GPR for river dyke - GSB

GOVERNMENT OF THE PEOPLE’S REPUBLIC OF BANGLADESH
MINISTRY OF POWER, ENERGY AND MINERAL RESOURCES
ENERGY AND MINERAL RESOURCES DIVISION
GEOLOGICAL SURVEY OF BANGLADESH
153, Pioneer Road, Segunbagicha, Dhaka
Geo-Hazard Risk Reduction in Bangladesh- On Site Training of
Using GPR (Ground Penetrating Radar) for Turag River Dyke
Monitoring in Bangladesh
Prepared by
Geological Survey of Bangladesh (GSB)
&
Norwegian Geotechnical Institute (NGI)
ENVIRONMENTAL GEOLOGY AND NATURAL HAZARDS ASSESMENT BRANCH
March, 2012

Contributors
Contributors from Geological Survey of Bangladesh (GSB)
Reshad Md. Ekram Ali, Director & Project Director.
Salma Akter, Deputy Director.
Md. Mahmood Hossain Khan, Assistant Director.
Shahtaj Karim, Assistant Director.
Mohammad Zohir Uddin, Assistant Director.
Contributors from Norwegian Geotechnical Institute (NGI)
Dr. Rajinder Kumar Bhasin
Dr. Lloyd Tunbridge
Dr. Fan-Nian Kong
Mr. Pawel Jankowski

Summary
The Norwegian Geotechnical Institute and the Geological Survey of Bangladesh have
carried out a Ground Penetrating Radar (GPR) Survey along a River dyke in Dhaka
Bangladesh. The purpose of this survey was to build-up capacity at GSB in using advanced
technologies for sub-surface geotechnical investigations.
Two different types of (GPR) antennas were used to demonstrate the capacity of detecting
sub-surface materials. It was demonstrated that GPR can serve as a quick and non
destructive method of testing that can be readily used in Bangladesh for detecting subsurface
structures such as weak spots in dykes or rebars in concrete.

Contents
1 Introduction 5
2 Principle of using GPR for river dyke monitoring 6
3 NGI GPR system 6
4 Test results 8
4.1 Indoor test results 8
4.2 GPR tests on dyke
4.3 GPR tests on road 9
5 GPR theory revisited: Ground material property, wave speed and propagation
attenuation 17
6 Conclusions 18
Appendices
Appendix A: Wave propagation in ground medium
Appendix B: Application on GPR for subsurface investigations
Control and review page

1 Introduction
The Geological Survey of Bangladesh (GSB) and the Norwegian Geotechnical (NGI)
Institute are co-operating in an institutional co-operation project titled “Geodisaster Risk
Reduction in Bangladesh”. In this project, one of the key Tasks is to build-up capacity at
GSB in using advanced geophysical tools for investigation of river dykes in Dhaka. The main
goal of the project is to reduce the risks from natural hazards in Bangladesh by increasing
the capacity of GSB through collaborative pilot project studies.
Ground penetrating Radar is a useful tool for geophysical survey, which can be used to
detect locations and extents of weak zone areas of a river dyke. Hence building up the
ability for GSB personnel to use GPR for river dyke controlling becomes an important topic
of this co-operation project. GSB personnel were trained at NGI in Norway July last year
(2010) on how to use the NGI GPR. In March 2011 an on site training in Dhaka, Bangladesh
took place.
GSB personnel have the background of performing seismic survey. GPR has similar record
as seismic survey. The main difference between GPR and seismic methods is that seismic
reflection comes from a boundary where there is an acoustic impedance contrast and in the
case of GPR the reflection comes from a boundary where there is a dielectric constant
contrast. So the geoscientists of GSB are trying to extrapolate their knowledge from their
training and seismic survey experience to GPR because in both cases the data (GPR &
seismic) have similar 2D view. However, in seismic reflection method, the depth of
penetration is very large so the geoscientists think about gross lithology but in case of GPR
they have to think very minutely about the lithology and other variation within a very shallow
depth in the range of 10 m. For any geophysical methods, the detection resolution is a
portion (for example, one tenth) of the test distance. Therefore, GPR normally gives
detection results with much higher resolution than the seismic method. The other
advantages of using GPR are that for detection using electromagnetic wave, the sensors (or
antennas) do not need to have good contact with the ground, which enables a fast survey
speed, and that the GPR signal is generated by electronics, and is coherent and repetitive,
which allows to use advanced signal and image processing methods.Figure 1 shows the
GSB and NGI team at the office in Dhaka. In this Figure Professor Lars Ottemoller from the
University of Bergen, second from right, is also present to plan the Task related to the
seismic component of the project.
Figure 1: GSB – NGI team at the GSB office in Dhaka

2 Principle of using GPR for river dyke monitoring
The main purpose of on site training is to demonstrate the ability of using GPR to identify for
example, the disturbed and weak zones in the dyke, as well as re-bars in the concrete etc.
The weak zones include loosely compacted layers, joints in different ground materials within
the dykes, presence of highly compressive layers, etc.
GPR is a non-destructive method that can perform quick surveys on top of a river dyke. The
conventional GPR test mode uses the reflection wave, where two antennas with a fixed
spacing are used, one for transmitting and the other for receiving signals. They are moved
along the river dyke. Electromagnetic waves can be transmitted through the ground with the
help of the transmitting antenna. The energy of the wave is reflected back towards the
surface where the electric or magnetic properties change (due to the weak zones etc.) in the
river dyke. A second antenna receives the reflected wave. The ground materials will
attenuate the radar wave. The detection distance is therefore dependent on the types of river
dyke materials. Generally speaking, materials such as concrete, sand, rock, gravel, sandy
soil etc are good materials for radar wave to penetrate. However silty clay and wet clay etc
are the difficult materials for radar wave to penetrate. For good materials the GPR detection
distance can be longer than 10 m.
Even though the detection distance is limited in clay material etc, the GPR reflection
measurement on top of the river dyke is a commonly accepted method used for the dyke
monitoring. This is because river dyke material changes at shallow distances may imply that
disturbances may take place in the internal part and thus deform the shallow layered
structure.
As mentioned, the GPR survey is non-destructive investigation method, as well as being
quick. Survey speed can be as fast as 2-3 km/hour. For an experienced operator, the
interpretation of the survey results can be made on site without further data processing
(inversion etc.). Hence, the GPR survey is low cost geotechnical investigation method.
3 NGI GPR system
At NGI there has been a continuous activity in developing and testing radars for different
subsurface mapping tasks since 1989. The work has included development of hardware and
software for data acquisition and algorithms for data processing and interpretation. In parallel
with the development work, NGI has carried out more than hundred service jobs, gaining a
lot of practical experience. The NGI Ground Penetrating Radar (GPR) employs an Agilent
E5062 network analyser as the transmitter and receiver. It uses step frequency radar
signals instead of the impulse signals that most of the commercial GPR use. NGI GPR has
been successfully used for over 100 field tests in around 20 countries since 1989. The NGI
radar systems have been exported to USA, South Korea, France, Taiwan, Malaysia, India,
Denmark and Bhutan.

Figure 2: GSB personnel operating the NGI radar system
The advantages of using a network analyser based GPR system are:
The NGI system can work within the frequency band 0.3 MHz to 3000 MHz, which has a
wider bandwidth coverage than most of the commercial GPR systems.
The radar signal bandwidth can be adjusted by software at the test site in order to best
match the ground conditions to the antenna resonance frequency. Conventional GPR
systems, which use impulse signals, need to change hardware to modify radar signals.
The frequency step number can be adjusted by software to make the best use of signal
power for target detection under different ground conditions.
The receiver of the NGI system has a higher processing dynamic range than
conventional GPR systems.
The NGI system has a perfect frequency response, flat and no-dispersion within the
assigned bandwidth. It is in itself a good tool for measuring ground material parameters
and antenna response.
The NGI system makes it simpler to employ frequency-domain signal processing method
such as filtering, deconvolution, Hilbert transform, etc.
Figure 3: Layout of measuring floor at GSB office building where horn antennas of length 10
cm are used

Figure 4: Layout of measuring on Prottasha bridge where a pair of 1.2 m antennas are used
4 Test results
NGI has designed two types of tests for this on site training of GSB personnel. One is the
indoor test to detect re-bars in the concrete and brick wall etc. The other is the outdoor test
on a river dyke. In the indoor test, horn (Transmitting & Receiving) antennas of length 10 cm
have been used (see Figure 3). The outdoor tests use two antennas of length 1.2 m: one for
transmitting the signal and the other for receiving the signal. Both antennas are tied with a
rope with 1 m separation between them (see Figure 4).
4.1 Indoor test results
Figure 5 shows the test result obtained while the radar antennas moved on a concrete floor
(Figure 3). A figure like Figure 5 is called as a radargram. The horizontal axis of the figure is
the trace number or the measurement number. In this test we move 2 cm to perform one
measurement. Therefore in the figure, each horizontal unit is 2 cm. The vertical axis is the
travelling time. In the figure the red arrow points to the time reference corresponding to the
ground surface. The blue arrow points to the time when the reflection from the bottom of the
floor concrete block arrives. The travelling time difference between the reflections from the
ground surface and the bottom of the concrete plate is about t = 2.5 ns. If we assume the
wave speed in the concrete is v =14 cm/ns. The thickness of the concrete block is s = v t/2
= 17.5 cm. According to theory, the reflection from a point target appears as a hyperbolic
curve in a radargram. In the radargram, we find many hyperbolas (yellow arrows in the
figure), which are the reflections from the re-bars inside the concrete floor.

Figure 5: Radargram for the test on 6 th floor ground
Figure 6: Radargram for the test along a verticalwWall of Reshad’s room.
This radargram in Figure 6 has also been collected by 10cm length antenna. Again we use
the red arrow to represent the time reference corresponding to the ground (wall) surface and
the blue arrow to point the time of the reflection from the other side of the wall. The travelling
time between those two reflections is about 3 ns. Assuming the wave speed 12 cm/ns and
using s = v t/2 , we obtain the wall thickness of 18 cm. Here we have used the typical value
of the wave speed in brick material. The actual wave speed can be obtained by a calibration
test on a known thickness material layer. In Figure 6, we have clearly seen the interface
between two layers of bricks (brown arrow).
4.2 GPR tests on dyke
The project area is located in the North and north-western part of the Dhaka city on Turag
dyke.This dyke is approximately 30 kilometres of earthen embankment along Tongi khal,

Turag River, and Buriganga River. A total of 11 regulators at the outfall of khals to the
surrounding rivers along the embankment. One pumping station at the outfall of Kallyanpur
khal to the Turag River, and another one at the outfall of Dholai khal to the Buriganga River.
These pump stations are for draining rainwater from parts of Dhaka West. The main goal of
the test project is to detect risk and flood hazards by details GPR (Ground Penetration
Radar) survey on Turag-Buriganga dyke detecting weak zones due to fracture/cracks,
organic material, and void space etc. in dyke. Locations of test survey areas are shown in
figure 7
4.2.1 GPR survey on Tongi Dyke
The geoscientists of GSB have carried out a GPR survey on Tongi Dyke. They have
selected the sluice gate area for their survey location. The starting and ending positions of
data acquisition are 23º49′48″N, 90º49′44″E to 23º49′44″N, 90º20′39″E. The main purpose
of this survey is to locate the sluice gate zone with GPR. Figure 8 shows a radargram across
the measurement line.
For collecting a radargram in the sluice gate area a pair of 1.2 m length antennas such as
shown in Figure 4 have been used with 1m spacing between receiving and transmitting
antennas.
Figure 7: Location map of 3 test study areas (Sluice Gate, Rostampur and Prottasha Bridge
areas) on the Turag dykes in Dhaka.
In Figure 8 there are four distinct reflections observed in the radargram: the 1 st is the
reflection at the interface between air and layer 1, the 2 nd reflection is at the interface
between layer 1 and layer 2, the third reflection is at the interface between layer 2 and layer
3 and the fourth reflection is at the interface between layer 3 and layer 4.
It is certain from our survey that the uppermost layer (layer 1) represents combine pitch (tar),
brick, brick chips with sand. The layer 2 may represent the pre-existing footpath over which

new pitch road has been made. The third layer may represent compacted soil below the preexisting
footpath. Below the third layer earth fill that is the layer 4 formed by silty clay to clay
materials found with slight variation of materials, compactness and wetness.
Each division of the vertical axis is 20 ns representing a depth of 1m with the assumption of
the wave speed 10 cm/ns. So the approximate depth of layers 1, 2 and 3 are about 0.5 m
each. Due to their thinness, it is very hard to differentiate among pitch, brick and brick chips
zone within the uppermost layer (layer 1). Horn antenna or 60cm length antenna may give
good result for identification of pitch layer in the uppermost layer.
Figure 8: Radargram Over Mirpur Sluice Gate.
Figure 9: The Sluice gate zone in between yellow mark (Red Circle).

In the radargram shown in Figure 9, two sets of anomaly are found. One is between the
trace number 355 to 380 and the other is in between 320 to 340. The anomaly in the red
circle represents the sluice gate. The lines representing the reflections from the ground
layers suddenly end from both side of red circle indicating the existence of the sluice gate.
The anomaly in between traces 320 to 350 shows a different reflection pattern from the other
one. This discontinuity of signal may be due to the beginning part of the sluice gate
structure.
4.2.2 GPR survey on Rostampur area
Figure 10: GPR test layout at Rostampur area
A GPR Survey was performed in Rostampur Cross dyke (23º52′32″ N, 90º21′01″ E) to
identify subsurface lithology (Fig 11) and the test result is shown in Figure 10. In this survey
the 1.2m antenna has been used. Between the traces #1 and #110, three lithological layers
can easily be identified from the radargram. The uppermost unit is mainly composed of brick
chips and silty clay materials with a thickness 0.5m. Below the top layer there is another
layer of thickness 0.5m. The lithology of this layer has not been identified. Below the second
layer the main dam materials silty clay to clay are found. From trace #111 to the end of the
radargram, we can only see two layers.
Figure 11: Test result in Rostampur area.

4.2.3 GPR survey on Prottasha bridge and on highway nearby
Figure 12 shows the GPR survey at Prottasha Bridge (23º54′21″ N, 90º22′35″ E) using the
1.2 m antenna used in all of the outdoor tests.
Figure 12: GPR team in action (test on Prottasha Bridge)
Figure 13: Radargram along connecting road of Prottasha bridge.

Figure 14: Side view of Prottasha bridge and the connecting road
In Figure 13 the uppermost red line represents the response from base of pitch and stone
chips. The 2 nd red line marks the response from base of brick and below this the whole
material might be silty clay.
The area within trace nos. 10-25 marked by the red circle may indicate the junction between
the highway and the bridge connecting road which can be identified by the criss-crossed
nature of the signal. Here, on the sloping side of the highway, new materials have been
dumped for the bridge connecting road construction. From trace no. 133-173 the signal
shows a concave nature. The area within this section might be more compacted compared
to its adjacent sides.
Figure 14 shows the side view of the bridge and the connecting road. The measurement is
taken from the right hand side to the left hand side.
Figure 15: Results of GPR measurement on the highway beside Prottasha bridge

Figure 15 shows the radargram for GPR measurement on the highway beside Prottasha
bridge. From this figure one can clearly see that the ground structure of the section between
two yellow lines (trace number 125 to 265) is different to the adjacent sections at both sides.
This is the section when the GPR moves close to the bridge connecting road. The red lines
show the reflections from a shallow layer (brick to sand interface?). At trace numbers 30, 85,
330, there may be reflections from pipes or cables under the highway.
Figure 16 shows the working environment when testing on the main dike which is a highway.
Figure 16: GPR test on a heavy traffic road connected to Prottasha bridge
4.3 GPR tests on road
GPR Survey was performed at two places on Tongi-Pubail highway. The first survey was
carried out road on Madhupur clay deposits which fall within 23º56′07″ N to 23º56′06″N
Latitude and 90º28′19″ E to 90º28′19″ E Longitude (Figure 17). The second survey was road
on depression deposit in between 23º54′54″ N to 23º54′51″N Latitude and 90º25′28″ E to
90º25′23″ E Longitude (Figure 19). In this survey, 1.2m antenna has been used. The total
length of survey area was about 203m and 133m respectively.
Figure 17: GPR survey, road on Madhupur Clay deposit

Figure 18: Results of GPR measurement road on Madhupur Clay deposit
In Figure 18, there are some layerings at the upper part. These layering are due to the
minute variation of Madhupur Clay. There is a channel fill like deposit in between 75 and 45
trace. The depth of the filling materials is about 12feet. The channel is filled by sandy
material which is confirmed by the depth penetration of the radar. There is also a warping
near to the No 10 trace. It seems that the left side of warping push upward compared to the
right side. The left site of the warping is more sandy than the right side. Presence of warping
in the radar gram indicates that the site has effected by neo-tectonic activity.
Figure 19: GPR survey, road on depression deposits
Figure 20: Results of GPR measurement road on depression deposit

In Figure 20, there is no continuous layering of either sediment or road materials. This may
be due to plastic nature of the underlying sediment. When heavy traffic moves over the road
it initiates a vibration of the underlying materials. Due to its plastic nature (Density lower than
the overlying sediment), it try to move upward (indicated by red arrow) and destroy the
continuous layering of the sediment as well as road materials.
5 GPR theory revisited: Ground material property, wave speed and propagation
attenuation
There are two types of the material properties, which have effects on the electromagnetic
wave propagating inside the material. One is the relative dielectric constant ε r, and the other
is the resistivity. Tables 1 and 2 show respectively the dielectric constants εr and the
resistivities R for common ground materials.
Table 1: Material dielectric constants
Material Dielectric
Constant εr
Air 1 0.3
Sand 4 0.15
Granite 5 0.13
Sandy soil 6 0.12
Wet clay soil 12 0.09
Fresh water 80 0.03
Table 2: Material resistivities
Wave speed
m/ns
At 100 MHz

In Appendix A, we have derived how the wave speed and propagation attenuation are
related to the ground material properties. According to equation A4 in Apendix A, the wave
velocity can be written
v
2 f
k
real
1
0
r
c
r
where c is the speed of light in air and c = 0.3 m/ns. The list of wave speeds in ground
materials shown in Table 1 is calculated by using the above equation.
A general rule is that for detecting deep targets one uses low frequency and long antennas.
Assuming the detection distance is 5 wavelengths and knowing the antenna length is half
wavelength, we obtain that the detection distance is 10 times of the antenna length. The
above empirical relationship can help to choose antenna lengths according to the detection
depth. The resolution is also related to the wavelength used. Here we adapt the criterion that
the detection resolution is ¼ wavelength, i.e., half of the antenna length. Hence our 10 cm
antenna is best suited for detection distance 1 m with a resolution 5 cm, while our 1.2 m
antenna is best suited for detection distance 12 m with a resolution 60 cm.
6 Conclusions
The tests during this on-site training period show that GPR can make quick and continuous
measurements on river dyke etc. GPR can provide high-resolution image of the ground
layering information.
Two types of antennas are used for measurements, one is 10 cm antenna, and the other is
1.2 m length antenna. It is known that on top of the dyke there is 6 cm thick asphalt
carpeting with stone chips. Below the asphalt layer there is 13 cm thick compacted layer of
brick chips mixed with sand (Locals inform maximum thickness of the brick chips layer is
about 40 cm). Below brick chips there is sand. There is no brick soling below the brick chips
layer. GPR test using 1.2 m antenna cannot resolve the layers with thickness 6cm and
13cm. However, Figures 3 and 5 show that using 10 cm antennas can detect those layers.
GPR can be used to detect the material discontinuities along the measurement direction
(e.g., see Figure 13 and 15). Those discontinuity locations may suggest that the ground is
built by different materials or the ground is deformed, and may indicate weak locations of a
dyke.

Kontroll- og referanseside/
Review and reference page
Dokumentinformasjon/Document information
Dokumenttittel/Document title
On site training of using GPR (ground penetrating radar) for river
dyke monitoring in Bangladesh
Dokumenttype/Type of
document
Rapport/Report
Teknisk notat/Technical
Note Oppdragsgiver/Client
Geological Survey of Bangladesh
Distribusjon/Distribution
Fri/Unlimited
Begrenset/Limited
Ingen/None
Emneord/Keywords
Geophysics, Ground Penetrating Radar, river dykes
Stedfesting/Geographical information
Land, fylke/Country, County
Bangladesh
Kommune/Municipality
Dhaka
Sted/Location
Dhaka
Kartblad/Map
UTM-koordinater/UTM-coordinates
Dokument nr/Document
No.
20092021-00-6-R
Dato/Date
22 Nov. 2011
Rev.nr./Rev.No.
0
Havområde/Offshore
area
Feltnavn/Field name
Sted/Location
Felt, blokknr./Field,
Block No.
Dokumentkontroll/Document control
Kvalitetssikring i henhold til/Quality assurance according to NS-EN ISO9001
Rev.
/
Rev.
Revisjonsgrunnlag/Reason for revision
Egenkontroll
/ Self
review
av/by:
Sideman
ns-
kontroll/
Colleagu
e review
av/by:
0 Original document FK RKB LT
Uavheng
ig
kontroll/
Independ
ent
review
av/by:
Tverrfag
lig
kontroll/
Interdisciplin
ary
review
av/by:
Skj.nr. 043

Kontroll- og referanseside/
Review and reference page
Dokument godkjent for
utsendelse/
Document approved for
release
Dato/Date
22/11/2011
Sign. Prosjektleder/Project Manager
Rajinder Bhasin
Skj.nr. 043

Kontroll- og referanseside/
Review and reference page
Skj.nr. 043