**Utilization of Ground-Penetrating Radar and Frequency Domain Electromagnetic for Investigation of Sewage Leaks**

Goldshleger Naftaly and Basson Uri

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/62156

#### **Abstract**

Fact 1: Underground sewage pipe systems deteriorate over time, developing cracks and joint defects; therefore, leakage is inevitable. Fact 2: The massive worldwide urbanization process, together with rural development, has meaningfully increased the length of sew‐ age pipelines. Result: The concomitant risk of sewage leaks exposes the surrounding land to potential contamination and environmental harm. It is therefore important to locate such leaks in a timely manner, enabling damage control. Advances in active remote-sens‐ ing technologies (GPR and FDEM: ground-penetration radar and frequency domain elec‐ tromagnetic) were used to identify sewage leaks that might cause pollution and to identify minor spills before they cause widespread damage.

**Keywords:** Active remote sensing, FDEM, GPR, Sewage leak, Contamination, Water pol‐ lution

#### **1. Introduction**

Water pollution is the contamination of bodies of water such as aquifers, lakes, ponds, rivers and oceans. This contamination occurs due to direct or indirect discharge of pollutants into the water bodies, without a suitable treatment to remove harmful compounds (pollutants may simply be defined as substances added to the environment that do not belong there). A substantial proportion of water and environmental contaminants are due to leaks from underground sewage pipeline systems in rural, urban and industrial areas, since any sewage pipeline system deteriorates over time, developing cracks and joint defects. Therefore, if sewage pipeline systems are not maintained properly, it is only a matter of time before the sewage leaks out and contaminates the surrounding groundwater and surface water.

© 2016 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Here, we suggest detecting sewage leaks from pipeline systems using two orthogonal active remote-sensing methods: (I) ground-penetrating radar (GPR) and (II) frequency domain electromagnetic (FDEM). Our hypothesis is that GPR and FDEM screening, which creates subsurface images around and along pipeline systems, will enable the extraction of residual signals and the detection of meaningful leaks. Like most complex near-surface detection missions, detection of sewage leaks in an urban environment requires a professional under‐ standing of the regional setting, from geomorphological, environmental and engineering perspectives.

Advances in remote-sensing technologies now enable their use to identify leakage that is potentially responsible for pollution and to identify minor spills before they can cause widespread damage. The detection of pollutants using GPR [1], was based on the research of Basson [2]. Basson [3] presented a combination of GPR and FDEM methods to detect and monitor saline contaminants in agricultural fields. Goldshleger [4, 5] demonstrated the ability to detect saline-affected soils using remote-sensing methods, toward improved management of these soils. Basson [6] described the detection of subsurface water/sewage/drainage pipe systems and leaks/contamination from such pipes. Ben-Dor [7, 8] reviewed remote-sensing– based methods to assess soil salinity and improve the management of salinity-affected soils. Ly and Chui [9] developed accurate representations of weep holes and leaky sewage pipes, and further showed the systems' long-term and short-term responses to rainfall events. Their simulation results provided a better understanding of local-scale migration of sewage leaks from a sewage pipe to nearby storm water drains. The last few years in Israel have seen increasing use of new methods based on active remote-sensing tools to study subsoil quality. These tools include GPR and underground monitoring systems measuring spatial moisture content, such as FDEM in the subsurface. The use of GPR is based on a method that was originally developed for measuring sand dunes of medium moisture content at an unsaturated resolution of a few percentage points [2]. The GPR helped define the possible reason for emerging high-salinity areas, such as a subsurface regional structure that reduces water infiltration into the deeper groundwater position [5]. The FDEM method provided a very important view of salt contamination in the soil layers (except the root zone layer) and also pinpointed areas with salinity problems. The images obtained from FDEM readings provided a subsurface view that also helped identify the reason for the high salinity in certain areas. In the soil salinity experiment in Israel, a severe defect in the drainage pipelines could be observed, which helped the farmers solve the problem before the subsequent season [5].

The present study focuses on the development of these electromagnetic (EM) methods to replace conventional acoustic methods for the identification of sewage pipe leaks. EM methods provide an additional advantage in that they allow mapping the fluid transport system in the subsurface. Leak-detection systems using GPR and FDEM are not limited to large amounts of water, but can also detect leaks of tens of liters per hour, because they can locate increases in pipes' or tanks' environmental moisture content that amount to only a few percentage points. The importance and uniqueness of this research lies in the development of practical tools to provide a snapshot of the spatial changes in soil moisture content to depths of about 3–4 m (in areas with asphalt overlay) at relatively low cost, in real time or close to real time. Spatial measurements performed using GPR and FDEM systems allow monitoring many tens of thousands of measurement points per hectare, thus providing a picture of the spatial situation along the pipelines. The main purpose of this study was to develop a method for detecting sewage leaks using the above-proposed geophysical methods, as the resultant contaminants can severely affect public health. We focused on identifying, locating and characterizing such leaks in sewage pipes in residential and industrial areas. 4

> detecting s the resulta n identifyin ntial and in

sewage lea ant contam ng, locating dustrial are

ks using th minants can g and char eas.

he above severely acterizing

eptibility (F

FDEM).

#### **2. Methods** 96 affec 97 such

94 study 95 prop

y was to de osed geop ct public he h leaks in se

evelop a m physical me ealth. We ewage pipe

method for d ethods, as focused on es in residen

ile and mag

uctivity profi

trical condu

102 elect

100 101

102 103

Here, we suggest detecting sewage leaks from pipeline systems using two orthogonal active remote-sensing methods: (I) ground-penetrating radar (GPR) and (II) frequency domain electromagnetic (FDEM). Our hypothesis is that GPR and FDEM screening, which creates subsurface images around and along pipeline systems, will enable the extraction of residual signals and the detection of meaningful leaks. Like most complex near-surface detection missions, detection of sewage leaks in an urban environment requires a professional under‐ standing of the regional setting, from geomorphological, environmental and engineering

Advances in remote-sensing technologies now enable their use to identify leakage that is potentially responsible for pollution and to identify minor spills before they can cause widespread damage. The detection of pollutants using GPR [1], was based on the research of Basson [2]. Basson [3] presented a combination of GPR and FDEM methods to detect and monitor saline contaminants in agricultural fields. Goldshleger [4, 5] demonstrated the ability to detect saline-affected soils using remote-sensing methods, toward improved management of these soils. Basson [6] described the detection of subsurface water/sewage/drainage pipe systems and leaks/contamination from such pipes. Ben-Dor [7, 8] reviewed remote-sensing– based methods to assess soil salinity and improve the management of salinity-affected soils. Ly and Chui [9] developed accurate representations of weep holes and leaky sewage pipes, and further showed the systems' long-term and short-term responses to rainfall events. Their simulation results provided a better understanding of local-scale migration of sewage leaks from a sewage pipe to nearby storm water drains. The last few years in Israel have seen increasing use of new methods based on active remote-sensing tools to study subsoil quality. These tools include GPR and underground monitoring systems measuring spatial moisture content, such as FDEM in the subsurface. The use of GPR is based on a method that was originally developed for measuring sand dunes of medium moisture content at an unsaturated resolution of a few percentage points [2]. The GPR helped define the possible reason for emerging high-salinity areas, such as a subsurface regional structure that reduces water infiltration into the deeper groundwater position [5]. The FDEM method provided a very important view of salt contamination in the soil layers (except the root zone layer) and also pinpointed areas with salinity problems. The images obtained from FDEM readings provided a subsurface view that also helped identify the reason for the high salinity in certain areas. In the soil salinity experiment in Israel, a severe defect in the drainage pipelines could be observed, which helped the farmers solve the problem before the subsequent season [5].

The present study focuses on the development of these electromagnetic (EM) methods to replace conventional acoustic methods for the identification of sewage pipe leaks. EM methods provide an additional advantage in that they allow mapping the fluid transport system in the subsurface. Leak-detection systems using GPR and FDEM are not limited to large amounts of water, but can also detect leaks of tens of liters per hour, because they can locate increases in pipes' or tanks' environmental moisture content that amount to only a few percentage points. The importance and uniqueness of this research lies in the development of practical tools to provide a snapshot of the spatial changes in soil moisture content to depths of about 3–4 m (in areas with asphalt overlay) at relatively low cost, in real time or close to real time. Spatial

perspectives.

262 Environmental Applications of Remote Sensing

In recent years, there has been an increase in the use of active remote-sensing tools, such as GPR (Figure 1a) and subsurface FDEM (Figure 1b), for measuring the subsurface's EM velocity and dielectric constant (GPR), and its electrical conductivity profile and magnetic susceptibility (FDEM). 95 **2. Me** 99 In re 100 tools 101 meas **ethods** cent years, s, such as suring the , there has s GPR (Fi subsurface been an in gure 1a) e's EM velo crease in t and subsu ocity and di he use of a urface FDE electric con active remo EM (Figure nstant (GPR te-sensing e 1b), for R), and its

gnetic susc

(b)

**Figure 1.** Taking measurements with the RAMAC GPR (a) and Gem-2 FDEM (b) in the study area.

Passive remote-sensing spectroscopy of ground surface and cross-sections using an optical fiber termed SPSP (subsurface-penetrating spectral probe), developed [10] and have been conducted as well. This study focuses on remote-sensing tools to replace acoustic methods [11, 12, 13]. EM methods provide the added advantage of being able to map underground liquidcarrying pipelines. Ground leak-detection systems using GPR and FDEM are not limited to large amounts of water: small leaks of tens of liters per hour can be detected in the environment by comparing medium-dry to minimum moisture content in the pipeline and the canal zone.

Our aim was to develop practical tools that would provide a snapshot of changes in spatial soil moisture content to depths of about 3–4 m in areas covered with asphalt at relatively low cost and in real time. The spatial measurements were performed with FDEM and GPR systems that allow measuring tens of thousands of points per hectare and thus enable monitoring the spatial situation along the pipeline.

#### **2.1. FDEM**

Traditionally, the electrical method "measures" apparent resistivity using electrodes that require ground contact in a DC electrical survey, while the EM method "measures" apparent conductivity without ground contact. The EM method, known as a "potential method", involves transmitting and receiving EM fields, commonly using a set of coils. The common unit of resistivity is ohm-m and conductivity is its inverse, in Siemen/m. The apparent resistivity *ρa* is defined in DC resistivity as:

$$
\rho\_a = 2\pi G \frac{\Delta V}{I} \tag{1}
$$

where *ΔV* is the voltage between a pair of potential electrodes, *I* is the current that flows through another pair of source electrodes, and *G* is the geometric factor that depends on the geometry of the electrodes. For a Wenner array that uses four equally spaced electrodes, for instance, *G* is the electrode spacing itself. Even for this simple array, each electrode spacing generates a different apparent resistivity because the spacing controls the volume of the subsurface sampled by the measurement. It is only when the earth is a homogeneous half space that the apparent resistivity is the same as the true resistivity.

Similarly, apparent conductivity is only same as the true conductivity when the earth is a homogeneous half space. As an example, consider a pair of horizontal coils separated by a distance *r.* A routinely measured quantity is called the *mutual coupling ratio* which, for a horizontal coplanar (or vertical dipole) coil configuration over a layered earth as derived by [14, 15, 16, 17], among others is written as:

$$\mathbf{Q} = \frac{H\mathbf{s}}{H\mathbf{p}} = -r^3 \int\_0^\eta \lambda^2 R\left(\lambda\right) f\_0\left(\lambda r\right) e^{-\lambda h} d\lambda \tag{2}$$

*H*p and *H*s are the primary and secondary fields at the receiver coil; *J*0 is the 0th order Bessel function; *r* is the coil separation and *h* is the sensor height above the ground. *Q* represents the secondary field normalized against the primary field at the receiver coil. Most frequencydomain sensors measure *Q* in parts per million (ppm). The kernel *R* corresponding to a homogeneous half space is:

Passive remote-sensing spectroscopy of ground surface and cross-sections using an optical fiber termed SPSP (subsurface-penetrating spectral probe), developed [10] and have been conducted as well. This study focuses on remote-sensing tools to replace acoustic methods [11, 12, 13]. EM methods provide the added advantage of being able to map underground liquidcarrying pipelines. Ground leak-detection systems using GPR and FDEM are not limited to large amounts of water: small leaks of tens of liters per hour can be detected in the environment by comparing medium-dry to minimum moisture content in the pipeline and the canal zone. Our aim was to develop practical tools that would provide a snapshot of changes in spatial soil moisture content to depths of about 3–4 m in areas covered with asphalt at relatively low cost and in real time. The spatial measurements were performed with FDEM and GPR systems that allow measuring tens of thousands of points per hectare and thus enable monitoring the

Traditionally, the electrical method "measures" apparent resistivity using electrodes that require ground contact in a DC electrical survey, while the EM method "measures" apparent conductivity without ground contact. The EM method, known as a "potential method", involves transmitting and receiving EM fields, commonly using a set of coils. The common unit of resistivity is ohm-m and conductivity is its inverse, in Siemen/m. The apparent

> *<sup>V</sup> <sup>G</sup> I*

where *ΔV* is the voltage between a pair of potential electrodes, *I* is the current that flows through another pair of source electrodes, and *G* is the geometric factor that depends on the geometry of the electrodes. For a Wenner array that uses four equally spaced electrodes, for instance, *G* is the electrode spacing itself. Even for this simple array, each electrode spacing generates a different apparent resistivity because the spacing controls the volume of the subsurface sampled by the measurement. It is only when the earth is a homogeneous half space

Similarly, apparent conductivity is only same as the true conductivity when the earth is a homogeneous half space. As an example, consider a pair of horizontal coils separated by a distance *r.* A routinely measured quantity is called the *mutual coupling ratio* which, for a horizontal coplanar (or vertical dipole) coil configuration over a layered earth as derived by

> ( ) ( ) 3 2 0

*H*p and *H*s are the primary and secondary fields at the receiver coil; *J*0 is the 0th order Bessel function; *r* is the coil separation and *h* is the sensor height above the ground. *Q* represents the

 l l

 l


0

¥

*<sup>H</sup> <sup>h</sup> Q r R J re d <sup>H</sup>*

ll

<sup>D</sup> <sup>=</sup> (1)

2 *<sup>a</sup>*

 p

r

that the apparent resistivity is the same as the true resistivity.

s p

spatial situation along the pipeline.

264 Environmental Applications of Remote Sensing

resistivity *ρa* is defined in DC resistivity as:

[14, 15, 16, 17], among others is written as:

**2.1. FDEM**

$$R\left(\mathcal{\lambda}\right) = \frac{\mathcal{\lambda} - \sqrt{\mathcal{\lambda}^2 + \iota 2\pi f \mu \sigma}}{\mathcal{\lambda} + \sqrt{\mathcal{\lambda}^2 + \iota 2\pi f \mu \sigma}}\tag{3}$$

where *f* is the transmitter frequency in Hz, *µ* the magnetic permeability and *σ* the half-space conductivity. Based on *Q* measured at a particular frequency over a real (heterogeneous) earth, we can invert Equation (2) to obtain the *apparent* half-space conductivity *σa*. It is obvious from Equation (2) that the resulting *σ* depends on coil separation, sensor height and frequency. In addition, each coil configuration (vertical coplanar, coaxial, etc.) has a different formula for *Q.* Figure 2 shows a coplanar coil pair at height h above layered earth [18], and a damped leastsquares inversion based on singular value decomposition to solve the nonlinear inverse problem.

**Figure 2.** Geometry of the horizontal coplanar electromagnetic sensor over layered earth where *σ* is the conductivity, t is the thickness of each layer, the subscripts stand for the number of layers, *s* is the coil separation and *h* is the sensor height [18].

Figure 3 shows the responses of the Gem-2 sensor over a half space as a function of induction number:

$$\theta = \left(\sigma \mu \alpha / 2\right)^{1/2} s \tag{4}$$

where *ω* is the angular frequency, *µ* is the magnetic permeability and *s* is coil separation.

**Figure 3.** The in-phase and quadrate responses as a function of induction number (from Huang and Won, 2003).

The depth of investigation of an EM system can be estimated using the skin depth *δ*, which is defined in classical EM theory as the distance in a homogeneous medium over which the amplitude of a plane wave is attenuated by a factor of 1/*e*, or about 37% of its original ampli‐ tude. The skin depth *δ* is:

$$
\delta = \sqrt{\frac{2}{\sigma \mu \rho \sigma}} \tag{5}
$$

The skin depth and the ability to transmit in several frequencies allows us to perform "fre‐ quency sounding" using a multifrequency sensor, thereby resolving different depths of penetration as sketched in Figure 4.

**Figure 4.** Frequency sounding for various depths using a multifrequency FDEM sensor such as Gem-2.

#### **2.2. GPR**

GPR, a reflection-scattering imaging method, is widely used for subsurface imaging in geophysics. GPR uses high frequencies (wavelengths; MHz–GHz). EM waves may form images of the subsurface by transmitting radar pulses into the ground and receiving the deflected waves from the interfaces below. Using wave methods and analysis, GPR images can be analyzed for their derived electrical properties and subsurface characteristics and for spatial mapping of water content [2, 3]. The range resolution is a function of the subsurface dielectric constants and the wave's frequency. It may vary from several centimeters to several tens of centimeters at the relevant effective frequencies [19, 20] For a certain wavelength, the penetration of GPR waves into the subsurface is mainly a function of the host material's conductivity, and therefore GPR waves decay significantly in conductive and saline soils. Using wave methods and analysis, GPR images can be analyzed for their derived electrical properties and subsurface characteristics and for spatial mapping of water content [2,3], as described in the following model.

The connection between the EM velocity and dielectric constant is expressed as:

$$
v = \frac{c}{\sqrt{k}}\tag{6}$$

where *c* is the speed of light in a vacuum and *k* is the dielectric constant.

The dielectric constant of water (*k*w) is about 80. The dielectric constant of air (*k*a) is 1. The dielectric constant of common "dry" soil (*k*dry soil) with residual moisture content can range between 6 and 15 (the effective dielectric constant of dry soil is determined according to volumetric mixing ratios between soil, water and air components).

The difference in the effective dielectric constant of "dry" and "wet" soils is mainly a function of the ratio between the air and water volumes, when the volumes are normalized to:

$$\mathbf{V\_{dry soil}} + \mathbf{V\_w} + \mathbf{V\_a} = \mathbf{1} \tag{7}$$

then:

The depth of investigation of an EM system can be estimated using the skin depth *δ*, which is defined in classical EM theory as the distance in a homogeneous medium over which the amplitude of a plane wave is attenuated by a factor of 1/*e*, or about 37% of its original ampli‐

**Figure 3.** The in-phase and quadrate responses as a function of induction number (from Huang and Won, 2003).

2 *µ*

The skin depth and the ability to transmit in several frequencies allows us to perform "fre‐ quency sounding" using a multifrequency sensor, thereby resolving different depths of

<sup>=</sup> (5)

s w

d

**Figure 4.** Frequency sounding for various depths using a multifrequency FDEM sensor such as Gem-2.

tude. The skin depth *δ* is:

266 Environmental Applications of Remote Sensing

penetration as sketched in Figure 4.

$$k\_{\rm eff} = k\_{\rm dry\ soil} V\_{\rm dry\ soil} + k\_{\rm w} V\_{\rm w} + k \left(1 - V\_{\rm w} \right) \tag{8}$$

The maximal soil–water absorbency is a strong function of the effective porosity.

#### **3. Leak detection in Ariel**

Ariel is a small city (about 20,000 residents) in Israel, located in the central highland region known as the Samarian Hills. It is situated 40 km (25 miles) east of Tel Aviv and 40 km west

of the Jordan River. It is situated 700 m (more than 2000 feet) above sea level. The city stretches over 12 km (8 miles) in length and 2 km in width. The research was performed with Yuvalim, the company that is responsible for maintaining the water and sewage network in the Ariel area and for supplying available water to residents. The mutual research was performed to identify sewage leaks before they pollute and damage the surrounding area. The research was supported by the Israeli Water Authority. The work was performed in several stages. 10 230 resea 231 dama 232 Wate arch was p age the su er Authority performed urrounding y. The work to identify area. The was perfor y sewage le research rmed in sev eaks befor was suppo veral stages re they po orted by th s. llute and he Israeli

#### **3.1. Selecting study sites** 237 Area as were sele ected in Ar riel for syst

**study sites**

**s**

nd geologica

al and pedo

**Selecting s**

ography an

**3.1 S** <sup>231</sup>

242 photo

238

239

242 sewe

Areas were selected in Ariel for system calibration (Figure 5). Two areas were chosen for the method calibration: the first was an industrial area and the second a residential area, both with well-mapped networks of water and sewage pipes. These areas were selected on the basis of information from computerized data, observations, field visits, use of orthophotos, aerial photography and geological and pedological data. 238 chos 239 seco 240 sewa 241 comp sen for the ond a resid age pipes. puterized method c dential are These are data, obse calibration: ea, both w eas were s ervations, the first w with well-ma selected on field visits was an ind apped netw n the basis s, use of dustrial area works of w s of inform orthophot a and the water and ation from tos, aerial

ta.

tion (Figure

e 5). Two a

areas were

em calibrat

ological dat

**Figure 5.** Maps showing Ariel's location (a) and the drainage infrastructure, sewerage and water supply for this city (b).

and the dra

ainage infra

astructure,

cation (a) a ity (b).

#### 241 **Figu ure 5.** Map s showing **3.2. Soil characterization**

water suppl

Ariel's loc ly for this ci

erage and w

242 **3.2 S Soil charac cterization** To characterize the pedological structure of the subsurface layers, excavations were performed. We sampled grain size, void content and porosity, moisture content, soil density and soil characteristics. We dug a channel in an underground sewage pipe replacement area at the experimental sites. Figure 6 presents the characterization of the sub layer.

Utilization of Ground-Penetrating Radar and Frequency Domain Electromagnetic for Investigation of Sewage Leaks http://dx.doi.org/10.5772/62156 269

**Figure 6.** Soil subsurface cross-section at site 1. Wooden pegs mark the changing soil layers.

The soil in the area is red Mediterranean, also known as Terra Rossa [21] and Lithic ruptic Xerochrept [22]. Terra Rossa occurs in areas where heavy rainfall dissolves carbon from the parent calcium carbonate rock and silicates are leached out of the soil, leaving residual deposits that are rich in iron hydroxides, causing the red color. Such areas are usually depressions within limestone. The soil was sampled in a 0.5-m-wide ditch at a depth of 2 m. The area has an easterly aspect, with an average elevation of 400 m above sea level. The local slopes vary between 7% and 25%. Soil texture was clay loam with an average composition of 45% sand, 25% silt and 30% clay. The sand content increased toward the lower part of the area. The average lime content was 30%. Rock fragments of up to 40 cm appeared together with the soil.

#### **3.3. GPR calibration**

of the Jordan River. It is situated 700 m (more than 2000 feet) above sea level. The city stretches over 12 km (8 miles) in length and 2 km in width. The research was performed with Yuvalim, the company that is responsible for maintaining the water and sewage network in the Ariel area and for supplying available water to residents. The mutual research was performed to identify sewage leaks before they pollute and damage the surrounding area. The research was

eaks befor was suppo veral stages

re they po orted by th

llute and he Israeli

areas were a and the water and ation from tos, aerial

e 5). Two a dustrial area works of w s of inform orthophot

s.

tion (Figure was an ind apped netw n the basis s, use of

Areas were selected in Ariel for system calibration (Figure 5). Two areas were chosen for the method calibration: the first was an industrial area and the second a residential area, both with well-mapped networks of water and sewage pipes. These areas were selected on the basis of information from computerized data, observations, field visits, use of orthophotos, aerial

ta.

supported by the Israeli Water Authority. The work was performed in several stages.

em calibrat the first w with well-ma selected on field visits ological dat

y sewage le research rmed in sev

**3.1. Selecting study sites**

arch was p age the su er Authority

performed urrounding y. The work

268 Environmental Applications of Remote Sensing

to identify area. The was perfor

riel for syst calibration: ea, both w eas were s ervations, al and pedo

**study sites**

ected in Ar method c dential are These are data, obse nd geologica

**Selecting s**

as were sele sen for the ond a resid age pipes. puterized ography an

10

230 resea 231 dama 232 Wate

**3.1 S** <sup>231</sup>

237 Area 238 chos 239 seco 240 sewa 241 comp 242 photo

238

239

241 **Figu** 242 sewe

**ure 5.** Map erage and w

**3.2. Soil characterization**

s showing water suppl

(a

a)

Ariel's loc ly for this ci

cation (a) a ity (b).

experimental sites. Figure 6 presents the characterization of the sub layer.

**Figure 5.** Maps showing Ariel's location (a) and the drainage infrastructure, sewerage and water supply for this city (b).

To characterize the pedological structure of the subsurface layers, excavations were performed. We sampled grain size, void content and porosity, moisture content, soil density and soil characteristics. We dug a channel in an underground sewage pipe replacement area at the

and the dra

ainage infra

astructure,

(b)

**cterization**

**Soil charac**

242 **3.2 S**

photography and geological and pedological data.

**s**

Calibration of the GPR system to the subsurface properties of the cross-section in a dry state (without leakage) is shown in Figure 7. The depth to the pipe was measured in a nearby manhole.

**Figure 7.** Part of the GPR profile performed for calibration of the GPR system in the Ariel industrial zone, on the road close to a rubber factory. The black circle displays diffraction created by the drain pipe. Above it, the trench is detected as well. The horizontal scale describes the measurement location (in meters) along the profile. The vertical scales de‐ scribe the time (in nanoseconds) and depth (in meters). The amplitude–intensity scale is shown as well.

Figure 7 shows the results of advanced processing of a cross-section for calibration of the system in the industrial area. On the horizontal scale, simulations are described above the measurement location along the incision in meters; the vertical scales describe the time and depth of the reflections on a timescale of 50 ns and scale depth of 2.5 m below the surface (the strength of the reflections is graded according to the color scale in Figure 7, where the diffrac‐ tion created by the drainage pipe can be deduced from a return time from the pipe of approx‐ imately 32 ns). The diffraction depth is 2.45 m, and the data from the system matches the data measured on the ground. This adaptation makes it possible to determine the velocity of the EM wave. The average measured subsurface speed of the EM wave (*v*) was 0.093 ± 0.001 m/ns at the Ariel industrial site. It is important to note that the speed of the wave depends on the directly calculated form and location of the anomaly and thus data processing is critical to the research results.

#### **3.4. The experimental site**

The experimental site for sewage pipeline and manhole leaks was located near Ariel's old stadium, not far from HaAtsmaut Street (Figure 8, blue rectangle), where a project for the replacement of old sewer pipes has been initiated.

Utilization of Ground-Penetrating Radar and Frequency Domain Electromagnetic for Investigation of Sewage Leaks http://dx.doi.org/10.5772/62156 271

14

291

292

293 **Figure 8**. The experimental site is located at the western end of the sewage 294 line adjacent to HaAtsmaut Street (blue rectangle). It includes 12-in. diameter iron pipes carrying on the order of 1000–1200 m<sup>3</sup> <sup>295</sup>sewage water per day, and an average 100 m<sup>3</sup> <sup>296</sup>/h during peak flow. **Figure 8.** The experimental site is located at the western end of the sewage line adjacent to HaAtsmaut Street (blue rectangle). It includes 12-in. diameter iron pipes carrying on the order of 1000–1200 m3 sewage water per day, and an average 100 m3 /h during peak flow.

297 Leakage was initiated in two places at the western site by cracking the sewer 298 pipes close to their bottom side. One crack was made about 6 m from the 299 sewage pit in the northern iron pipe using an electrical disk that created a Leakage was initiated in two places at the western site by cracking the sewer pipes close to their bottom side. One crack was made about 6 m from the sewage pit in the northern iron pipe using an electrical disk that created a wedge-shaped hole 15–20 cm in diameter; the second crack was also a circle of 15–20 cm diameter in the lower part of the pipe (Figure 9). The experimental site was monitored daily by radar and FDEM before the start of and during the controlled leakage.

**Figure 9.** Pictures of the two cracks made in the sewer pipes for the controlled leakage experiment.

#### **4. Results**

**Figure 7.** Part of the GPR profile performed for calibration of the GPR system in the Ariel industrial zone, on the road close to a rubber factory. The black circle displays diffraction created by the drain pipe. Above it, the trench is detected as well. The horizontal scale describes the measurement location (in meters) along the profile. The vertical scales de‐

Figure 7 shows the results of advanced processing of a cross-section for calibration of the system in the industrial area. On the horizontal scale, simulations are described above the measurement location along the incision in meters; the vertical scales describe the time and depth of the reflections on a timescale of 50 ns and scale depth of 2.5 m below the surface (the strength of the reflections is graded according to the color scale in Figure 7, where the diffrac‐ tion created by the drainage pipe can be deduced from a return time from the pipe of approx‐ imately 32 ns). The diffraction depth is 2.45 m, and the data from the system matches the data measured on the ground. This adaptation makes it possible to determine the velocity of the EM wave. The average measured subsurface speed of the EM wave (*v*) was 0.093 ± 0.001 m/ns at the Ariel industrial site. It is important to note that the speed of the wave depends on the directly calculated form and location of the anomaly and thus data processing is critical to the

The experimental site for sewage pipeline and manhole leaks was located near Ariel's old stadium, not far from HaAtsmaut Street (Figure 8, blue rectangle), where a project for the

scribe the time (in nanoseconds) and depth (in meters). The amplitude–intensity scale is shown as well.

research results.

**3.4. The experimental site**

270 Environmental Applications of Remote Sensing

replacement of old sewer pipes has been initiated.

Daily monitoring with the FDEM method included five cross-sections: four were parallel to the sewer pipeline and the fifth was above it, running on each side of the pipeline at a distance of 0.5 m. During the experiment, FDEM scanning was performed to qualify the effect of moisture on the soil cross-section. Figure 10 shows the status of the subsurface before the start of the controlled leak; it was in a relatively dry state characteristic of the month of May at this site.

**Figure 10.** Map of the integrated electrical conductivity at 60,025 Hz before the start of the controlled leak at the west‐ ern site (locations of the measurements are shown by the blue rectangle in Figure 8). The map is based on measure‐ ments performed with a GEM-2 FDEM sensor. The location of the sewer pipe is marked with a black line. Data were collected prior to the leak with dimensional scanning of approximately 30 m × 25 m. Lower conductivity (*σ*) values (11 mS/m) appear in blue-green in the southwestern corner of the area, while the highest conductivity appears in red-pur‐ ple (55 mS/m) in the northeastern part of the map. These conductivity changes suggest anomalous subsurface moisture from the water pipe near the old stadium, as well as the accumulation of water from the slope, where there is a garden.

Figure 11 shows a pronounced increase in electrical conductivity of about 40 mS/m after 4 days of controlled leakage. The area has high conductivity because of changes in wetness due to a significant increase in liquid as a result of the sewage flow.

The results of the FDEM measurements conducted 10 days after the beginning of the controlled leak are presented in Figure 12. This picture may look similar to Figure 11 in terms of colors, but their intensity has increased due to an increase in the conductivity values to about 152 mS/m.

On the map in Figure 12, low visibility, reflecting low electrical conductivity, is shown in bluegreen shades, high visibility in red-colored shades. Purple indicates sewer leakage on the background of the driest area, highlighting the differences in moisture. A wide area can be seen west of the pipe (black line in Figure 12) with relatively low electrical conductivity compared to the rest of the region. Northeast of the pipe, there is high electrical conductivity resulting from the spillover of sewage water.

Utilization of Ground-Penetrating Radar and Frequency Domain Electromagnetic for Investigation of Sewage Leaks http://dx.doi.org/10.5772/62156 273

**Figure 11.** Map of the electrical conductivity at 60,025 Hz after about 4 days of leakage. Measurements were collected during the sewage leak, under wet conditions, with the GEM-2 sensor (locations of the measurements are shown by the blue rectangle in Figure 8). The location of the sewer pipe is marked with a black line. The highest conductivity value was about 95 mS/m. The significant increase in electrical conductivity is a result of the sewer liquids that were spilled during the 4 days of the controlled leak, in both the southwestern and northeastern sides of the area, probably due to a subsurface topography gradient.

**Figure 10.** Map of the integrated electrical conductivity at 60,025 Hz before the start of the controlled leak at the west‐ ern site (locations of the measurements are shown by the blue rectangle in Figure 8). The map is based on measure‐ ments performed with a GEM-2 FDEM sensor. The location of the sewer pipe is marked with a black line. Data were collected prior to the leak with dimensional scanning of approximately 30 m × 25 m. Lower conductivity (*σ*) values (11 mS/m) appear in blue-green in the southwestern corner of the area, while the highest conductivity appears in red-pur‐ ple (55 mS/m) in the northeastern part of the map. These conductivity changes suggest anomalous subsurface moisture from the water pipe near the old stadium, as well as the accumulation of water from the slope, where there is a garden.

Figure 11 shows a pronounced increase in electrical conductivity of about 40 mS/m after 4 days of controlled leakage. The area has high conductivity because of changes in wetness due to a

The results of the FDEM measurements conducted 10 days after the beginning of the controlled leak are presented in Figure 12. This picture may look similar to Figure 11 in terms of colors, but their intensity has increased due to an increase in the conductivity values to

On the map in Figure 12, low visibility, reflecting low electrical conductivity, is shown in bluegreen shades, high visibility in red-colored shades. Purple indicates sewer leakage on the background of the driest area, highlighting the differences in moisture. A wide area can be seen west of the pipe (black line in Figure 12) with relatively low electrical conductivity compared to the rest of the region. Northeast of the pipe, there is high electrical conductivity

significant increase in liquid as a result of the sewage flow.

resulting from the spillover of sewage water.

about 152 mS/m.

272 Environmental Applications of Remote Sensing

**Figure 12.** Map of integrated electrical conductivity at 60,025 Hz. Measurements were collected with the FDEM system, under wet conditions, after 10 days of controlled leakage (locations of the measurements are shown by the blue rectan‐ gle in Figure 8). Electrical conductivity ranged from 0 to 152 mS/m. Low conductivity is expressed in blue-green shades, high conductivity in purple-red colors.

Figure 13 shows maps made by FDEM monitoring of electrical conductivity at various frequencies in the first tested area. The maps are arranged, from left to right, at increasing frequencies and depth: the frequencies were 2,025 Hz, 4,725 Hz, 11,025 Hz, 25,725 Hz and 60,025 Hz, each frequency representing a 30 cm increase in depth. The low-visibility electrical conductivity is represented by blue-green hues, and the high-visibility electrical conductivity by red-purple hues. There were a few quantitative differences in the map scales.

**Figure 13.** Maps made by FDEM monitoring of electrical conductivity at 2,025 Hz, 4,725 Hz, 11,025 Hz, 25,725 Hz and 60,025 Hz. The lower EC values are represented by blue-green hues, and the higher EC values by red-purple hues. There were a few quantitative differences between the maps' scales.

Four sections, two on each side of the sewer, were monitored by GPR and are shown in Figure 14. The distance between the main radar cross-sectional cuts was approximately 0.5 m. The radar sections shown in Figure 14 were collected with an antenna at a nominal frequency of 250 MHz over the location of the underground sewage pipe at the first (western) test site. The first cross-section was obtained before the leak started and reflects the typical dry state of the ground in May. An incision was made a few days after the initiation of the leak and shows a relatively wet subsoil. The right cross-section shows an incision made at a lower depth, 10 days after leak initiation, indicating a further increase in wetness. Similar data processing was carried out for the three cross-sections to highlight their differences.

Utilization of Ground-Penetrating Radar and Frequency Domain Electromagnetic for Investigation of Sewage Leaks http://dx.doi.org/10.5772/62156 275

**Figure 14.** Soil moisture reflected by GPR cross-section (locations of the measurements are shown by the blue rectangle in Figure 8).

#### **5. Modeling subsurface moisture content**

Figure 13 shows maps made by FDEM monitoring of electrical conductivity at various frequencies in the first tested area. The maps are arranged, from left to right, at increasing frequencies and depth: the frequencies were 2,025 Hz, 4,725 Hz, 11,025 Hz, 25,725 Hz and 60,025 Hz, each frequency representing a 30 cm increase in depth. The low-visibility electrical conductivity is represented by blue-green hues, and the high-visibility electrical conductivity

**Figure 13.** Maps made by FDEM monitoring of electrical conductivity at 2,025 Hz, 4,725 Hz, 11,025 Hz, 25,725 Hz and 60,025 Hz. The lower EC values are represented by blue-green hues, and the higher EC values by red-purple hues.

Four sections, two on each side of the sewer, were monitored by GPR and are shown in Figure 14. The distance between the main radar cross-sectional cuts was approximately 0.5 m. The radar sections shown in Figure 14 were collected with an antenna at a nominal frequency of 250 MHz over the location of the underground sewage pipe at the first (western) test site. The first cross-section was obtained before the leak started and reflects the typical dry state of the ground in May. An incision was made a few days after the initiation of the leak and shows a relatively wet subsoil. The right cross-section shows an incision made at a lower depth, 10 days after leak initiation, indicating a further increase in wetness. Similar data processing was

There were a few quantitative differences between the maps' scales.

carried out for the three cross-sections to highlight their differences.

by red-purple hues. There were a few quantitative differences in the map scales.

274 Environmental Applications of Remote Sensing

Moisture content was computed on the basis of subsurface GPR and FDEM measurements and its spatial spread was obtained for calibration and wetness testing with water- and sewagecarrying pipelines. In these experiments, radar velocities were measured and dielectric constants were computed. Their correlations were used to measure the moisture content from data collected in the residential and industrial neighborhoods.

The computation of moisture content using GPR was based on the method developed by Basson [2] From the calibration measurements conducted at the end of May 2012, the average subsurface EM wave velocity was 0.093 ± 0.001 m/ns. The calculated dielectric constant during this period was about 10.4. This value is low but not minimal, as minimal moisture content is typically found in the mid-to-late summer months (according to data from the Israel Mete‐ orological Service, the rain that accumulated in the area in the months before the GPR measurements amounted to about 161 mm).

The velocity of EM waves in a substance is mainly a function of that substance's bulk dielectric properties and moisture content. When a substance is composed of a mixture of materials, the velocity is a function of their mixing ratios. In the case of a subsurface environment, we can treat the substance as a bulk property composed of soil, rock, minerals and organic materials mixed with air and water. When the rate of air increases, the velocity increases as well. However, when the moisture content increases, the average dielectric constant decreases as well and fro equation (6) it can be seen that the EM velocity (v) decreases as well.

The difference in the effective dielectric constant of "dry" and "wet" soils is mainly a function of the ratio between the air and water volumes, when the volumes are normalized according to equations (7) and (8). The maximal soil–water absorbency is a strong function of the effective porosity. For soils in the Ariel region, the effective porosity can vary from 40% to 60%. We used an average effective porosity of 50% in our computations. Therefore, the possible mixing ratios relative to the normalized volume are:

$$V\_{\text{dry soil}} = 0.5 V\_{\text{tot}} \tag{9}$$

$$V\_{\rm av} + V\_{\rm a} = 0.5V\_{\rm tot} \tag{10}$$

Since *k*dry soil is the effective dielectric constant measured using GPR imaging for a soil with residual moisture content and since *k*a = 1:

$$k\_{\text{top soil}} = 0.5k\_{\text{dry soil}} + 80V\_{\text{w}} + 0.5 - V\_{\text{w}} \tag{11}$$

The radar wave velocity for "dry" soil at the surface will be measured and is expected to vary with the GPR and its value, *v*top soil ~ 0.07–0.14 m/ns. From Equation (1), this velocity range can reflect dielectric constant values of ~4.6–18.4 for *k*top soil*.* For example, for maximal dielectric constant values of 5–14 for delicate quartz-based soils and for the presented computations, the moisture content in the surface can vary as *V*w top soil ~ 0.4%–2.1%. In the same way, we can investigate deeper soils where the moisture content is expected to be greater. The average radar wave velocity (*v*humid soil) measured by the GPR at the calibration site in Ariel at the end of May is 0.093 m/ns. Using Equation (1), this velocity reflects a dielectric constant value (*k*humid soil) of 10.406. The additional volume of water needed to increase the dielectric constant from 4.6 to 10.41 can be computed as:

$$
\Delta k = 5.81 \tag{12}
$$

Utilization of Ground-Penetrating Radar and Frequency Domain Electromagnetic for Investigation of Sewage Leaks http://dx.doi.org/10.5772/62156 277

$$
\Delta k = 80 \Delta V\_{\text{w}} \tag{13}
$$

$$
\Delta V\_w = 7.26\tag{14}
$$

We develop a moisture content model using relative values of the moisture content (based on Equations (6–14)) causing an increase in electrical conductivity as measured by the FDEM. We had to consider the overall subsurface features, such as texture, density and effective porosity, as well as the content of salts in soils irrigated with brackish effluent water. The model results are presented in the graph in Figure 15.

**Figure 15.** Volumetric moisture content calculated from measurements and from the FDEM model in the experimental zones in Ariel (accuracy ±10% of the measured value).

#### **6. Discussion and conclusions**

subsurface EM wave velocity was 0.093 ± 0.001 m/ns. The calculated dielectric constant during this period was about 10.4. This value is low but not minimal, as minimal moisture content is typically found in the mid-to-late summer months (according to data from the Israel Mete‐ orological Service, the rain that accumulated in the area in the months before the GPR

The velocity of EM waves in a substance is mainly a function of that substance's bulk dielectric properties and moisture content. When a substance is composed of a mixture of materials, the velocity is a function of their mixing ratios. In the case of a subsurface environment, we can treat the substance as a bulk property composed of soil, rock, minerals and organic materials mixed with air and water. When the rate of air increases, the velocity increases as well. However, when the moisture content increases, the average dielectric constant decreases as

The difference in the effective dielectric constant of "dry" and "wet" soils is mainly a function of the ratio between the air and water volumes, when the volumes are normalized according to equations (7) and (8). The maximal soil–water absorbency is a strong function of the effective porosity. For soils in the Ariel region, the effective porosity can vary from 40% to 60%. We used an average effective porosity of 50% in our computations. Therefore, the possible mixing

Since *k*dry soil is the effective dielectric constant measured using GPR imaging for a soil with

The radar wave velocity for "dry" soil at the surface will be measured and is expected to vary with the GPR and its value, *v*top soil ~ 0.07–0.14 m/ns. From Equation (1), this velocity range can reflect dielectric constant values of ~4.6–18.4 for *k*top soil*.* For example, for maximal dielectric constant values of 5–14 for delicate quartz-based soils and for the presented computations, the moisture content in the surface can vary as *V*w top soil ~ 0.4%–2.1%. In the same way, we can investigate deeper soils where the moisture content is expected to be greater. The average radar wave velocity (*v*humid soil) measured by the GPR at the calibration site in Ariel at the end of May is 0.093 m/ns. Using Equation (1), this velocity reflects a dielectric constant value (*k*humid soil) of 10.406. The additional volume of water needed to increase the dielectric constant from 4.6 to

dry soil tot *V V* = 0.5 (9)

w a tot *VV V* + = 0.5 (10)

D = *k* 5.81 (12)

top soil dry soil w <sup>w</sup> *k k VV* = + +- 0.5 80 0.5 (11)

well and fro equation (6) it can be seen that the EM velocity (v) decreases as well.

measurements amounted to about 161 mm).

276 Environmental Applications of Remote Sensing

ratios relative to the normalized volume are:

residual moisture content and since *k*a = 1:

10.41 can be computed as:

We introduced a combination of GPR and FDEM orthogonal methods to detect subsurface leaks from a sewage pipeline system. The rationale for this combination is to increase the probability of detection, especially in complex urban environments and when the soil–rock setting can vary from relatively resistive to relatively conductive. The results of our study indicate that even minor leaks, such as the minor controlled leaks created in the experiment, and changes in the subsurface moisture content can be accurately detected. We could detect sewage leakage, as well as its progress. The combination of the two methods enabled not only the detection of the leak but also a qualitative assessment of its size. Factors affecting the ability to detect leaks were limited by the soil–rock conductivity, as well as the density of the terrain and subterrain systems and structures. The geophysical methods may detect sewage effluent flow paths as well as the contaminant in the soil.

The limestone and dolomite bedrock in the Ariel area is suitable for GPR mapping. The clarity of the GPR profile enabled analysis and interpretation of the physical data with good accuracy. We could detect sewage leakage, as well as its progress. The anomalous moisture of the leakage accumulating around the sewage pit in the southwest research area validated the efficiency of the methods.

### **Acknowledgements**

The research was supported by the Israel water authority and by the Water Cooperation of Yuvalim. We would like to thank Mr. Omer Shamir from GeoSense for assistance with the data collection.

### **Author details**

Goldshleger Naftaly1,2\* and Basson Uri3


#### **References**

[1] Basson, U., and Ben-Avraham, Z., 1994. Subsurface spatial mapping of pollutants concentrations using ground penetrating radar. Proceedings of the 25th Annual Meet‐ ingoftheIsraelSocietyforEcologyandEnvironmentalQualitySciences,Tel-Aviv,Israel, 3–4 May 1994, p. 29.

[2] Basson, U., 1992. Mapping of Moisture Content and Structure of Unsaturated Sand Layers with Ground Penetrating Radar. M.Sc. thesis.Tel-Aviv University, Tel-Aviv, Israel.

probability of detection, especially in complex urban environments and when the soil–rock setting can vary from relatively resistive to relatively conductive. The results of our study indicate that even minor leaks, such as the minor controlled leaks created in the experiment, and changes in the subsurface moisture content can be accurately detected. We could detect sewage leakage, as well as its progress. The combination of the two methods enabled not only the detection of the leak but also a qualitative assessment of its size. Factors affecting the ability to detect leaks were limited by the soil–rock conductivity, as well as the density of the terrain and subterrain systems and structures. The geophysical methods may detect sewage effluent

The limestone and dolomite bedrock in the Ariel area is suitable for GPR mapping. The clarity of the GPR profile enabled analysis and interpretation of the physical data with good accuracy. We could detect sewage leakage, as well as its progress. The anomalous moisture of the leakage accumulating around the sewage pit in the southwest research area validated the efficiency of

The research was supported by the Israel water authority and by the Water Cooperation of Yuvalim. We would like to thank Mr. Omer Shamir from GeoSense for assistance with the data

[1] Basson, U., and Ben-Avraham, Z., 1994. Subsurface spatial mapping of pollutants concentrations using ground penetrating radar. Proceedings of the 25th Annual Meet‐ ingoftheIsraelSocietyforEcologyandEnvironmentalQualitySciences,Tel-Aviv,Israel,

flow paths as well as the contaminant in the soil.

278 Environmental Applications of Remote Sensing

the methods.

collection.

**Author details**

**References**

Goldshleger Naftaly1,2\* and Basson Uri3

1 Civil Engineering Ariel University, Israel

3 Geosense Ltd., Even Yehuda, Israel

3–4 May 1994, p. 29.

2 Israel Ministry of Agriculture, Beit Dagan, Israel

\*Address all correspondence to: Goldshleger1@gmail.com

**Acknowledgements**

