Nat. Hazards Earth Syst. Sci., 10, 1807–1814, 2010www.nat-hazards-earth-syst-sci.net/10/1807/2010/doi:10.5194/nhess-10-1807-2010© Author(s) 2010. CC Attribution 3.0 License.

Natural Hazardsand Earth

System Sciences

Monitoring subsidence effects in the urban area of ZonguldakHardcoal Basin of Turkey by InSAR-GIS integration

H. Akcin 1, H. S. Kutoglu1, H. Kemaldere1, T. Deguchi2, and E. Koksal1

1Zonguldak Karaelmas University, Department of Geodesy and Photogrammetry Engineering, 67100, Zonguldak, Turkey2Nitetsu Mining Consultants Co., Ltd., Tokyo, Japan

Received: 23 November 2009 – Revised: 1 July 2010 – Accepted: 29 July 2010 – Published: 1 September 2010

Abstract. Zonguldak Hardcoal Basin is the largest bitu-minous coal region in Turkey where extensive undergroundmining activity exists. Because of this activity subsidence ef-fects have been experienced in different locations of the city.In this study, surface deformations caused by the subsidencehave been observed by D-InSAR technique using C-BandRADARSAT data. InSAR data process of 16 RADARSATimages acquired between 24 July 2005–23 October 2006 hasresulted in significant deformations in the order of about 6 cmin the most populated region of the city. The deformationmap obtained has been integrated with digitized mine pro-duction maps and Quickbird Orthoimage into GIS. Accord-ing to GIS analysis, there are three mine seams at differentlevels driven below the deformed zone. Many governmentaland private buildings located in this area have a high poten-tial risk of subsidence damage. Also, this area covers ap-proximately 12 km of transportation routes.

1 Introduction

Zonguldak is the capital city of the Zonguldak province, lo-cated at the Western Black Sea coast of Turkey, 360 km eastof Istanbul and 270 km north of Ankara (Fig. 1). The topo-graphy around the city is very undulating and steep; 56% ofthe city’s land is surrounded by mountains, and only 29%with the slope with less than 20% suitable for settlementand agriculture. Also, very heavy forests cover the land sur-face in the immediate vicinity of the city centre (ZonguldakProvince Environment and Forestry Directorate, 2006).

Forestry and fishing are among the important sources ofincome in the city with a population of 200 thousand whilethe major industry is the hard coal mining, so that Zongul-

Correspondence to:H. S. Kutoglu(kutogluh@hotmail.com)

dak is the most famous mining region of Turkey. Under-ground coal mine extraction in the basin was initiated in 1848when the Ottoman Empire, before Republic of Turkey, ruledthe land. Now, Turkish Hardcoal Enterprises (TTK) is thecompany authorized for mine production in the basin. Ac-cording to the official records of TTK, the hard coal pro-duction is about 2.5 million tons per year, and has totallyreached 400 million tons since 1848. There are widespreadcoal seams located between the levels of +155 m and−550 munder the city (Turkish Hardcoal Enterprise, 2008).

Due to the above-mentioned mining activities, major sub-sidence events occurred at some locations of the city in thepast (Fig. 2). Some locations are still suffering from subsi-dence phenomena, and some locations are under the risk ofsubsidence damage. For the safety of human life and pro-perty, monitoring the temporal evolution of the subsidenceeffects for the city of Zonguldak is very crucial. Also, sucha monitoring program can facilitate defining the locations atrisk, early warning before major occurrences, and proper ur-ban planning for the future.

Regarding the deformation monitoring, GPS is the mostpowerful geodetic technique producing the most precise, re-liable and exact results to detect pointwise surface deforma-tions. However, to keep wide grounds under control, Differ-ential InSAR is today’s most useful geodetic technique. GPSmay need thousands of site points to monitor an area of in-terest which can be controlled only through a pair of InSARimages.

Therefore, D-InSAR was decided as the main techniquefor this study, since subsidence coming out in the basin isformerly not known. GPS was preferred for validation ofthe results obtained from InSAR. The surface deformationmaps obtained from D-InSAR analysis were integrated withQuickBird Ortho-image into Geographical Information Sys-tem (GIS). Finally, the deformation areas in different levelswere determined, and buildings, roads etc. located in thoseareas were documented by a GIS analysis.

Published by Copernicus Publications on behalf of the European Geosciences Union.

1808 H. Akcin et al.: Monitoring subsidence effects in Zonguldak, Turkey, by InSAR-GIS

MONITORING SUBSIDENCE EFFECTS IN THE URBAN AREA OF ZONGULDAK

HARDCOAL BASIN OF TURKEY BY INSAR-GIS INTEGRATION

H. Akcin1, H.S. Kutoglu

1, H. Kemaldere

1, T. Deguchi

2 and E. Koksal

1

1Zonguldak KaraelmasUniversity, Department of Geodesy and Photogrammetry Engineering, 67100,

Zonguldak, TURKEY, (hakanakcin@hotmail.com, kutogluh @hotmail.com, kemaldere@hotmail.com) 2 Tokyo, JAPAN, Nitetsu Mining Consultants Co., Ltd., Japan. (t_degu@nifty.com)

Abstract

Zonguldak Hardcoal Basin is the largest bituminous coal region in Turkey where extensive underground mining

activity exists. Because of this activity subsidence effects have been experienced in different locations of the city. In

this study, surface deformations caused by the subsidence have been observed by D-InSAR technique using C-Band

RADARSAT data. InSAR data process of 16 RADARSAT images acquired between 2005/7/24-2006/10/23 has

resulted in significant deformations in the order of about 6 cm in the most populated region of the city. The

deformation map obtained has been integrated with digitized mine production maps and Quickbird Orthoimage into

GIS. According to GIS analysis, there are three mine seams at different levels driven below the deformed zone.

Many governmental and private buildings located in this area have a high potential risk of subsidence damage. Also,

this area covers approximately 12 km long transportation routes.

1. Introductıon

Zonguldak is the capital city of the Zonguldak province, located at the Western Black Sea coast

of Turkey, 360 km east of Istanbul and 270 km north of Ankara (Fig. 1). The topography around

the city is very undulating and steep; 56% of the city lands are surrounded by mountains, and

only 29% with the slope of less than %20 is suitable for settlement and agriculture. Also, very

heavy forests cover the land surface in the immediate vicinity of the city center [1].

Figure 1. Location of Zonguldak city and TTK hard coal basin (source: GoogleEarth)

Fig. 1. Location of Zonguldak city and TTK hard coal basin(source: GoogleEarth).

2 Detecting surface deformations usingInSAR technique

InSAR technique has found a wide use since the late 1990sfor monitoring ground deformations caused by landslide,tectonic motions, natural and mining-induced subsidenceevents, and etc. InSAR applications for landslides canbe exemplified by Colesanti et al. (2003), Colesanti andWasowski (2006), Hilley et al. (2004), Kimura and Yam-aguchi (2000). This technique has been mostly appliedfor monitoring tectonic motions (Kimura and Yamaguchi,2000; Burgmann et al., 2006; Delouis et al., 2002; Johan-son and Burgmann, 2005; Wright et al., 2004; Burgmannet al., 2002). It has also been used successfully for sur-face effects of subsidence induced by natural causes or manmade activities (Poland et al., 2006; Amelung et al., 1999;Tomas et al., 2005; Dixon et al., 2006; Fernandez et al.,2009). However, InSAR technique, for deformation moni-toring, depends a lot on local conditions such as topography,vegetation, atmosphere as well as resolution of data. There-fore, each study can be regarded unique. For instance, inTurkey, many successful studies were carried out for moni-toring tectonic movements, however, no successful study todetect mining-induced deformations has been carried out yet.

Forestry and fishing are among the important sources of income in the city with a population of

200 thousand while the major industry is the hard coal mining, such that Zonguldak is the most

famous mining region of Turkey. Underground coal mine extraction in the basin was initiated in

1848 when Ottoman Empire before Republic of Turkey ruled the land. Now, Turkish Hardcoal

Enterprises (TTK) is the company authorized for mine production in the basin. According to the

official records of TTK, the hard coal production is about 2.5 million tons per year, and has

totally reached 400 million tons since 1848. There are widespread coal seams located between

the levels of +155 m and -550 m under the city [2].

Due to the above-mentioned mining activities, major subsidence events occurred at some

locations of the city in the past (Fig. 2). Some locations are still suffering from subsidence

phenomena, and some locations are under the risk of subsidence damage. For the safety of

human life and property, monitoring the temporal evolution of the subsidence effects for the city

of Zonguldak is very crucial. Also, such a monitoring program can facilitate definition of the

locations under the risk, early warning before major occurrences, and proper urban planning for

the future.

Figure 2. Subsidence scenes from different places of the city

As regards deformation monitoring, GPS is the most powerful geodetic technique producing the

most precise, reliable and exact results to detect pointwise surface deformations. However, to

keep wide grounds under control Differential InSAR is today’s most useful geodetic technique.

GPS may need thousands of site points to monitor an area of interest which can be controlled

through only a pair of InSAR images.

Therefore, D-InSAR was decided as the main technique for this study since where subsidence

comes out in the basin is formerly not known. GPS was preferred for validation of the results

obtained from InSAR. The surface deformation maps obtained from D-InSAR analysis were

Fig. 2. Subsidence scenes from different places of the city.

2.1 Data choice

Surface deformations in heavily forested areas can be de-tected best through L-band SAR data, which can penetratevegetation (Gens, 2006; European Space Agency, 2009). Forthat reason, L-band ALOS Palsar data was considered firstto detect the surface deformations in the Zonguldak basin.However, sufficient data pairs having a baseline less than500 m could not be found. Therefore, it was decided to uti-lize the data from C-band RADARSAT, because orbit stabil-ity is quite good. Figure 4 shows the acquisition dates of16 RADARSAT images used for the study.

The image dated 24 July 2005 was chosen as the mas-ter image and others were considered slave images (Fig. 3).Therefore, 15 InSAR data pairs were composed to detect sur-face deformations in relation to the master image. Each pairhas the baseline less than 500 m.

2.2 InSAR processing for deformation monitoring

InSAR uses phase information of SAR data. The distancechange between a sensor and the ground can be measuredfrom phase difference of two observations using phase pro-perty in slant-range length:

φ = φorbit+φtopo+φatm+φdef+φnoise (1)

whereφorbit is the orbit fringe caused by baseline distanceobtained by two observations whileφtopo is the topographicfringe with respect to terrain. These are described by thefollowing equations

φorbit =4πBpara

λ(2)

φtopo=4πhBperp

λρsinα. (3)

Nat. Hazards Earth Syst. Sci., 10, 1807–1814, 2010 www.nat-hazards-earth-syst-sci.net/10/1807/2010/

H. Akcin et al.: Monitoring subsidence effects in Zonguldak, Turkey, by InSAR-GIS 1809

integrated with QuickBird Ortho-image into Geographical Information System (GIS). Finally,

the deformation areas in different levels were determined, and buildings, roads etc. located in

those areas were documented by a GIS analysis.

2. Detecting surface deformations using InSAR technique

InSAR technique has found a wide use since the late 1990s for monitoring ground deformations

caused by landslide, tectonic motions, natural and mining-induced subsidence events, and etc.

InSAR applications for landslides can be exemplified by [3-6]. This technique has been mostly

applied for monitoring tectonic motions [6-11]. It has also been used successfully for surface

effects of subsidence induced by natural causes or man made activities [12-16]. However, InSAR

technique for deformation monitoring highly depends on local conditions such as topography,

vegetation, atmosphere as well as resolution of data. Therefore, each study can be regarded

unique. For instance, in Turkey, many successful studies were carried out for monitoring tectonic

movements; however, any successful study to detect mining-induced deformations had not been

carried out yet.

Data Choice

Surface deformations in heavily forested areas can be detected best through L-band SAR data,

which can penetrate vegetation [17, 18]. For that reason, L-band ALOS Palsar data was

considered first to detect the surface deformations in the Zonguldak basin. However, sufficient

data pairs having baseline less than 500 m could not be found. Therefore, it was decided to

utilize the data from C-band RADARSAT, of which orbit stability is quite good. Figure 4 shows

the acquisition dates of 16 RADARSAT images used for the study.

Figure 3. Radarsat data used for the study

The image dated 2005/7/24 was chosen as the master image and others were considered slave

images (Fig 3). Therefore, 15 InSAR data pairs were composed to detect surface deformations in

relation to the master image. Each pair has the baseline less than 500m.

InSAR processing for deformation monitoring

InSAR uses phase information of SAR data. The change of distance between a sensor and the

ground can be measured from phase difference of two observations using phase property in

slant-range length:

noisedefatmtopoorbit φφφφφφ ++++= (1)

where orbitφ is the orbit fringe caused by baseline distance obtained by two observations while

topoφ is the topographic fringe with respect to terrain. These are described by the following

equations

16 images

05/7/24 05/9/10 05/10/28 06/3/21 06/5/8 06/6/25 06/8/12 06/9/29 05/8/17 05/10/4 05/11/21 06/4/14 06/6/1 06/7/19 06/9/5 06/10/23

RADARSAT

No observation (snow season)

Fig. 3. Radarsat data used for the study.

education campus located in the most densely populated area of the city. Also, the university

campus is located in the blue zone. The maximum rate of the deformation is obtained as 5.6 cm

per 15 months. A detailed temporal development of the deformation, detected from the each pair,

is given in Fig. 6 for the foregoing area.

Figure 4. D-InSAR processing procedure used in this study [22].

Figure 5. Final deformation map around Zonguldak Urban Area for 15 months

5.6 cm

Black Sea

City centre

Kozlu

Kilimli

Master image

Slave image

Fig. 4. D-InSAR processing procedure used in this study (Deguchi et al. 2006).

In these equations,BparaandBperprepresent parallel and per-pendicular components of baseline, respectively. Alsoh, λ,ρ andα represent elevation, wavelength, slant-range lengthand incidence angle, respectively. Finally,φatm is a phasedelay caused by the reflection of microwaves in the watervapor layer,φnoise is the error component caused by ther-mal noise, or temporal and spatial decorrelation associatedwith baseline distance or scattering characteristic change,andφdef represents the amount of surface deformation dur-ing the period between two observations (see Massonnet andFeigl, 1998; Franceschetti and Lanari, 1999; Hanssen, 2001).Radar Interferometry - Data Interpretation and Error Analy-sis, Springer Verlag, New York. The whole process in whichthe deformation interferograms are obtained is explained bya scheme in Deguchi et al. (2006), illustrated in Fig. 4.

In order to measure the time series of deformation, we ap-plied the smoothness-constrained least-squares method to thepixels satisfying (a) and (b) below:

(a) capable of phase unwrapping,

(b) coherence values greater than 0.1.

The amount of deformation since 24 July 2005 was solvedas an unknown parameter under the condition thatU in theEq. (4) was minimized.

U =

∑(φ

(k)i,j − t

(k)i,j

)2

+ α2∑(

first or second difference oft(k)i,j

)2(4)

whereφ(k)i,j is the amount of deformation ofk-th periodic ob-

servation since 24 July 2005. The pixel location is indicated

www.nat-hazards-earth-syst-sci.net/10/1807/2010/ Nat. Hazards Earth Syst. Sci., 10, 1807–1814, 2010

1810 H. Akcin et al.: Monitoring subsidence effects in Zonguldak, Turkey, by InSAR-GIS

education campus located in the most densely populated area of the city. Also, the university

campus is located in the blue zone. The maximum rate of the deformation is obtained as 5.6 cm

per 15 months. A detailed temporal development of the deformation, detected from the each pair,

is given in Fig. 6 for the foregoing area.

Figure 4. D-InSAR processing procedure used in this study [22].

Figure 5. Final deformation map around Zonguldak Urban Area for 15 months

5.6 cm

Black Sea

City centre

Kozlu

Kilimli

Master image

Slave image

Fig. 5. Final deformation map around Zonguldak Urban Area for15 months.

using i and j . The amount of deformation ofk-th peri-odic observation estimated by smoothness-constrained con-dition is described int (k)

i,j . Finally, α2 is a parameter to con-

trol the smoothness oft (k)i,j and the optimum value is de-

cided by minimizing Akaike’s Bayesian Information Crite-rion (ABIC) given below.

ABIC = nlog2π +nlogσ 2+n+2nlogdet

(I +α2DT D

)− nlogα2 (5)

wheren is the number of unknown parameters andσ 2 is de-fined byU/n. Also, I is the identity matrix andD is the matrixof first or second difference oft (k)

i,j (Goldstein and Werner,1998; Akaike and Kitagawa, 1999).

2.3 Results

Applying the above-mentioned procedure for theRADARSAT data pairs have resulted in the deforma-tion map given by Fig. 5. Since RADARSAT is a C-bandmicrowave sensor, it is difficult to provide high coherencebetween master and slave images in densely, vegetatedareas. Consequently, fringe with spatial continuity wasobserved only in the urban district. As seen from the figure,there are significant surface deformations near Kozlu district(Fig. 5). The core (maximum) area of the deformation inred, coincides with the secondary and high school educationcampus located in the most densely populated area of thecity. Also, the university campus is located in the blue zone.The maximum rate of the deformation is obtained as 5.6 cm

Figure 6. Temporal deformation developments from the D-InSAR pairs

3. GPS Validation

D-InSAR can reveal areal deformations over resolution limits of data depending on the factors

stated above. InSAR technique can catch an existing deformation, but can not verify non-existent

deformation. Moreover, deformation rates can be slightly different point to point in equi-

deformation area detected by InSAR. Last, but not least deformations from InSAR are in

direction of slant range. Therefore, the InSAR results must be validated through a more exact

method such as GPS.

In this study, a geodetic network with fifteen site points was established in the study area

(Fig.10). This network was observed 4 times through GPS technique, which is the most effective

surveying method for monitoring pointwise deformations. Fig. 7 shows the time schedule for the

GPS observations.

Fig. 6. Temporal deformation developments from the D-InSARpairs.

per 15 months. A detailed temporal development of thedeformation, detected from the each pair, is given in Fig. 6for the foregoing area.

3 GPS validation

D-InSAR can reveal areal deformations over resolution lim-its of data depending on the factors stated above. InSARtechnique can catch an existing deformation, but can not ver-ify non-existent deformation. Moreover, deformation ratescan be slightly different point to point in equi-deformationarea detected by InSAR. Last, but not least deformationsfrom InSAR are in the direction of slant range. Therefore,the InSAR results must be validated through a more exactmethod such as GPS.

In this study, a geodetic network with fifteen site pointswas established in the study area (Fig. 10). This networkwas observed 4 times through GPS technique, which is themost effective surveying method for monitoring pointwisedeformations. Figure 7 shows the time schedule for the GPSobservations.

The GPS observations were conducted in static modewith one hour sessions. For datum definition, the networkpoints were connected to the four points of Turkish NationalFundamental GPS Network (TUTGA). The observations in

Nat. Hazards Earth Syst. Sci., 10, 1807–1814, 2010 www.nat-hazards-earth-syst-sci.net/10/1807/2010/

H. Akcin et al.: Monitoring subsidence effects in Zonguldak, Turkey, by InSAR-GIS 1811

Figure 7. Time schedule for GPS campaigns

The GPS observations were conducted in static mode with one hour sessions. For datum

definition, the network points were connected to the four points of Turkish National

Fundamental GPS Network (TUTGA). The observations in all the periods were processed into a

batch adjustment with respect to the principle of free network adjustment to obtain the temporal

coordinates of the site points. In the first process, two TUTGA points were found unstable. Then,

these points were removed from the datum definition, and the final results were obtained

regarding the other two TUTGA points.

-10

-8

-6

-4

-2

0

2

05/1

0

06/3

06/6

06/1

0

Period

Defo

rmati

on

rate

in

cm

113 114 115

Figure 8. Temporal deformations at GPS site points in the significant deformation area. The point 112 is not listed

here because it was destroyed during a building construction after the second GPS observation period.

Fig. 8 shows the surface deformations at the site points located in the significant deformation

area determined through InSAR technique. The GPS results validate the existence of a vertical

deformation in this area. The total deformations are -8.2 cm/ 13 months, -7.3 cm/ 13 months and

-4.4 cm/13 months at the points 113, 114 and 115, respectively. In this region, the deformation

rate was obtained 5.6 cm/15 months by InSAR. As stated above, the deformations obtained from

InSAR are areal while those from GPS are pointwise. For a comparison of both methods, an

average deformation rate is obtained 6.6 cm/13 months for the three GPS points. If two months

time difference is ignored coherence of the average GPS rate with InSAR result is %85. In this

respect, the results from both techniques can be regarded as comparable.

4. GIS integration and analysis

After InSAR processing, the underground mine production maps for the related area were

digitized and incorporated to GIS to establish the relationship between the obtained surface

deformations and the mining activity. Also, the final deformation map was integrated to the

system, and equivalent deformation zones with 1 cm intervals have been vectorized for defining

different levels of deformation (Fig. 9). Then, they were both displayed on QuickBird

Orthoimage for the region (Fig. 10).

GPS campaign

2005/10 2006/3 2006/6 2006/10

4 times

Fig. 7. Time schedule for GPS campaigns.

Figure 7. Time schedule for GPS campaigns

The GPS observations were conducted in static mode with one hour sessions. For datum

definition, the network points were connected to the four points of Turkish National

Fundamental GPS Network (TUTGA). The observations in all the periods were processed into a

batch adjustment with respect to the principle of free network adjustment to obtain the temporal

coordinates of the site points. In the first process, two TUTGA points were found unstable. Then,

these points were removed from the datum definition, and the final results were obtained

regarding the other two TUTGA points.

-10

-8

-6

-4

-2

0

2

05/1

0

06/3

06/6

06/1

0

Period

Defo

rmati

on

rate

in

cm

113 114 115

Figure 8. Temporal deformations at GPS site points in the significant deformation area. The point 112 is not listed

here because it was destroyed during a building construction after the second GPS observation period.

Fig. 8 shows the surface deformations at the site points located in the significant deformation

area determined through InSAR technique. The GPS results validate the existence of a vertical

deformation in this area. The total deformations are -8.2 cm/ 13 months, -7.3 cm/ 13 months and

-4.4 cm/13 months at the points 113, 114 and 115, respectively. In this region, the deformation

rate was obtained 5.6 cm/15 months by InSAR. As stated above, the deformations obtained from

InSAR are areal while those from GPS are pointwise. For a comparison of both methods, an

average deformation rate is obtained 6.6 cm/13 months for the three GPS points. If two months

time difference is ignored coherence of the average GPS rate with InSAR result is %85. In this

respect, the results from both techniques can be regarded as comparable.

4. GIS integration and analysis

After InSAR processing, the underground mine production maps for the related area were

digitized and incorporated to GIS to establish the relationship between the obtained surface

deformations and the mining activity. Also, the final deformation map was integrated to the

system, and equivalent deformation zones with 1 cm intervals have been vectorized for defining

different levels of deformation (Fig. 9). Then, they were both displayed on QuickBird

Orthoimage for the region (Fig. 10).

GPS campaign

2005/10 2006/3 2006/6 2006/10

4 times

Fig. 8. Temporal deformations at GPS site points in the significantdeformation area. The point 112 is not listed here because it wasdestroyed during a building construction after the second GPS ob-servation period.

all the periods were processed into a batch adjustment withrespect to the principle of free network adjustment to obtainthe temporal coordinates of the site points. In the first pro-cess, two TUTGA points were found unstable. Then, thesepoints were removed from the datum definition, and the fi-nal results were obtained regarding the other two TUTGApoints.

Figure 8 shows the surface deformations at the sitepoints located in the significant deformation area determinedthrough InSAR technique. The GPS results validate the ex-istence of a vertical deformation in this area. The total de-formations are−8.2 cm/13 months,−7.3 cm/13 months and−4.4 cm/13 months at the points 113, 114, and 115, respec-tively. In this region, the deformation rate was obtained5.6 cm/15 months by InSAR. As stated above, the deforma-tions obtained from InSAR are areal while those from GPSare pointwise. For a comparison of both methods, an averagedeformation rate is obtained 6.6 cm/13 months for the threeGPS points. If two months time difference is ignored coher-ence of the average GPS rate with InSAR result is 85%. Inthis respect, the results from both techniques can be regardedas comparable.

4 GIS integration and analysis

After InSAR processing, the underground mine productionmaps for the related area were digitized and incorporated toGIS to establish the relationship between the obtained sur-face deformations and the mining activity. Also, the final de-formation map was integrated to the system, and equivalent

As seen from Fig. 10, there are mining activities at three different mining levels (-425, -485 and -

560 meters) under the deformed zone. This might be the main reason of the subsidence effects

monitored on the ground.

Figure 9. Different deformation levels vectorized from InSAR data

Figure 10. InSAR deformation map, digitized mine production map and Quickbird Orthoimage combined in GIS

Fig. 9. Different deformation levels vectorized from InSAR data.

As seen from Fig. 10, there are mining activities at three different mining levels (-425, -485 and -

560 meters) under the deformed zone. This might be the main reason of the subsidence effects

monitored on the ground.

Figure 9. Different deformation levels vectorized from InSAR data

Figure 10. InSAR deformation map, digitized mine production map and Quickbird Orthoimage combined in GIS

Fig. 10. InSAR deformation map, digitized mine production mapand Quickbird Orthoimage combined in GIS.

deformation zones with 1 cm intervals have been vectorizedfor defining different levels of deformation (Fig. 9). Then,they were both displayed on QuickBird Orthoimage for theregion (Fig. 10).

As seen from Fig. 10, there are mining activities at threedifferent mining levels (−425,−485, and−560 m) under thedeformed zone. This might be the main reason for the subsi-dence effects monitored on the ground.

For the risk assessment, superstructure over the regionsuch as buildings, roads etc. was extracted from the Quick-bird image, thus, the final product was obtained for furtherGIS analysis (Fig. 11). All the GIS integration processes ap-plied for the study is illustrated in Fig. 12.

www.nat-hazards-earth-syst-sci.net/10/1807/2010/ Nat. Hazards Earth Syst. Sci., 10, 1807–1814, 2010

1812 H. Akcin et al.: Monitoring subsidence effects in Zonguldak, Turkey, by InSAR-GIS

Figure 11. Final GIS product

Figure 12. GIS integration process

XITRF96

ZITRF96

ZITRF96

Mine production map

InSAR deformation map

QuickBird

Orthoimage

Extracted buildings

from Quick Birth

ortho-image

Final product for

hazard analysis

Roads

Mine

tunnel (-200m)

tunnel (-300m)

tunnel (-425m)

tunnel (-485m)

tunnel (-560m)

panel

Building

Government

Residence

Fig. 11. Final GIS product.

Figure 11. Final GIS product

Figure 12. GIS integration process

XITRF96

ZITRF96

ZITRF96

Mine production map

InSAR deformation map

QuickBird

Orthoimage

Extracted buildings

from Quick Birth

ortho-image

Final product for

hazard analysis

Roads

Mine

tunnel (-200m)

tunnel (-300m)

tunnel (-425m)

tunnel (-485m)

tunnel (-560m)

panel

Building

Government

Residence

Fig. 12. GIS integration process.

Using the final product, statistics of the structures affectedby the subsidence were studied by the buffer zone analysis.Thus, buildings and roads located in the different degrees ofthe deformation zones were determined (Fig. 13).

Based on the analysis, the size of the deformed area de-tected by InSAR is totally 1.83 km2. A total of 51 govern-ment and 1040 private buildings (residence) are located in thedetected deformation zones. Also total length of the roads inthese zones is 11.8 km. A detailed distribution of the struc-tures into the deformation zones can be seen in Table 1.

Figure 13. Structures located in the deformation zones

For the risk assessment, superstructure over the region such as buildings, roads etc. was extracted

from the Quickbird image; thus the final product was obtained for the further GIS analysis (Fig.

11). All the GIS integration process applied for the study is illustrated in Fig. 12.

Using the final product, statistics of the structures affected by the subsidence were studied by the

buffer zone analysis. Thus, buildings and roads located in the different degrees of the

deformation zones were determined (Fig. 13).

Based on the analysis, the size of the deformed area detected by InSAR is totally 1.83 km2. A

total of 51 government and 1040 private buildings (residence) are located in the detected

deformation zones. Also total length of the roads in these zones is 11.8 km. A detailed

distribution of the structures into the deformation zones can be seen in Table 1.

1st degree deformation zone 4th degree deformation zone

2nd degree deformation zone 5th degree deformation zone

3th degree deformation zone 6th degree deformation zone

Fig. 13. Structures located in the deformation zones.

Finally, the buildings were investigated on site; the damaged ones were documented, and the

documents were relocated into the GIS (Fig. 14). In Fig 14, a high school damaged due to the

subsidence is illustrated from the designed GIS.

Table 1. Distribution of the structures into the deformation zones

Deformation zone Area (m2) Government

building

Private building Road length (m)

1st degree 44 216.04 1 (High school) 26 525

2nd

degree 83 805.19 4 (1 high school, 2

gyms)

35 782

3rd

degree 138 937.46 2 (1 high school) 73 1 461

4th

degree 180 140.77 6 (2 university

buildings, 1 primary

school, 1 hostel)

121 2 969

5th

degree 290 715.80 17(7 university

building

162 5 107

6th

degree 1 094 307.47

(totally 1 832 122.73)

21(12 university

buildings, 1 gym, 4

hostels

623 7 239

Figure 14. An example of the damaged buildings documented on site (from a high school)

5. Conclusions

As stated previously, Zonguldak lands have very steep topography with heavy forests cover. In

this difficult geography, not only a heavy mining activity is carried out underground but also a

dense urban life survives on the ground. In this context, Zonguldak city is a rare place in the

world. This is the first study that the subsidence effects around the city were successfully

observed through the InSAR technique and interpreted by GIS. At the same time, this is also the

first successful study in Turkey in which D-InSAR is applied for monitoring the mining-induced

surface deformations. Furthermore, an extensive mining activity carried out under a densely

populated city makes InSAR-GIS integration necessary. In this regard, the study is one of the

first studies in which urban deformations monitored by InSAR results are analyzed within GIS.

For instance, a similar investigation was performed for the city of Naples, Italy [25].

Fig. 14. An example of the damaged buildings documented on site(from a high school).

Finally, the buildings were investigated on site; the dam-aged ones were documented, and the documents were relo-cated into the GIS (Fig. 14). In Fig. 14, a high school dam-aged due to the subsidence is illustrated by the designed GIS.

5 Conclusions

As stated previously, Zonguldak lands have very steep topo-graphy with heavy forest cover. In this difficult geography,not only is a heavy mining activity carried out undergroundbut also a dense urban life survives on the ground. In thiscontext, Zonguldak city is a rare place in the world. This is

Nat. Hazards Earth Syst. Sci., 10, 1807–1814, 2010 www.nat-hazards-earth-syst-sci.net/10/1807/2010/

H. Akcin et al.: Monitoring subsidence effects in Zonguldak, Turkey, by InSAR-GIS 1813

Table 1. Distribution of the structures into the deformation zones.

Deformation zone Area Government building Private building Road length(m2) (m)

1st degree 44 216.04 1 (high school) 26 525

2nd degree 83 805.19 4 (1 high school, 2 gyms) 35 782

3rd degree 138 937.46 2 (1 high school) 73 1461

4th degree 180 140.77 6 (2 university buildings, 1 primary school, 1 hostel) 121 2969

5th degree 290 715.80 17 (7 university building) 162 5107

6th degree 1 094 307.47 21(12 university buildings, 1 gym, 4 hostels) 623 7239(totally 1 832 122.73)

the first study that the subsidence effects around the city weresuccessfully observed through the InSAR technique and in-terpreted by GIS. At the same time, this is also the first suc-cessful study in Turkey in which D-InSAR is applied formonitoring the mining-induced surface deformations. Fur-thermore, an extensive mining activity carried out under adensely populated city makes InSAR-GIS integration neces-sary. In this regard, the study is one of the first studies inwhich urban deformations monitored by InSAR results areanalysed within GIS. For instance, a similar investigationwas performed for the city of Naples, Italy (Lanari, et al.,2004).

In this study, processing of the 15 RADARSAT data pairsresulted in subsidence of about 6 cm/15 months between24 July 2005–5 September 2006. This is quite a rapid defor-mation rate. The deformation area monitored surrounds themost populated area in the city, where many buildings androads are located. The mine production maps of TTK showthat there are mining activities at the three different min-ing levels (−425, −485, and−560 m) under the deformedzone. These activities might be the main reason of the sub-sidence effects monitored on the ground. In addition, theintense building stock in the region might be another factortriggering the subsidence. According to the GIS analysis, itwas determined that 51 government, 1040 private buildingsand 11.8 km long roads are in the 1.8 km2 risky area. Thosestatistics reveal the dimension of the problem experiencedin Zonguldak city. If the deformation rate progresses at thesame rate the structures in the zone can be exposed to de-structive effects.

Acknowledgements.The authors thank the anonymous reviewers.Their contributions greatly improved the paper. Another thank goesto Hasan Gercek from Zonguldak Karaelmas University for editingthe language of the paper.

Edited by: G. G. R. IovineReviewed by: M. Berber and four other anonymous referees

References

Akaike, H. and Kitagawa, G.: The practice of time series analysis,Springer-Verlag, New York, 327–339, 1999.

Amelung, F., Galloway, D. L., Bell, J. W., Zebker, H. A., and Lacz-niak, R. J.: Sensing the ups and downs of Las Vegas: InSARreveals structural control of land subsidence and aquifer-systemdeformation, Geology, 27(6), 483–486, 1999.

Burgmann, R., Ayhan, M. E., Fielding, E. J., Wright, T. J., Mc-Clusky, S., Aktug, B., Demir, C., Lenk, O., and Turkezer, A.: De-formation during the 12 November 1999 Duzce, Turkey, Earth-quake, from GPS and InSAR, B. Seismol. Soc. Am., 92(1), 161–171, 2002.

Burgmann, R., Hilley, G., Ferretti, A., and Novali, F.: Resolvingvertical tectonics in the San Fransisco Bay Area from perma-nent scatterer InSAR and GPS analysis, Geology, 34(3), 221–224, 2006.

Colesanti, C. and Wasowski, J.: Investigating landslides with space-borne Synthetic Aperture Radar (SAR) Interferometry, Eng.Geol., 88(1–2), 173–199, 2006.

Colesanti, C., Ferretti, A., Prati, C., and Rocca, F.: Monitoring land-slides and tectonic and with the Permanent Scatterers Technique,Eng. Geol., 68(1–2), 3–14, 2003.

Deguchi, T., Kato, M., Akcin, H., and Kutoglu, H. S.: AutomaticProcessing of Interferometric SAR and accuracy of surface de-formation measurement, SPIE Europe Remote Sensing, Stock-holm, Sweden, 2006.

Delouis, B., Giardini, D., Lundgren, P., and Salichon, J.:, (2002).Joint inversion of InSAR, GPS, teleseismic, and strong-motiondata for the spatial and temporal distribution of earthquake slip:application to the 1999 Izmit mainshock, B. Seismol. Soc. Am.,92(1), 278–299, 2002.

Dixon, T. H., Amelung, F., Ferretti, A., Novali, F., Rocca, F.,Dokka, R., Sellall, G., Kim, S.-W., Wdowinski, S., and Whit-man, D.: Subsidence and flooding in New Orleans, Nature, 441,587–588, 2006.

European Space Agency: Synthetic Aperture Radar Land Applica-tions, Tutorial, 2009.

Fernandez, P., Irigaray, C., Jimenez, J., et al.: First delimitation ofareas affected by ground deformations in the Guadalfeo RiverValley and Granada metropolitan area (Spain) using the DInSARtechnique, Eng. Geol., 105, 84–101, 2009.

www.nat-hazards-earth-syst-sci.net/10/1807/2010/ Nat. Hazards Earth Syst. Sci., 10, 1807–1814, 2010

1814 H. Akcin et al.: Monitoring subsidence effects in Zonguldak, Turkey, by InSAR-GIS

Franceschetti, G. and Lanari, R.: Synthetic Aperture Radar Process-ing, CRC Press, New York, 1999.

Gens, R.: Data Selection and Interferometric Baselines, in InSARand its Applications, Lecture Notes, University of Alaska, Fair-banks, 2006.

Goldstein, R. M. and Werner, C. L.: Radar interferogram filteringfor geophysical applications, Geophys. Res. Lett., 25(21), 4035–4038, 1998.

Hanssen, R. F.: Radar Interferometry – Data Interpretation and Er-ror Analysis, Springer Verlag, New York, 2001.

Hilley, G. E., Burgmann, R., Ferretti, A., and Novali, F.: Dynam-ics of slow-moving landslides from permanent scatterer analysis,Science, 304, 1952–1955, 2004.

Johanson, I. A. and Burgmann, R.: Creep and quakes onthe northern transition zone of the San Andreas fault fromGPS and InSAR data, Geophys. Res. Lett., 32, L14306,doi:10.1029/2005GL023150, 2005.

Lanari, R., Zeni, G., Manunta, M., Guarino, S., Berardino, P., andSansosti, E.: An integrated SAR/GIS approach for investigatingurban deformation phenomena: a case of study of the city ofNaples, Italy, Int. J. Remote Sens., 25(14), 2855-2862, 2004.

Kimura, H. and Yamaguchi, Y.: Detection of landslide areas us-ing satellite radar interferometry, Photogramm. Eng. Rem. Sens.,66(3), 337–344, 2000.

Massonnet, D. and Feigl, K. L.: Radar Interferometry and its Appli-cation on Changes in the Earth’s Surface, Rev. Geophys., 36(3),441–500, 1998.

Poland, M., Burgmann, R., Dzurisin, D., Lisowski, M., Masterlark,T., Owen, S., and Fink, J.: Constraints on the mechanism of long-term, steady subsidence at Medicine Lake volcano, northern Cal-ifornia, from GPS, leveling, and InSAR, J. Volcanol. Geoth. Res.,150(1–3), 55–78, 2006.

Tomas, R., Marquez, Y., Lopez-Sanchez, J. M., Delgado, J., Blanco,P., Mallorqui, J. J., Martinez, M., Herrera, G., and Mulas, J.:Mapping ground subsidence induced by aquifer overexploitationusing advanced Differential SAR Interferometry: Vega Media ofthe Segura River (SE Spain) case study, Remote Sens. Environ.,98, 269–283, 2005.

Turkish Hardcoal Enterprise: Annual Report 2008, Turkish Hard-coal Enterprise, Zonguldak, 2008.

Wright, T. J., Parsons, B., England, C., and Fielding, E. J.: InSARobservations of low slip rates on the major faults of western Ti-bet, Science, 305(5681), 236–239, 2004.

Zonguldak Province Environment and Forestry Directorate:Zonguldak Province Environmental Condition Report, Ministryof Environment and Forestry, Ankara, 2006 (in Turkish).

Nat. Hazards Earth Syst. Sci., 10, 1807–1814, 2010 www.nat-hazards-earth-syst-sci.net/10/1807/2010/