**5. Results**

#### **5.1 "Southern Conveyor Project" pipeline**

The detection of the buried water pipe was initially performed using the multi-temporal Google Earth images (Figure 12). As shown, the success rate for the detection of the water pipe can vary depending on the period of observation. The interpretation could be performed much easier in areas with no coverage (bare soil) while in cultivated areas the interpretation was a difficult task. In addition, images taken just after rainfall or after watering crops, tend to provide better results since soil marks could be easily spotted.

Moreover, the tree pattern could reveal the footprint of the water pipe (see Figure 13). This pattern can be used for the detection of buried water pipes or can be used for monitoring possible problems resulting from tree roots.

**Figure 13.** The footprint of the water pipe as a result of the tree pattern.

also along the entire length of the water pipe. The results were mapped and statistical analysis

**no Satellite Overpass no Satellite Overpass** Landsat ETM+ 07/05/2007 7 Landsat ETM+ 14/09/2008 Landsat ETM+ 23/05/2007 8 Landsat ETM+ 30/09/2008 Landsat ETM+ 27/08/2007 9 Landsat ETM+ 16/10/2008 Landsat ETM+ 28/07/2008 10 Landsat ETM+ 22/12/2009 Landsat ETM+ 13/08/2008 11 Landsat ETM+ 07/01/2010 Landsat ETM+ 29/08/2008 12 Landsat ETM+ 13/04/2010

was performed.

**Table 3.** Satellite images used for this study

166 Remote Sensing of Environment: Integrated Approaches

(a) (b)

(c) (d)

23-10-2003; (b): 13-06-2004; (c): 29-05-2008; and (d): 30-05-2009.

**5.1 "Southern Conveyor Project" pipeline**

**5. Results**

**Figure 12.** Google Earth satellite image used for the detection of the buried water pipe during different periods: (a):

The detection of the buried water pipe was initially performed using the multi-temporal Google Earth images (Figure 12). As shown, the success rate for the detection of the water pipe can vary depending on the period of observation. The interpretation could be performed much The IKONOS image used for this case study was able to maximize the visible footprint of the water pipe. Indeed, using the VNIR part of the spectrum and false colour composites (Figure 14) made possible the detection of both soil and crop marks. The IKONOS multispectral image was able to detect other parts of the water pipe network of the area, as shown in Figure 14 (right arrow). Spatial filter and PCA analysis applied to the image data (Figure 15) were able further to enhance the interpretation.

In an attempt to evaluate if an automatic detection of such crop marks could be performed (e.g. classification), spectral profiles were examined. Spectral signatures from the image were evaluated as shown in Figure 15, which features areas of crop marks and of healthy vegetation. Scatter plots from these two areas (Figure 16) indicate that a spectral difference exists between these areas, especially in the VNIR part of the spectrum.

**Figure 14.** IKONOS VNIR-R-G pseudo colour composite

the green part of the spectrum (520-600nm) and 25% in the very near infrared (750-900nm) in

**Figure 16.** Scatter plots from crop marks (red square) and healthy vegetation (yellow square) for Bands 1-3 and Bands

Detection of Water Pipes and Leakages in Rural Water Supply Networks Using Remote Sensing Techniques

http://dx.doi.org/10.5772/39309

169

Figure 17: Ground spectral signatures over dry and wet soil in the *'Lakatameia'* pipeline

**Figure 17.** Ground spectral signatures over dry and wet soil in the *'Lakatameia'* pipeline

Wet soil

Figure 18: Ground spectral signatures of different targets in the *'Lakatameia'* pipeline

Figures 19 and 20 present the spectral signatures over the same areas from different heights, using the low altitude system. Reflectance initially increases as the system is raised above ground level (until 10 meters) while a small decrease of the reflectance is observed afterwards (16 meters) which can be associated with the larger area covered from the spectroradiometer. However it should be noted that these differences (~5%) are similar to the total relative uncertainties of calibration for satellite sensors (within 5%) (Trishchenko et

al. 2002).

contrast to 12% and 35% respectively for the wet grass.

1-4 combinations (left and right respectively).

Dry soil

**Figure 15.** IKONOS 3 x 3 high pass filter (left) and PCA analysis (right)

#### **5.2.** *"Lakatameia"* **pipeline**

The results found that water leakages could be monitored using remote sensing techniques. As shown in Figure 17, the spectral signatures of dry and wet soil is easily recognized in the visible range of the spectrum (400 -700 nm) and in the very near infrared range (750-900nm). Wet soil tends to give 20-25% lower reflectance values compare to the dry soil. This difference is also maximized in the very near infrared range of the spectrum. Similarly, Figure 18 indicates spectral signature profiles of several targets before (dry) and after (wet) a leakage event. Similar findings also applied to vegetation. Dry grass tends to give approximately 5% reflectance in Detection of Water Pipes and Leakages in Rural Water Supply Networks Using Remote Sensing Techniques http://dx.doi.org/10.5772/39309 169

**Figure 16.** Scatter plots from crop marks (red square) and healthy vegetation (yellow square) for Bands 1-3 and Bands 1-4 combinations (left and right respectively).

the green part of the spectrum (520-600nm) and 25% in the very near infrared (750-900nm) in contrast to 12% and 35% respectively for the wet grass.

Figure 17: Ground spectral signatures over dry and wet soil in the *'Lakatameia'* pipeline

Figure 18: Ground spectral signatures of different targets in the *'Lakatameia'* pipeline

Figures 19 and 20 present the spectral signatures over the same areas from different heights, using the low altitude system. Reflectance initially increases as the system is raised above ground level (until 10 meters) while a small decrease of the reflectance is observed afterwards (16 meters) which can be associated with the larger area covered from the spectroradiometer. However it should be noted that these differences (~5%) are similar to the total relative uncertainties of calibration for satellite sensors (within 5%) (Trishchenko et

al. 2002).

**Figure 17.** Ground spectral signatures over dry and wet soil in the *'Lakatameia'* pipeline

**5.2.** *"Lakatameia"* **pipeline**

**Figure 15.** IKONOS 3 x 3 high pass filter (left) and PCA analysis (right)

**Figure 14.** IKONOS VNIR-R-G pseudo colour composite

168 Remote Sensing of Environment: Integrated Approaches

The results found that water leakages could be monitored using remote sensing techniques. As shown in Figure 17, the spectral signatures of dry and wet soil is easily recognized in the visible range of the spectrum (400 -700 nm) and in the very near infrared range (750-900nm). Wet soil tends to give 20-25% lower reflectance values compare to the dry soil. This difference is also maximized in the very near infrared range of the spectrum. Similarly, Figure 18 indicates spectral signature profiles of several targets before (dry) and after (wet) a leakage event. Similar findings also applied to vegetation. Dry grass tends to give approximately 5% reflectance in

**Figure 18.** Ground spectral signatures of different targets in the *'Lakatameia'* pipeline

Figures 19 and 20 present the spectral signatures over the same areas from different heights, using the low altitude system. Reflectance initially increases as the system is raised above ground level (until 10 meters) while a small decrease of the reflectance is observed afterwards (16 meters) which can be associated with the larger area covered from the spectroradiometer. However it should be noted that these differences (~5%) are similar to the total relative uncertainties of calibration for satellite sensors (within 5%) (Trishchenko et al. 2002).

**Figure 19.** Spectral signatures of wet soil in the *'Lakatameia'* pipeline at different heights using the low altitude sys-

Detection of Water Pipes and Leakages in Rural Water Supply Networks Using Remote Sensing Techniques

http://dx.doi.org/10.5772/39309

171

**Figure 20.** Spectral signatures of dry soil in the *'Lakatameia'* pipeline at different heights using the low altitude system

Based on the findings of the *"Lakatameia"* water pipe, satellite images where used for the detection of known water leakages using archive satellite images. In order to examine the capabilities of satellite remote sensing images for the detection of water leakages, several algorithms and analyses were carried out. At first, reflectance values of all datasets (see Table

**5.3. "Frenaros — Choirokoitia " water pipe**

tem

The above results are well supported in the literature. Nocita et al. (2011), Ouillon et al. (2002), Dobos (2003), Kaleita et al. (2005) and Garcia-Rodriguez (2011) found that moisture affects the reflectance value of soil. There is a notable decrease in reflectance with increasing moisture in the ground (Bowers and Hanks, 1965; Baumgardner et al., 1985; Twomey et al., 1986; Ishida et al., 1991; Whiting et al., 2000; Bogrekci and Lee, 2005; Lesaignoux et al. 2007). However, the rate of decrease in relative reflectance becomes more moderate with increasing ground moisture, since at very high moisture contents, the soil is already quite dark and further moisture added to the soil has less of an effect on the reflectance (Kaleita et al., 2005). Moisture dominates the spectral reflectance of soils in the 340-2500 nm wavelengths (Somers et al., 2010; Bogrekci and Lee, 2005). Moisture affects the reflection of shortwave radiation from ground surfaces in the visible and near-infrared - VNIR (400-1100nm) and shortwave infrared - SWIR (1100-2500nm) regions of the spectrum (Bowers and Hanks, 1965; Skidmore et al., 1975). It is notable that, althoughprecipitation affects the reflectance value for each target, itdoes not change the typical spectral signature between wet and dry conditions (Philpot, 2010).

The results indicate that the detection of a leakage event is possible using remote sensing techniques.Indeed, the use of the very nearinfrared range of the spectrum can be used on areas with bare soil or with vegetation.The findings from this pipeline were therefore compared with data from actual cases studies of water leakage in the '*Freanaros-Choirokoitia*' pipeline.

Detection of Water Pipes and Leakages in Rural Water Supply Networks Using Remote Sensing Techniques http://dx.doi.org/10.5772/39309 171

**Figure 19.** Spectral signatures of wet soil in the *'Lakatameia'* pipeline at different heights using the low altitude system

**Figure 20.** Spectral signatures of dry soil in the *'Lakatameia'* pipeline at different heights using the low altitude system

#### **5.3. "Frenaros — Choirokoitia " water pipe**

**Figure 18.** Ground spectral signatures of different targets in the *'Lakatameia'* pipeline

170 Remote Sensing of Environment: Integrated Approaches

spectral signature between wet and dry conditions (Philpot, 2010).

Figures 19 and 20 present the spectral signatures over the same areas from different heights, using the low altitude system. Reflectance initially increases as the system is raised above ground level (until 10 meters) while a small decrease of the reflectance is observed afterwards (16 meters) which can be associated with the larger area covered from the spectroradiometer. However it should be noted that these differences (~5%) are similar to the total relative

The above results are well supported in the literature. Nocita et al. (2011), Ouillon et al. (2002), Dobos (2003), Kaleita et al. (2005) and Garcia-Rodriguez (2011) found that moisture affects the reflectance value of soil. There is a notable decrease in reflectance with increasing moisture in the ground (Bowers and Hanks, 1965; Baumgardner et al., 1985; Twomey et al., 1986; Ishida et al., 1991; Whiting et al., 2000; Bogrekci and Lee, 2005; Lesaignoux et al. 2007). However, the rate of decrease in relative reflectance becomes more moderate with increasing ground moisture, since at very high moisture contents, the soil is already quite dark and further moisture added to the soil has less of an effect on the reflectance (Kaleita et al., 2005). Moisture dominates the spectral reflectance of soils in the 340-2500 nm wavelengths (Somers et al., 2010; Bogrekci and Lee, 2005). Moisture affects the reflection of shortwave radiation from ground surfaces in the visible and near-infrared - VNIR (400-1100nm) and shortwave infrared - SWIR (1100-2500nm) regions of the spectrum (Bowers and Hanks, 1965; Skidmore et al., 1975). It is notable that, althoughprecipitation affects the reflectance value for each target, itdoes not change the typical

The results indicate that the detection of a leakage event is possible using remote sensing techniques.Indeed, the use of the very nearinfrared range of the spectrum can be used on areas with bare soil or with vegetation.The findings from this pipeline were therefore compared with

data from actual cases studies of water leakage in the '*Freanaros-Choirokoitia*' pipeline.

uncertainties of calibration for satellite sensors (within 5%) (Trishchenko et al. 2002).

Based on the findings of the *"Lakatameia"* water pipe, satellite images where used for the detection of known water leakages using archive satellite images. In order to examine the capabilities of satellite remote sensing images for the detection of water leakages, several algorithms and analyses were carried out. At first, reflectance values of all datasets (see Table 3) were calculated based on the metadata file using equations 3 and 4. Following this, several vegetation indices were calculated. In addition, different false colour composites were produced to assess the ability of the system to detect the known leakages from the satellite images.

Normalized Difference Vegetation Index (NDVI);

Ratio Vegetation Index (RVI) were calculated based on the formulas shown in equations 5, 6

Detection of Water Pipes and Leakages in Rural Water Supply Networks Using Remote Sensing Techniques

Figure 22 presents the NDVI development during the examined 12 dates of satellite overpasses (see Table 3). Figure 23 presents the SAVI development during the examined 12 dates of

**Point 1,2,3: "Frenaros-Choirokoitia" pipeline**

1 2 3 4 5 6 7 8 9 10 11 12

**Dates**

**Figure 22.** NDVI refl. (calculated with Reflectance values) development during the examined 12 dates (Landsat im-

(p – p / p + p NIR red NIR red ) ( ) (3)

red NIR p / p (5)

http://dx.doi.org/10.5772/39309

173

Point 1 Point 2 Point 3

(1+0.5 p - p / p + p +0.5 ) ( NIR rb NIR red ) ( ) (4)

Soil Adjusted. Vegetation Index (SAVI) and the

and 7.

Where:

pNIR is the near infrared reflectance

pred is the red reflectance

satellite overpasses.


ages) in Points 1, 2 and 3

**NDVI values**

For Point 1 at *Pyla* area, leakage detection was difficult using medium resolution images. Monitoring of the pipeline using the red and the near infrared part of the spectrum for Point 1 did not reveal any significant changes of reflectance due to the water leakage. Similarly, vegetation indices (NDVI) did not show any differences for Point 3 (*Anglisides* area).

However, for Point 2, Landsat 7 ETM+, promising results were found. As shown in Figure 21, the Landsat satellite image dated January 7, 2010, tends to give higher vegetation index values, prior to the water leakage being repaired on February 18, 2010. However, the above hypothesis is applicable to other areas of the water pipe as well. The above results have shown the limitations of using medium resolution satellite images for the detection of water leakages, especially when these are rare and small.

**Point 2: "Frenaros-Choirokoitia" pipeline**

**Figure 21.** NDVI values using Landsat 7 ETM+ images used over Point 2. The red square highlights the area where the leakage was observed.

In an effort to explore further the information extracted using satellite data the three pilot areas were examined separately. Three vegetation indices, the

Normalized Difference Vegetation Index (NDVI);

Soil Adjusted. Vegetation Index (SAVI) and the

Ratio Vegetation Index (RVI) were calculated based on the formulas shown in equations 5, 6 and 7.

$$(\mathbf{p}\_{\text{NIR}} - \mathbf{p}\_{\text{red}}) \;/\; \left(\mathbf{p}\_{\text{NIR}} + \mathbf{p}\_{\text{red}}\right) \tag{3}$$

$$(1 \text{+0.5}) \ (\mathbf{p}\_{\text{NIR}} \text{--} \mathbf{p}\_{\text{rb}}) / \left(\mathbf{p}\_{\text{NIR}} \text{--} \mathbf{p}\_{\text{red}} \text{+0.5}\right) \tag{4}$$

$$\mathbf{P\_{red}} / \mathbf{P\_{NIR}} \tag{5}$$

Where:

3) were calculated based on the metadata file using equations 3 and 4. Following this, several vegetation indices were calculated. In addition, different false colour composites were produced to assess the ability of the system to detect the known leakages from the satellite

For Point 1 at *Pyla* area, leakage detection was difficult using medium resolution images. Monitoring of the pipeline using the red and the near infrared part of the spectrum for Point 1 did not reveal any significant changes of reflectance due to the water leakage. Similarly,

However, for Point 2, Landsat 7 ETM+, promising results were found. As shown in Figure 21, the Landsat satellite image dated January 7, 2010, tends to give higher vegetation index values, prior to the water leakage being repaired on February 18, 2010. However, the above hypothesis is applicable to other areas of the water pipe as well. The above results have shown the limitations of using medium resolution satellite images for the detection of water leakages,

**Point 2: "Frenaros-Choirokoitia" pipeline**

55530

**Figure 21.** NDVI values using Landsat 7 ETM+ images used over Point 2. The red square highlights the area where the

In an effort to explore further the information extracted using satellite data the three pilot areas

**Distance (m)**

22/12/2009 7/1/2010 13/4/2010

56002

Day watepipe fixed: 18-02-2010

56446

vegetation indices (NDVI) did not show any differences for Point 3 (*Anglisides* area).

images.


leakage was observed.


0,000

54506

55065

were examined separately. Three vegetation indices, the

0,100

0,200

0,300

**NDVI value**

0,400

0,500

0,600

0,700

especially when these are rare and small.

172 Remote Sensing of Environment: Integrated Approaches

pNIR is the near infrared reflectance

pred is the red reflectance

Figure 22 presents the NDVI development during the examined 12 dates of satellite overpasses (see Table 3). Figure 23 presents the SAVI development during the examined 12 dates of satellite overpasses.


**Point 1,2,3: "Frenaros-Choirokoitia" pipeline**

**Figure 22.** NDVI refl. (calculated with Reflectance values) development during the examined 12 dates (Landsat images) in Points 1, 2 and 3

**Points 1,2,3: "Frenaros-Choirokoitia" pipeline**

the soil around Point 2, and subsequently dried after the repair of the water pipe and the

Detection of Water Pipes and Leakages in Rural Water Supply Networks Using Remote Sensing Techniques

In addition, meteorological data provided from the Meteorological Service of Cyprus, indi‐ cate that significant rainfall was recorded on 25, 26 and 27 February 2010, after the pipe line repair date of Point 2 (18 February, 2010). During March and April of 2010, only 1.0 and 2.1 mm of rain were recorded for the same location. Such information provides additional validation that the main factor affecting the NDVI, SAVI and RVI values is the presence or absence of vegetation as a result of soil moisture before and after the pipeline leakage repair. Regarding Point 3,intheAnglisides area, Septemberprecipitationdatadidnot affectthepipe leakage since no significant rainfall was recorded before and after the pipeline repair (17 September, 2008).

**Points 1,2,3: "Frenaros-Choirokoitia" pipeline**

Point 1 Point 2 Point 3

http://dx.doi.org/10.5772/39309

175

1 2 3 4 5 6 7 8 9 10 11 12 **Dates**

**Figure 24.** RVI refl. (calculated with Reflectance values) development during the examined 12 dates (Landsat images)

Remote sensing techniques have been found to be effective both for the detection of the water pipes and for the detection of water leakages. The preliminary results of this study have shown that remote sensing techniques are able to detect areas of the pipeline with water leakages. Ground spectroradiometric data along with the low altitude spectroradiometer system indicate significant differences in the reflectance values in areas where leakage is observed. In addition, crop and soil marks can be used for mapping the actual footprint of the water pipe. Although the use of medium resolution satellite images for monitoring extensive pipelines may be problematic, such as in Points 1 and 3 in the "*Franaros - Choirokoitia*" pipeline, this may

evaporation of the soil water.

0,000

**6. Discussion and remarks**

1,000

2,000

3,000

**RVI value**

in Points 1, 2 and 3

4,000

5,000

6,000

**Figure 23.** SAVI refl. (calculated with Reflectance values) development during the examined 12 dates (Landsat images) in Points 1, 2 and 3

Based on the graphs of Figure 22, the NDVI values present the following pattern: during May 2007, in all 3 points of the known water leakage, NDVI decreases significantly with values close to -0.8, when almost in all cases NDVI is above zero with similar values. After September 2008, the NDVI values increase until April 2010 when they decline again. Such results indicate that the vegetation of the area around the study points reflects soil moisture resulting from rainfall as it can differentiate according to season. Detailed examination of each point related to the pipeline repair indicates that NDVI in Points 1 and 2, water leakage ceased just after the 2nd and the 11th date in correspondence: (a) Point 1: -0,72 and 0,17 for days 2 and 3 and (b) Point 2: 0,60 and 0,09 for days 11 and 12 respectively.

For Point 1, there is a significant change of NDVI value before and after the repair date of the pipeline. In Point 2, the NDVI value decreased significantly (from 0, 60 to 0, 09) following the repair of the pipeline.

However, in Point 3 there is no significant change of the NDVI value before and after the repair date of the pipeline. Although there is a slight decrease in NDVI values immediately following the repair, there is a significant increase within 2 weeks: Point 3: 0,16; 0,13 and 0,42 for days 7 -9 respectively.

The results indicate that only at Point 2 is there a significant decline of NDVI values as a result of lack of soil moisture around the pipe. Another factor can be that due to the temporal difference between the two measurements, of 7 January 2010 and 13 April 2010, respectively, as lack of rainfall may have resulted in moisture evaporation. The same conclusion is reached with SAVI data (Figure 23). The value of SAVI in Point 2 was 0,35 in January 2010 and declined to 0,07 just after the pipeline repair.

Figure 24 presents RVI data which were calculated using equation 7. The RVI index indicates the effect of soil moisture around Point 2. The RVI value in Point 2, in January 2010 was 4,02 and after the pipeline repair, it decreased to 1,21. It seems that the vegetation developed on the soil around Point 2, and subsequently dried after the repair of the water pipe and the evaporation of the soil water.

In addition, meteorological data provided from the Meteorological Service of Cyprus, indi‐ cate that significant rainfall was recorded on 25, 26 and 27 February 2010, after the pipe line repair date of Point 2 (18 February, 2010). During March and April of 2010, only 1.0 and 2.1 mm of rain were recorded for the same location. Such information provides additional validation that the main factor affecting the NDVI, SAVI and RVI values is the presence or absence of vegetation as a result of soil moisture before and after the pipeline leakage repair. Regarding Point 3,intheAnglisides area, Septemberprecipitationdatadidnot affectthepipe leakage since no significant rainfall was recorded before and after the pipeline repair (17 September, 2008).

**Points 1,2,3: "Frenaros-Choirokoitia" pipeline**

**Figure 24.** RVI refl. (calculated with Reflectance values) development during the examined 12 dates (Landsat images) in Points 1, 2 and 3
