–7.6
DP12 0.0–2.0 and 4.6–9.0
n.a., not applicable.
MICROGRAVITY FOR CAVITY DETECTION 375
Discussion
With the possible exception of Anomaly 5, the intrusive
investigation confirmed that the low gravity anomalies
identified by the microgravity survey do represent areas
of low-density ground and/or voided ground. The
dynamic probing has also shown that away from the
anomalies, the ground strength profile is ‘normal’ for
the site.
Of the anomalies investigated, none would appear to
present an immediate hazard to site users. However, past
experience has shown that all such features in this area
are metastable and further loss of strength and associ-
ated settlement or collapse may occur in the future,
particularly where changes occur in groundwater flows
or load conditions.
Case study 2
Introduction
Case study 2 is an investigation of a section of the A590
bypass road constructed at High Newton, Cumbria, and
was undertaken on behalf of Laing O’Rourke, who were
principal contractors for the project. The geology of the
site comprises c. 14 m of glacial till over the Dinantian
Limestone. The water table was proved to lie between
5 and 10 mbgl in the area surveyed. High Newton is
located to the north of Grange-over-Sands, and lies on
the southern edge of the Lake District fells (Fig. 4). The
area surveyed is on the western slope of a valley,
immediately to the west of the present A590 road.
Low-density ground thought to be associated with the
presence of solution features within the limestone was
discovered during initial excavation and ground investi-
gation works at the site. A preliminary geophysical
investigation was undertaken over a small section of the
route. The survey highlighted three main areas for
further investigation, one of which resulted in the identi-
fication of low-density ground in the near-surface.
A second phase of geophysical investigation was
commissioned to extend to the north and south of the
area surveyed in the first phase of investigation. The
principal aim of the geophysical investigation was to
identify variations in subsurface density associated with
voids or low-density ground within the limestone or
overlying glacial till.
Microgravity survey
The gravity survey was conducted on a 5 m by 5 m grid
with three independent readings collected at each station
to reduce error. Data recorded on site include accurate
position and height measurements for each recorded
station. An RTK differential global positioning system
(GPS) was used to survey each station location with a
repeatability of <12 mm.
To account for time-variant effects on gravity read-
ings such as Earth tides and instrument drift, a base
station reading was recorded every hour during the
survey so that the appropriate corrections could be
made during the data processing stage. A Scintrex CG5
instrument was used to collect the data. The survey was
undertaken in normal working hours. The site was
cleared of surface vegetation. Although construction
works were continuing on site, these were restricted to
areas away from the survey location. Site traffic occa-
sionally generated problematic ground vibrations, but
data quality checks undertaken in real time allowed
for affected data to be rejected and the observation
repeated. Repeat readings taken as part of the survey
provided repeatability of less than 8 µGal.
Over the two phases of the project four separate
survey grids were defined to allow for access restrictions
from continuing construction works. Data from Phase 1
were combined with the Phase 2 data to provide a map
with continuous coverage. Continuing site activity made
it impossible to reoccupy the base station location or
any other observation points from the Phase 1 survey,
therefore the relative gravity residual values between the
two phases have been matched by equating the average
readings from each survey grid where they overlap. The
overlap between the survey grids is shown in Figure 4.
Errors between adjacent observation points from the
two phases of the survey are estimated to be less than
14 µGal.
A regional trend was clearly identifiable in the data,
with values increasing or decreasing over a length scale
of 100 m or more. This ‘regional’ trend was calculated
and removed from the data to provide a ‘flatter’ residual
Fig. 4. Location map of survey area located on a section of the
proposed route of the A590 High to Low Newton bypass,
Cumbria, UK.
G. TUCKWELL ET AL.376
value map from which local anomalous lows could more
easily be identified.
The data collected from the area to the south of the
access road show consistently higher values than those
for the rest of the survey areas. The data as presented for
this area have been processed separately from those for
the rest of the areas to remove the effect of the regional
difference, and therefore allowing local lows in the data
to be identified. The resulting map (Fig. 5a) represents
the gravitational variations associated with subsurface
density variations, and is the map from which interpre-
tations can be made.
A number of negative anomalies can be identified in
the data based on their relatively low values compared
with their immediate surroundings, and these are high-
lighted in Figure 5a. The anomalies themselves are
irregular in plan view, suggesting complex ground con-
ditions. This is not unexpected in areas of glacial till,
where the heterogeneous nature of the material can
affect its response to subsidence. The anomalies do,
however, align along a postulated lineament running
approximately north–south through the survey area.
Geological faults strike approximately north–south or
NW–SE through this area, with fault blocks down-
thrown to the west. It is possible that a structural feature
associated with the regional tectonics has in the past
localized the natural processes of void formation.
Borehole data and synthetic gravity model
Further analysis has been made possible by an extensive
programme of drilling undertaken around the central
anomaly (highlighted in Fig. 5). A significant volume of
‘soft’ or ‘broken’ ground has been identified in the
borehole logs. In these logs ‘soft’ ground refers to
low-density soils and glacial till material, and ‘broken’
refers to extensively fractured (low-density) limestone
bedrock. A summary of the borehole logs is provided in
Table 2. The measured microgravity data show two
main lows in the area (Fig. 5a and b). Figure 5c shows
the depth to bedrock determined from the borehole logs,
with simplified stick logs superimposed at each borehole
Fig. 5. (a) Microgravity residual Bouguer anomaly plot for the survey area in case study 2. Inset indicates area detailed in (b–e). (b)
Detailed view of measured residual Bouguer anomaly data as shown in (a). (c) Contours of depth to bedrock as determined from
borehole data. Local borehole data annotated as stick diagrams at each hole location as marked. Green indicates ‘good’ ground
conditions; red indicates ‘soft’ or ‘broken’ ground. (d) Gravity anomaly predicted by ground model defined by borehole data. (e)
Measured residual Bouguer anomaly plot adjusted to remove the effects of variations in the depth to bedrock.
MICROGRAVITY FOR CAVITY DETECTION 377
location indicating good ground (green) and low-density
ground (red).
If a density contrast between the low-density ground
and the surrounding ‘intact’ ground is assumed then it is
possible to calculate a theoretical Bouguer residual
anomaly pattern that would be predicted to be measured
over the area. The calculations used here are based on a
simple Bouguer slab model (Robinson & Coruh 1988);
they assume a simplified geometry of the subsurface
defined by the borehole logs, and a constant density
contrast, and therefore do not accurately capture the
likely complexity of the ground in the area. They do,
however, provide a useful basis on which to compare
the predicted with the measured gravity values so as
to better understand the accuracy of the measured
gravity map in representing the presence of problem
ground conditions. The density contrast assumed is
0.4gcm
3
, and is the same regardless of whether the
low-density ground is within the soils, glacial till or
limestone.
Table 2. Summary of borehole data, case study 2
Exploratory
hole
Top of
feature
Base of
feature
Feature type Exploratory
hole
Top of
feature
Base of
feature
Feature type
3 0 15 Soil 3N 0 13.5 Soil
15 15.1 Bedrock 13.5 14.5 Broken ground
3A 0 14 Soft ground 14.5 15.4 Soil
14 14.1 Bedrock 15.4 15.5 Bedrock
3B 0 10.5 Soil 3O 0 9 Soil
10.5 11 Broken ground 9 9.6 Broken ground
11 11.1 Bedrock 9.6 14.3 Soil
3C 0 3 Soil 14.3 14.4 Bedrock
3 11 Soft ground 3P 0 12.9 Soft ground
11 11.2 Soil 12.9 15.4 Soil
11.2 11.3 Bedrock 15.4 17.5 Bedrock
3D 0 12.9 Soil 3Q 0 10.8 Soil
12.9 13 Bedrock 10.8 10.9 Bedrock
3E 0 7 Soil 3R 0 5.9 Soil
7 9.9 Soft ground 5.9 7.2 Soft ground
9.9 10.5 Broken ground 7.2 10.6 Soil
10.5 13.1 Soil 10.6 11.1 Broken ground
13.1 21.9 Broken ground 11.1 11.2 Bedrock
21.9 22 Bedrock 3S 0 7 Soil
3F 0 11.4 Soil 7 8.3 Soft ground
11.4 11.5 Bedrock 8.3 11 Soil
3G 0 8 Soil 11 11.8 Broken ground
8 10 Soft ground 11.8 13.1 Soil
10 10.1 Soil 13.1 13.2 Bedrock
10.1 10.2 Bedrock 3T 0 7 Soil
3H 0 5.1 Soil 7 22.5 Soft ground
5.1 10 Soft ground 22.5 22.6 Bedrock
10 12.1 Soil 3U 0 8.2 Soil
12.1 21.5 Broken ground 8.2 13.3 Soft ground
21.5 21.6 Bedrock 13.3 13.4 Bedrock
3I 0 10.2 Soil 3V 0 11.6 Soil
10.2 11.9 Soft ground 11.6 19.5 Broken ground
11.9 16 Broken ground 19.5 19.6 Bedrock
16 16.1 Bedrock 3W 0 13.1 Soil
3J 0 6 Soil 13.1 13.2 Bedrock
6 9.5 Soft ground 3X 0 9.3 Soil
9.5 14 Soil 9.3 10.2 Soft ground
14 14.1 Bedrock 10.2 16.7 Soil
3K 0 14.4 Soil 16.7 19.7 Broken ground
14.4 14.5 Bedrock 19.7 19.8 Bedrock
3L 0 11.5 Soil 3Y 0 7.6 Soil
11.5 11.6 Bedrock 7.6 12 Soft ground
3M 0 15.1 Soil 12 14 Soil
15.1 15.2 Bedrock 14 14.1 Bedrock
G. TUCKWELL ET AL.378
The predicted anomaly map is shown in Figure 5d.
Lows are located in approximately the same positions as
those identified in the measured data; however, the
predicted amplitude of the low centred on observation
point 18.1 is significantly greater than that observed.
Most notable is that whereas borehole 3T located a
significant thickness (15.5 m) of low-density ground ex-
tending from a depth of 7 mbgl the predicted gravity
anomaly is lower in amplitude than that calculated for
borehole locations 3A and 3P, with a thickness of
low-density ground of 14 m and 13 m, respectively,
extending from the surface. This highlights the effect of
the depth of the feature on the gravitational signature.
Features at or near surface will have a greater signature
in the gravity anomaly map than equivalent features at
greater depths. Although this depth effect is well known,
these results emphasize the need to provide a careful
analysis of the data as an interpretation of the gravity
signal to the engineer.
A further influence on the measured gravity values is
the depth to bedrock. Assuming that the overlying
glacial till has a lower density than the underlying
bedrock, then a deepening of the depth to bedrock will
manifest in the data as a relative low. The measured
depths to bedrock are shown in Figure 5c. The depth
varies considerably within the area defined by the group
of boreholes, from 10 to 22 m.
It is possible to estimate the effect of these variations
on the measured gravity data using equivalent calcula-
tions to those undertaken above to predict the signature
of low-density ground. Figure 5e shows the measured
residual gravity anomaly map adjusted to correct for the
calculated effect of variations in the depth to bedrock. A
density contrast of 0.2 g cm
3
was used in the calcula-
tions. The locations of the low anomalies have not
changed significantly from the uncorrected data (Fig. 5a
and b) but relative values have changed. Most sig-
nificantly, the magnitude of the anomaly centred on
observation point 18.1 has increased. As a result, the
corrected anomaly map shows a closer agreement with
the predicted anomaly distribution (Fig. 5d).
Discussion
In the context of this location, the principal factor
affecting the ability to detect poor ground is the depth at
which the affected ground lies beneath the surface. In
this example relatively large volumes of soft or broken
ground present at depth have been shown to manifest as
relatively low-amplitude negative anomalies in the
measured microgravity data. Although still detectable
by the microgravity technique, these features are super-
imposed on larger signals attributable to shallow ground
conditions. This produces a complex picture that re-
quires careful consideration by the geophysicist, and
also clear communication to the engineer. The detailed
borehole investigation of one of the major anomalies
identified in this dataset allowed, through some simple
and quick calculations, a link to be demonstrated be-
tween the proven ground conditions and the micro-
gravity survey results.
We note that more sophisticated data analysis tech-
niques are available, and may be appropriate in other
circumstances; for example, the removal of the gravita-
tional effects of above-ground structures and buildings.
However, these would be inappropriate in cases such as
those discussed here, where the major limitation on the
accuracy of a ground model is the spacing of reliable
borehole data.
Conclusions
Complex ground conditions comprising heterogeneous
soils such as made ground or glacial till in addition to
low-density and voided areas may give rise to irregular
and complex anomaly profiles that may subsequently be
verified by targeted intrusive investigations. Should it be
necessary or useful for the engineering requirements of
the project to further analyse complex gravity anomalies
it is possible to unravel these signals by calculating
and summing anomaly profiles from simple geometric
features.
Based on the analysis of the two microgravity datasets
described here, under the scrutiny of extensive follow-up
intrusive investigations, it has been possible to demon-
strate that the microgravity technique is capable of
locating low-density ground associated with natural
voiding processes. The method is effective even where
the causative void is either no longer present or lies at
some greater depth and is not itself detected by the
microgravity observations.
The technique has an established record of the detec-
tion of large natural and man-made voids at depth. It
has been demonstrated here that, provided the feature is
sufficiently close to the surface, it should be possible to
use microgravity techniques. It is therefore a valid and
useful technique for the detection of near-surface voids
and low-density ground.
The use of microgravity has, on some occasions, been
hampered in commercial settings by its relative expense
per unit area compared with other techniques; however,
it should be given proper consideration by the project
engineer. Other geophysical techniques (e.g. ground
penetrating radar, EM ground conductivity, resistivity
imaging, magnetic gradiometry, and seismic refraction
and stiffness measurements, to name a few) may provide
a more cost-effective method of covering large areas. An
informed consideration of all available methods is im-
portant; however, we emphasize that microgravity is, in
essence, the only non-intrusive technique that directly
measures the principal defining property of a void: the
absence of mass. For this reason, it should be considered
at the earliest stages of a project. Decisions on the
MICROGRAVITY FOR CAVITY DETECTION 379
investigative programme should be based on a careful
analysis of the relative risk (and economic cost) of using
a less reliable method of non-intrusive investigation.
Where the preferred course is an extensive programme
of boreholes, consideration should be given to undertak-
ing a microgravity survey to identify areas of ground to
target and a comparison made with the cost of obtaining
equivalent coverage and detail of information using
boreholes alone. A ‘geophysics first, boreholes later’
approach is likely in many cases to provide the most
detailed and cost-effective method to investigate the
ground conditions.
References
B, I., S, P., E,S.J.&F, N.S. 1997.
The detection of cavities using the microgravity technique:
case histories from mining and karstic environments. In:
MC, D.M., E, M., F, P.J. &
R, G.M. (eds) Modern Geophysics in Engineering
Geology. Geological Society, London, Engineering
Geology Special Publications, 12, 153–166.
B,M.W.&S, P. 2003. The application of
time-lapse microgravity for the investigation and monitor-
ing of subsidence at Northwich, Cheshire. Quarterly
Journal of Engineering Geology and Hydrogeology, 36,
231–244.
B S I 1990. BS 1377. Methods of
test for soil for engineering purposes. BSI, London.
C, A.H. 1998. Subsidence hazards caused by the
dissolution of Permian gypsum in England: geology,
investigation and remediation. In: M, J.G. &
E, M. (eds) Geohazards in Engineering Geology.
Geological Society, London, Engineering Geology Special
Publications, 15, 265–275.
C,R.J.&S, P. 1999. The application of micro-
gravity in industrial archaeology: an example from the
Williamson tunnels, Edge Hill, Liverpool. In: P,
A.M. (ed.) Geoarchaeology: exploration, environments,
resources. Geological Society, London, Special
Publications, 165, 41–60.
D, R.J. & C, G.R.J. 1998. EULDEP: a program
for the Euler deconvolution of magnetic and gravity data.
Computers and Geosciences, 24, 545–550.
E, C. N. 1987. The engineering geomorphology of karst
development and the prediction of subsidence risk upon the
chalk outcrop in England. PhD thesis, University of
London.
E, S.J. & B, I. 1997. Application of the micrograv-
ity technique to cavity location in the investigations of
major civil engineering works. In: MC, D.M.,
E, M., F,P.J.&R, G.M. (eds)
Modern Geophysics in Engineering Geology. Geological
Society, London, Engineering Geology Special Publica-
tions, 12, 183–192.
P, D.A., D, J.C., C,A.H.&F, J.K.
1995. The investigation of dissolution subsidence incorpo-
rating microgravity geophysics at Ripon, Yorkshire.
Quarterly Journal of Engineering Geology and
Hydrogeology, 28, 83–94.
R, E.S. & C, C. 1988. Basic Exploration
Geophysics. Wiley, New York.
S, P., MG, R., T,E.&C, N.J. 2005.
The use of microgravity for cavity characterization in
karstic terrains. Quarterly Journal of Engineering Geology
and Hydrogeology, 38, 155–169.
Received 26 July 2007; accepted 16 May 2008.
G. TUCKWELL ET AL.380
