Tunnelling and Underground Space Technology 35 (2013) 20–29

Contents lists available at SciVerse ScienceDirect

Tunnelling and Underground Space Technology

journal homepage: www.elsevier.com/locate/tust

Locational accuracy of underground utility mapping using ground penetrating radar

Siow Wei Jaw 1, Mazlan Hashim ⇑
Institute of Geospatial Science & Technology (INSTeG), Faculty of Geoinformation and Real Estate, Universiti Teknologi Malaysia, 81310 UTM Johor Bahru, Malaysia

a r t i c l e i n f o a b s t r a c t

Article history: Many of today’s underground utilities are reaching the end of their practical life and need to be replaced
Received 23 April 2012 or repaired. At the same time, new utility installations due to urban expansion and development of new
Received in revised form 16 September communication technologies such as broadband are in progress. Hence accurate information of these
2012
utilities is essential for utility owners, engineers, contractors or surveyors, particularly as reference for
Accepted 12 November 2012
Available online 3 January 2013
excavation. Ground penetrating radar has been widely used in extracting information of buried utilities
for better utility maintenance and management. The widely used scanning technique (i.e. ‘perpendicular-
to-pipe’ scanning) is limited for retrieving the precise position of targets due to the effects of surrounding
Keywords:
Underground utility mapping
media. This paper provides a solution for the prevention of failed excavations by means of precise under-
Ground penetrating radar ground utility mapping. This paper first of all examines the accuracy of the commonly used data acqui-
Locational accuracy sition scanning technique, by conduction a series of tests, and then developed a better method.
Root mean square error Subsequently, a real-life experiment was carried out to validate the performance of the proposed new
Scanning technique scanning technique, to demonstrate its accuracy and effectiveness. We found that our method was able
to do ‘along-pipe’ scanning with very high precision (i.e. less than +0.10 m, conforming to Quality Level A
utility data). Hence, the proposed method set as a new benchmark for using ground penetrating radar for
precisely locating buried utilities.
Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction lay new infrastructure, without knowing where other subsurface

cables and pipelines are located. This is not only inconvenient to
The demand for basic utility services such as water, gas, elec- the urban dweller, in the form of interruptions in electricity, tele-
tricity and communication are increasing dramatically in response phone, TV and water and gas supplies, but also causes dangerous
to rapid urban growth and development of new communication situations which have resulted in gas explosions and mortalities.
technologies such as broadband. The urban underground of the So, precise 3-D data must be available especially for maintenance,
most densely populated modern cities (United States, United King- repair and laying of new cables and pipelines. Without a well-orga-
dom, China, Japan and etc.) are characterized by many different nized underground spaces planning for supporting surface city life,
types of utility networks including electricity cables, gas pipelines, most of the modern cities will experience serious and unavoidable
fibre optics, water and sanitary sewer pipelines and street lighting problems with growing populations in the future (Aydin, 2008).
circuits. This growing demand for utility services has resulted in The conditions of underground utilities are difficult to assess
much construction of new utility installations, maintenance and from the ground surface (Hao et al., 2012). At present there are
repair systems (Lester and Leonard, 2007). but few publications about the accuracy of underground utility
Many of these utilities were placed under a street or footpath mapping. There are no specific procedures and approaches for
many years ago, and often by different utility companies or organi- underground utility mapping in the form of standard guidelines
zations. Locational data on depth and even maps of where they are that are available to the telecom and other infrastructure indus-
laid are either sparse, unavailable or in best case unreliable. tries involved. For this reason, the ‘‘hit and miss’’ principle is prac-
Although today, as and when new cables/pipes are laid under- ticed widely by the stakeholders in current underground utility
ground, precise 3-D maps are made, this has not been the usual mapping industries. Many failed excavation cases are reported
case. This situation has led to many cases of power interruptions, worldwide every year, where these have caused interruptions of
broken cables and pipelines when one organization is digging to utility services supply, hazardous pipeline explosions and others
serious accidents (Costello et al., 2007; Lester and Leonard, 2007;
Metje et al., 2007; Rogers et al., 2012). In this sense, humans re-
⇑ Corresponding author. Tel.: +60 19 7173861; fax: +60 7 5566163. quire a more detailed understanding regarding the shallow subsur-
E-mail addresses: [email protected], [email protected] face if they are to manage many of the Earth’s finite resources
(M. Hashim). (Neal, 2004). With this regards, comprehensive and accurate infor-
1
Tel.: +60 19 7173861; fax: +60 7 5566163.

0886-7798/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.tust.2012.11.007
 S.W. Jaw, M. Hashim / Tunnelling and Underground Space Technology 35 (2013) 20–29 21

mation of the shallow subsurface is much-needed by utility own- factors influencing accuracy. Subsequently, the results of these
ers, contractors, engineers and even decision makers during the experiments were applied to a real world situation for verification
excavation so as to avoid failed excavations during utility mainte- purposes. The last part of the paper describes the accuracy assess-
nance and rehabilitation. The non-destructive technology namely ment carried out for the detection of underground utilities using
ground penetrating radar (GPR) is therefore an important means proposed GPR scanning technique.
to extract the information of the shallow subsurface, particularly
for detecting and locating the buried utilities (Rubing, 2009). 2. The current status of underground utility mapping
GPR has been widely used in detecting and locating under-
ground utilities due to its many advantages such as: fast data 2.1. Standard guidelines of underground utility mapping
acquisition, cost effectiveness for mapping large areas, better re-
sults compared to other non-destructive technologies and because In each typical underground utility mapping work, the operator
it provides high resolution imagery, for improved interpretation needs to comply with the guidelines provided by the authorities.
(Lester and Leonard, 2007; Millington and Cassidy 2009; Rogers These guidelines mention the role and responsibility of the stake-
et al., 2009; Enes et al., 2010; Jorge et al., 2010; Jeng et al., 2011). holders in underground utility mapping, the quality level of the
According to Jeng et al. (2011) and Ni et al. (2010), GPR is widely utility data, utility map format and design, plus the development
used for examining man-made structures to determine the posi- of an underground utility database. Authorities such as the Depart-
tion and depth of underground utilities. However, to date, no ment of Survey and Mapping Malaysia (JUPEM) (2006), the Amer-
investigations have been conducted to test the accuracy of under- ican Society of Civil Engineers (ASCE) (2002), and the Mapping the
ground utility mapping using GPR. This is one of the main reasons Underworld (MTU) organization are responsible for providing all
for the increased cases of failed excavations, because the accuracy underground utility mapping related guidelines to the relevant
of the underground utility mapping using GPR, has often been industries. According to these guidelines, there are four quality lev-
overlooked by the utility owners or surveyors in a project. This els (A, B, C and D) of utility data which may be collected by differ-
has stifled efforts to develop new underground investigation tech- ent methods such as traditional drawing, on-site investigation or
nologies and the setting up of an underground 3-D cadastral sys- survey and geophysics and trenchless technology (see Fig. 1). The
tem, due to a lack of understanding on current GPR systems, a accuracy, reliability and cost of the utility data increase from Qual-
lack of a standard methodology for data collection and especially ity Level D to A where Quality Level A has the highest accuracy (i.e.
a lack of reliable accuracy assessment (Daniels et al., 2008; Jol, ±10 cm or better in both vertical and horizontal directions). A cen-
2009; Ariaratnam, 2010; Enes et al., 2010; Seyfi and Yaldiz, 2010). tralized underground utility database can be developed based on
Therefore, in order to reduce the number of failures in excava- these utility data of different quality levels. By having accurate
tions of subsurface infrastructure, this paper focuses on the loca- information of the underground utilities, it can avoid some of the
tional accuracy of mapping underground infrastructure using risks of catastrophic damage to underground utilities and adjacent
GPR. The first part of this paper studies the accuracy of three surface areas, as well as reduce the disruption of existing utility
pre-designed data acquisition scanning techniques (i.e. ‘perpendic- services resulting from ‘‘blind’’ excavation work.
ular-to-pipe’, ‘along-pipe’ and ‘variation-angles’ scanning) that are
commonly used in the industry. In doing this, a dual frequency GPR 2.2. Underground utility mapping technologies
system was used to scan underground utilities under laboratory
conditions. These tests were specially set up to understand the role The public-owned urban underground areas: pedestrian path,
of GPR in underground utility mapping, and evaluate the various pavement path, cycle path and road form a complete spider’s

• Review of exisitng records & verbal recollection

Quality Level • Analyse records for availablity of additional information
D

• Surveying & Plotting visible above ground features

Quality Level • Correlate relevant utility records to the surveyed features
C

• Surface Geophysical technique to identify the existence and horizontal

position of the subsurface utilities
Quality Level • Mark & indicate the location of the underground utilities on the ground
B surface for subsequent survey

• Non destructive excavation methods to determine precise horizontal and

vertical positions of subsurface utilities ( ≤ 10 cm)
Quality Level • Survey and exposure of the underground utilities at each specific location
A

Fig. 1. Utility data quality level.

22 S.W. Jaw, M. Hashim / Tunnelling and Underground Space Technology 35 (2013) 20–29

Fig. 2. Formation of hyperbola in GPR imaging.

web network of utility lines. As described by Hao et al. (2012) and For understanding the position of the target from reflected
Rogers et al. (2012), the underground utilities network that serves pulses of the interfaces that recorded by the GPR system, the time
our cities are the most complex in the world, and yet they are difference between the reflected pulses (t1 or t2) are used in con-
invisible from the ground surface. The tasks of locating and map- junction with the contrast in the dielectric properties of the sur-
ping the underground utilities are conducted by using different veyed layers. For computing the thickness of these layers, Eq. (2)
types of trenchless technologies. These include: (i) visual technol- by Al-Qadi and Lahour (2005), Eq. (3) by Shihab and Al-Nuaimy
ogy (closed-circuit television); (ii) sewer scanner and evaluation (2005) and Eq. (4) by Daniels et al. (2008) were used:
technology; (iii) pulsed technology (inductive line location, induc- ct i
tive line tracer, conductive line tracer and passive line tracer); (iv) Thickness; di ¼ pffiffiffiffiffiffi ð2Þ
2 er;i
magnetic technology (magnetic locator and magnetometer); (v)
electromagnetic and radio frequency technology (ground penetrat-  c 2
ing radar, magnetic flux leakage, hydroscope and etc.); (vi) technol- where er ¼ ð3Þ
ogy resistivity (resistivity locator) and (vii) acoustic technology
v
(acoustic pipe tracer, sonar and pipe cable locator) which are represents the electric permittivity of the materials (er)
widely used for locating these underground utilities without exca- c
vation (Cist and Schutz, 2001; Rogers et al., 2009; Ni et al., 2010; and v ¼ qffiffiffiffi
e
ð4Þ
m
Hao et al., 2012). Of all these technologies mentioned above, GPR eo
is the top choice technology for the subsurface investigation of
represents the electromagnetic wave travel velocity; di represents
man-made structures (He and Hiroshi, 2007; Rogers et al., 2009;
layer thickness; ti represents two-way travel time of the reflection;
Ni et al., 2010). For this reason, GPR was used in this study to ac-
c is the speed of light in free space (3  108 m/s); er,i represents
quire data of buried utilities in order to ascertain their accuracy
dielectric constant for each layer; em represents the material’s
for 3-D mapping.
dielectric permittivity and eo represents the free space permittivity.
In the operation of a GPR system, the transmitter transmits high
peak voltage non-sinusoidal electromagnetic (EM) pulses to the
underground. When the signal hits a target, it will reflect back to 3. Materials and methods
the receiver antenna. These reflected signals are known as a radar
trace. However, these reflected signals always contain unwanted 3.1. Data acquisition
echoes caused by heterogeneous media such as sand, clay, rock,
gravel and utilities. As such, these media appear as black and white For conducting this study, IDS DetectorDuo2 GPR system was
streaks in the radargram. For cylindrical targets such as pipelines used for data acquisition due to its optimal frequency (250 MHz
in underground utilities, they will appear in the form of black and 700 MHz) for locating the actual position of utilities through real
and white streaks having the shape of a hyperbola in the radar- time interpretation and automatic calibration for different soil type
gram. Fig. 2, shows how the hyperbola is formed when the GPR parameters (Ingegneria dei Sistemi S.p.A., 2007). This GPR system
scans over a line (from Xa to Xb), by connecting the end points of was used to acquire data over two different test beds (i.e.
the lines (YJ, YK, and YJ) that are orthogonal to the antenna’s tra- JUPEM_TestBed and UTM_TestBed) using three pre-designed scan-
jectory (XJ, XK, and XJ), with the antenna located at the position ning techniques, namely: (i) ‘perpendicular-to-pipe’ scanning; (ii)
of (x1, y1, z1) as it moves past this single scanline. The shortest line ‘along-pipe’ scanning; and (iii) ‘variation-angles’ scanning. These test
(YK in Fig. 2) represents the depth of the target. For computing the beds contain different types and sizes of utilities that are commonly
depth, Eq. (1) proposed by Gordon et al. (1998) and Motoyuki used in the utility industries. These test sites were specifically built
(2009) is used: to mimic the underground infrastructure in the real world, as well as
for calibrating the GPR system, and for understanding the capability
mt of GPR for underground utility mapping. For our experimental pur-
Depth; D ¼ ð1Þ poses, the utilities were buried at different depths as shown by the
2
arrangement depicted in Fig. 3a and b respectively.
where D is depth; m represents velocity of the electromagnetic
wave; and t is the two-way travel time of the reflection. 2
Trademarks of Ingegneria dei Sistemi S.p.A.
 S.W. Jaw, M. Hashim / Tunnelling and Underground Space Technology 35 (2013) 20–29 23

Fig. 3. Utility details of (a) JUPEM_TestBed and (b) UTM_TestBed.

3.2. Data pre-processing inaccuracies in the depth scale of the radargram. For this reason,
the start time of the transmitted signal is removed according to
The raw data acquired by the GPR system were pre-processed to the optimum velocity acquired from hyperbola fitting and hence
remove the unwanted echoes in order to produce a better focused the actual depth of the target is acquired. This produces a focused
radargram before further interpretation. In this sense, the data image with a better signal-to-noise ratio (SNR). However, the
were initially aligned to the depth scale of the radargram according radargram still contains some unwanted echoes due to ringing
to the actual measurement made from the ground surface. This is noise caused by non-target features such as rocks, cavities or soil.
because the recorded waveforms are always influenced by the re- By using the Clear-X filtering technique, we can suppress the un-
sponse from the surrounding medium and yet the medium’s tran- wanted background clutter (or noise) which obscured the reflec-
sition zone is separated from the real data, which causes tion of the actual target. According to Kim et al. (2007), the
24 S.W. Jaw, M. Hashim / Tunnelling and Underground Space Technology 35 (2013) 20–29

Fig. 4. Scanlines for data acquisition at (a) JUPEM_TestBed and (b) UTM_TestBed.
 S.W. Jaw, M. Hashim / Tunnelling and Underground Space Technology 35 (2013) 20–29 25

Table 1
Comparison of the pre-designed scanning techniques.

Experiment Number of targets detected RMSE values (m) Penetration depth (m)
Planimetric (x, y) Depth (z)
JUPEM_TestBed
Perpendicular-to-pipe scanning 6/9 ±0.109 ±0.107 <1.90
Along-pipe scanning 4/9 ±0.080 ±0.075 <1.82
Variation-angles scanning 1/9 ±0.118 ±0.117 <1.15
UTM_TestBed
Perpendicular-to-pipe scanning 5/5 ±0.112 ±0.109 <1.30
Along-pipe scanning 4/5 ±0.086 ±0.070 <1.15
Variation-angles scanning 3/5 ±0.113 ±0.115 <1.15

Clear-X filter is useful for removing unwanted continuous compo- formed to enhance the visual quality of the radargram. This image
nents along the X-axis because the noisy components regularly enhancement of the radargram was made by using this ‘‘clearing
happen in the horizontal and periodic direction. After the noisy filter’’, according to the region specified by the user, based on prior
background clutters were removed, a bandpass filter was per- knowledge of the target. The regions with frequency either lower

(a) Perpendicular to pipe scanning

(b) Along pipe scanning

(c) Variation angles scanning

*Note: grainy textures reflect to existence of high moisture contents due to shallow water
table (bottom part of radargram)
Fig. 5. Results of utility detection for the JUPEM_TestBed: (a) perpendicular to pipe scanning, (b) along pipe scanning, and (c) variation angles scanning.
26 S.W. Jaw, M. Hashim / Tunnelling and Underground Space Technology 35 (2013) 20–29

(a) Perpe ndicular to pipe

i scanningg (b) Along pipe scanningg

*Notee: grainy ttextures refflect to

existeence of highh moisture coontents
due too shallow wwater table (bbottom
part oof radargram
m)

(c) Varriation angles scanning

Fig. 6. Results of utility detection for the UTM_TestBed: (a) perpendicular to pipe scanning, (b) along pipe scanning, and (c) variation angles scanning.

or higher than the GPR signal bandwidth are hereby removed. 3.3. Feature detection
Lastly, the structure of the image was altered, based on the applied
gain function. Linear and smoothed gains were used in this study The real reflections of the buried utilities were extracted from
based on a multiplication or mathematical operator defined by the pre-processed radargram based on hyperbola reflection. Then,
the system itself. These pre-processing steps are based on the the position and depth of each target were retrieved for coordinate
author’s personal experience. In fact, users can tailor their own transformation purposes. The actual underground position for each
processing routines to suit their specific needs, using different target was computed by referring to the actual ground coordinate
GPR systems, according to their own data interpretation obtained using a real time kinematic (RTK) and global positioning
experience. system (GPS). After that, 10 sample points were collected from
 S.W. Jaw, M. Hashim / Tunnelling and Underground Space Technology 35 (2013) 20–29 27

each detected utility based on based on the Nyquist sampling cri- The direction of scanning is shown in Fig. 4. In order to determine
teria to assess the accuracy as shown in Section 3.4 (Thomas and the orientation of the buried utilities, a pipe cable locator was used.
Mike, 2009). Then, these data were pre-processed and the feature detection
steps as discussed in Sections 3.2 and 3.3 were carried out, prior
3.4. Accuracy assessment to the assessment.
Based on the detected utilities, we computed the overall accu-
In this study, the three criteria used for assessment were (i) root racy for all the sample points (10 sample points selected from each
mean square error (RMSE); (ii) target detectability; and (iii) pene- detected utility) collected from both test beds. According to the
tration power. For the accuracy assessment, we computed the root RMSE results for both the JUPEM_TestBed and the UTM_TestBed,
mean square error (RMSE) for both planimetric position and depth the along-pipe scanning yielded the highest accuracy (i.e. lowest
observed by the GPR system. By calculating the RMSE for both pla- RMSE) data in terms of both planimetric position and depth, as
nimetric position and depth, the achievable accuracy of the GPR compared to the other two techniques (see Table 1). The results
system for underground utility mapping is known. In this sense, prove that data acquired using the along-pipe scanning technique
Eqs. (5) and (6) proposed by Gonçalves et al. (2006) and Reyes is the best, as its accuracy is equivalent to Quality Level ‘‘A’’ utility
et al. (2010) were used for the RMSE for planimetric (RMSExy) data (i.e. accuracy better than ±0.10 m for both planimetric posi-
and depth (RMSEz), respectively: tion and depth). Based on the world standard guideline for utility
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi data quality level (see Fig. 1), any data acquired using non-destruc-
XN   tive technology with accuracy better than 10 cm is categorized as
RMSExy ¼ 1=N rx2i þ ry2i ð5Þ
i¼1 Quality Level ‘A’ (American Society of Civil Engineers (ASCE),
2002; James, 2003; Department of Survey and Mapping Malaysia
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
XN  (JUPEM), 2006).
RMSEz ¼ 1=N i¼1
jZ o  Zj2 ð6Þ Furthermore, we also found out that the along-pipe scanning
technique has the best target detectability as it detected the most
where N represents the number of points observed; (rxi, ryi) is the buried utilities in both test beds (see Figs. 5 and 6). In this sense,
residual of a sample point; Zo is the observed depth and Z is the the along-pipe scanning technique once again proved to be the
computed depth. After computing the RMSE for each point, the dif- best data acquisition scanning technique, because of its good target
ferences between the variance of the observed and the computed detectability. In addition, the along-pipe scanning technique also
planimetric positions and depths were assessed using analysis of has the best penetration depth capability, as shown in Table 1.
variance (ANOVA). The reliability between the observed value from Nevertheless, the penetration depth for all three scanning tech-
GPS and computed value from GPR system was then tested. niques were basically limited to 2 m, caused by the existence of
Target detectability and penetration power were analyzed using the high moisture content of the medium, due to the shallow water
the three pre-designed data acquisition scanning techniques table in the area. The comparison of the three parameters for the
namely: (i) ‘perpendicular-to-pipe’ scanning; (ii) ‘along-pipe’ scan- pre-designed scanning techniques is summarized in Table 1.
ning; and (iii) ‘variation-angles’ scanning. For perpendicular-to-
pipe scanning, the data were acquired in the direction perpendicu-
lar to the orientation of the buried utilities. While for the along- 4. Discussions
pipe scanning, the data were acquired in the direction parallel to
the orientation of the buried utilities, variation-angles scanning The findings regarding the detection accuracy of GPR using the
was used to acquire data in the direction according to the angles different data acquisition scanning techniques mentioned above
(30°, 45°, 60° and 90°) to the orientation of the buried utilities. were re-confirmed by the analysis of variance (ANOVA). ANOVA

Presence of three different hyperbolic patterns in a GPR image

Gas pipe Electrical cable Water reticulation pipe

Fig. 7. Case study based on theoretical data using GprMax simulator.

28 S.W. Jaw, M. Hashim / Tunnelling and Underground Space Technology 35 (2013) 20–29

Fig. 8. Results of utility detection under real world conditions.

was conducted to determine the relationship between the variance 5. Conclusion

of these pre-designed scanning techniques in order to prove that
survey line orientation and location affects the accuracy of target This research was carried out both in the laboratory, and in
detection. The finding showed that there are significant differences some real world examples, to find an optimum solution for precise
(p = 0.002) for both planimetric position and depth among these underground utility mapping. Through this study, the accuracy of
three scanning techniques. This confirms that different scanning GPR for 3-D underground utility mapping has been presented,
technique yields different data acquisition accuracy level. This and its accuracy is evaluated by use of RMSE and sample paired
has been proven in the work of Jaw and Hashim (2011) and Jorge t-test. We found that data acquisition using the along-pipe scan-
et al. (2010), as they both emphasized that it is important to select ning technique yields good results for underground utility map-
the most suitable scanning technique. Different scanning tech- ping. The data acquired using this technique provides ±0.10 m
niques come along with different surface antenna arrangements accuracy, which is equivalent to the Quality Level A utility data
which have the potential to improve underground images as as specified in the universal standard guidelines for underground
shown in these studies. utility mapping. By applying good practice of data acquisition as
In order to improve our ability to understand the radar’s detec- proposed in this study, it is useful and effective for collecting high
tion mechanism and obtain more information from GPR data, more precision 3-D subsurface utility data. This is significant in minimiz-
sophisticated data analysis tools were used. Hence, the assessment ing the risks of failed excavation. Therefore, this work has set a new
of the accuracy of the scanning techniques was aided by numerical benchmark related to good practice of GPR data acquisition tech-
experiments conducted on synthetic GPR images generated by niques for underground utility mapping industries. With continu-
means of the electromagnetic simulator GprMax3 developed on ous, rapid improvements in GPR hardware and software, we are
the basis of the Finite Difference Time Domain (FDTD) numerical confident that the use of this method will increase greatly in the
method (Giannopoulos, 2005). In order to study the limitations of near future.
the perpendicular-to-pipe scanning technique, we have reported a
case where it was difficult to distinguish multiple utilities that are
Acknowledgments
buried closely together (see Fig. 7). The GPR image clearly depicts
the presence of three different hyperbolic patterns, which, however,
This study was conducted as part of the E-Science Grant
cannot be easily detected as the pattern associated with these three
Scheme, Ministry of Science, Technology & Innovation, Malaysia.
adjacent buried objects is obscured by the patterns generated by
We are also thankful to the Universiti Teknologi Malaysia and
other neighbours. This suggests that the perpendicular-to-pipe scan-
Department of Survey and Mapping, Malaysia for providing the
ning technique cannot effectively handle multiple objects that bur-
facilities for this investigation. Cooperations from RDG Supply
ied closely together.
Sdn. Bhd. is also acknowledge.
By using these reliable findings, we then carried out an experi-
ment under real world conditions at Persiaran Kewajipan, Subang
Jaya, Selangor Darul Ehsan, Malaysia. In this experiment, we suc- Appendix A. Supplementary material
cessfully detected almost all the buried utilities (6 out of 8) in these
areas using the along-pipe scanning data acquisition technique as Supplementary data associated with this article can be found, in
recommended above (see Fig. 8). These results were further veri- the online version, at http://dx.doi.org/10.1016/j.tust.2012.11.007.
fied by on-site trial hole investigations and using ancillary infor- These data include Google map of the most important areas de-
mation such as the blueprint of the underground utilities scribed in this article.
provided by the utility owners. In this actual real world experi-
ment, the accuracy that we obtained was ±0.098 m (x, y) and References
±0.095 m (z). As such, this research has proven that the suggested
scanning technique is good for practices in underground utility Al-Qadi, I.L., Lahouar, S., 2005. Measuring layer thickness with GPR – theory to
practice. Journal of Construction and Building Materials 19, 763–772.
mapping to obtain high accuracy utility data, which satisfies the
American Society of Civil Engineering (ASCE), 2002. Standard Guideline
demanding requirement of the Quality Level ‘‘A’’, as specified by for the Collection and Depiction of Existing Subsurface Utility Data.
the American Society of Civil Engineers (ASCE) (2002). ASCE Code and Standards Activity Committee (CSAC), New York, US,
pp. 4–6.
Ariaratnam, S.T., 2010. Survey questionnaire results of the current level of
knowledge on trenchless technologies in China. Journal of Tunnelling and
3
Created by Antonis Giannopoulos (Giannopoulos, 2005). Underground Space Technology 25, 802–810.
 S.W. Jaw, M. Hashim / Tunnelling and Underground Space Technology 35 (2013) 20–29 29

Aydin, C.C., 2008. Usage of underground space for 3D cadastre purpose and related Kim, J.H., Cho, S.J., Yi, M.J., 2007. Removal of ringing noise in GPR data by signal
problems in Turkey. Sensor 8, 6972–6983. processing. Journal of Geosciences 11, 75–81.
Cist, D.B., Schutz, A.E., 2001. State of The Art for Pipe & Leak Detection. Geophysical Lester, J., Leonard, E.B., 2007. Innovative process to characterize buried utilities
Survey System, Inc., Salem, New Hampshire, USA, pp. 2–8. using ground penetrating radar. Journal of Automation in Construction 16, 546–
Costello, S.B., Chapman, D.N., Rogers, C.D.F., Metje, N., 2007. Underground asset 555.
location and condition assessment technologies. Journal of Tunnelling and Metje, N., Atkins, P.R., Brennan, M.J., Chapman, D.N., Lim, H.M., Machell, J.,
Underground Space Technology 22, 524–542. Muggleton, J.M., Pennock, S., Ratcliffe, J., Redfern, M., Rogers, C.D.F., Saul, A.J.,
Daniels, J., Ehsani, M.R., Allerd, B.J., 2008. Ground-Penetrating Radar Methods (GPR). Shan, Q., Swingler, S., Thomas, A.M., 2007. Mapping the underworld – state-of-
Taylor and Francis Group, LLC, USA, pp. 129–145. the-art review. Journal of Tunnelling and Underground Space Technology 22,
Department of Survey and Mapping Malaysia (JUPEM), 2006. Standard Guideline for 568–586.
Underground Utility Mapping. Department of Survey and Mapping Malaysia, Millington, T.M., Cassidy, N.J., 2009. Optimising GPR modelling: a practical, multi-
Kuala Lumpur, Malaysia, pp. 4–7. threaded approach to 3D FDTD numerical modelling. Journal of Computers and
Enes, Y., Sevket, D., Caner, O. 2010. On the imaging application of ground Geosciences 36, 1135–1144.
penetrating radar. In: Proceedings of the URSI International Symposium on Motoyuki, Sato., 2009. Principles of Mine Detection by Ground-penetrating Radar.
Electromagnetic Theory, 16–19 August, Berlin, Germany. Springer-Verlag, London, pp. 19–26.
Giannopoulos, A., 2005. Modelling ground penetrating radar by GprMax. Journal of Neal, A., 2004. Ground-Penetrating Radar and Its Use in Sedimentology: principle,
Construction and Building Materials 19, 755–762. problems and progress. Journal of Earth Science Reviews 66, 261–330.
Gonçalves, H., Gonçalves, J.A., Luís, C., 2006. Measurement for an objective Ni, S.H., Huang, Y.H., Lo, K.F., Lin, D.C., 2010. Buried pipe detection by ground
evaluation of the geometric correction processing quality. IEEE Geoscience penetrating radar using the discrete wavelet transform. Journal of Computers
and Remote Sensing Letters 6, 292–296. and Geotechnics 37, 440–448.
Gordon, M.O., Broughton, K., Hardy, M.S.A., 1998. The assessment of the value of Reyes, C., Hilaire, T., Paul, S., Mecklenbräuker, C.F. 2010. Evaluation of the root mean
GPR imaging of flexible pavements. Journal of NDT and E International 31, 429– square error performance of the PAST-consensus algorithm. In: Proceedings of
438. the International ITG Workshop on Smart Antennas, 23–24 February, Bremen,
Hao, T., Rogers, C.D.F., Metje, N., Chapman, D.N., Muggleton, J.M., Foo, K.Y., Wang, P., Germany.
Pennock, S.R., Atkins, P.R., Swingler, S.G., Parker, J., Costello, S.B., Burrow, M.N.P., Rogers, C.D.F., Chapman, D.N., Entwisle, D., Jones, L., Kessler, H., Metje, N., Mica, L.,
Anspach, J.H., Armitage, R.J., Gohn, A.G., Goddard, K., Lewin, P.L., Orlando, G., Morey, M., Pospisil, P., Price, S., Raclausky, J., Scott, H., Thomas, A.M. 2009.
Redfern, M.A., Royal, A.C.D., Saul, A.J., 2012. Condition assessment of the buried Predictive mapping of soil geophysical properties for GPR utility location
utility service infrastructure. Journal of Tunnelling and Underground Space surveys. In: Proceedings of the 5th International Workshop on Advanced
Technology 28, 331–344. Ground Penetrating Radar, 27–29 May, Granada, Spain.
He, Y., Hiroshi, M., 2007. Extraction method for ground penetrating radar. PIERS Rogers, C.D.F., Hao, T., Costello, S.B., Burrow, M.N.P., Metje, N., Chapman, D.N.,
Online 3, 701–703. Parker, J., Armitage, R.J., Anspach, J.H., Muggleton, J.M., Foo, K.Y., Wang, P.,
Ingegneria dei Sistemi S.p.A., 2007. DETECTORDUO System User Manual. Pisa, Italy, Pennock, S.R., Atkins, P.R., Swingler, S.G., Gohn, A.G., Goddard, K., Lewin, P.L.,
pp. 6–7. Orlando, G., Redfern, M.A., Royal, A.C.D., Saul, A.J., 2012. Condition assessment of
James, H.A., 2003. New National Utility Standards & Guidelines from AASHTo. ASCE the buried utility service infrastructure – a proposal for integration. Journal of
and FHWA. So-Deep, Inc., Manassas Park, Virginia, pp. 1–6. Tunnelling and Underground Space Technology 28, 331–344.
Jaw, S.W., Hashim, M. 2011. Accuracy of data acquisition approached with ground Rubing, Ge, 2009. New progress of GPR to detect underground pipelines. In:
penetrating radar for subsurface utility mapping. In: Proceedings of 2011 IEEE Proceedings of the International Conference on Pipelines and Trenchless
International RF and Microwave (RFM 2011), 12–14 December, Seremban, Technology, 18–21 October, Shanghai, China.
Malaysia. Seyfi, L., Yaldiz, E., 2010. A novel software for an energy efficient GPR. Journal of
Jeng, Y., Lin, C.H., Li, Y.W., Chen, C.S., Yu, H.M., 2011. Application of sub-image Advances in Engineering Software 41, 1195–1199.
multiresolution analysis of ground penetrating radar data in a study of shallow Shihab, S., Al-Nuaimy, W., 2005. Radius estimation for cylindrical objects detected
structure. Journal of Applied Geophysics 73, 251–260. by ground penetrating radar. Journal of Subsurface Sensing Technologies and
Jol, H.M., 2009. Ground Penetrating Radar: Theory and Application, 1st ed. Elsevier Applications 6, 151–165.
Science, Netherlands, UK, pp. 141–172. Thomas, B., Mike, E.D., 2009. Sampling theorems for signals from the union of finite-
Jorge, L.P., Slob, E., Robson, S.L., Leite, D.N., 2010. Comparing detection and location dimensional linear subspaces. IEEE Transactions on Information Theory 55,
performance of perpendicular and parallel broadside GPR antenna orientation. 1872–1882.
Journal of Applied Geophysics 70, 1–8.