PSI-based monitoring of structural deformations in Istanbul lagoon regions

Abstract

Structural health monitoring (SHM) is a critical concern, especially for metropolitan cities exposed to natural hazards and environmental influences. Remote sensing plays a vital role in monitoring surface movements by taking advantage of regional image acquisition at regular intervals. In this study, the persistent scatterer interferometric synthetic aperture radar (PSInSAR) method was applied to Sentinel-1 satellite data acquired between 2018 and 2021 to monitor surface deformations in the lagoon region between the Büyükçekmece and Küçükçekmece Lakes, Istanbul. This region is particularly significant due to its proximity to the North Anatolian Fault Zone, the presence of active landslides, and clay-rich geological formations. Analysis of line-of-sight (LOS) deformation results from both ascending and descending orbits revealed deformation rates ranging from −9.5 to +7.3 mm/year and −9.9 to +6.6 mm/year, respectively. Six regions containing various engineering structures were selected for investigation. The study identified the impacts of active landslides on structures as well as the seasonal effects associated with clayey geological formations. Additionally, the deformations observed in coastal structures (e.g., piers and breakwaters) and thermal effects on bridge structures (e.g., viaducts) were evaluated based on the obtained results. Overall, the findings demonstrate that PSI effectively captures both regional and structure-specific deformation patterns. The study highlights the potential of PSI-based multi-temporal InSAR techniques for establishing continuous monitoring frameworks that support early warning, maintenance planning, and informed decision-making in urban renewal and infrastructure management in Istanbul.

Introduction

Earthquakes are among the most dangerous geological disasters, posing significant threats to human life, property, and the economy both directly and indirectly. The North Anatolian Fault (NAF), stretching 1300 km from east to west, is the most seismically active in Türkiye. Historical records of major earthquakes along the NAF indicate a westward progression, resembling a domino effect. ¹ The last major earthquake on this fault occurred in Izmit in 1999, and the next significant event is expected in Istanbul. Istanbul, a metropolitan city with a population exceeding 15 million, serves as the country’s economic, touristic, and cultural hub. It is home to numerous critical engineering structures, including bridges, viaducts, ports, and airports, which are vital links in national and international transportation networks. Before the anticipated major Istanbul earthquake, it is essential to assess the current condition of these key structures, reinforce them if necessary, and ensure continuous monitoring.

Selecting the most suitable technique for measuring and monitoring deformations caused by structural, atmospheric, and environmental factors requires consideration of several criteria: (i) physical characteristics of the structure being monitored (e.g., size and location), (ii) environmental conditions (e.g., geological properties of the ground, regional tectonic activity, and prevailing atmospheric conditions), (iii) anticipated nature and magnitude of the structural movements (which determines whether continuous or periodic monitoring is required and the necessary measurement accuracy), and (iv) economic feasibility of the research. Based on these criteria, geodetic or nongeodetic (geotechnical) monitoring techniques can be applied separately or in combination. ²

The concepts of deformation and monitoring have a significant role in engineering structures. The deformations may result from human activities ³ or natural processes such as fault ruptures ⁴ and landslides. ⁵ The deformations can be categorized into two groups: those that progress slowly over time due to the usage of buildings and the effects of environmental factors, and those that occur suddenly due to natural disasters. In this context, structural health monitoring (SHM) plays a key role in monitoring slowly progressing deformations and deterioration of structures.^6,7

SHM relies on in-situ measurements, such as sensors that are permanently located on the structures. ⁸ However, point-based in-situ measurements including sensors, and traditional surveying methods, such as leveling and global navigation satellite system (GNSS), have high accuracy but do not effectively represent the whole structure and its surroundings, and are labor-intensive. ⁹ Remote sensing technologies offer a practical solution for temporal change analysis of engineering structures, benefiting from image acquisition from the same area at regular intervals. Developments in synthetic aperture radar (SAR) technology, especially in the last decade, have increased interest in the use of interferometric SAR (InSAR) technologies for the detection of deformations in structures and SHM applications.^10–14 InSAR technologies enable cost-effective, time-saving, and labor-saving analysis for various engineering structures such as bridges,^15,16 transportation infrastructures,^17,18 historical structures, ¹³ airports, ⁹ and buildings in residential areas.^19,20

Various multi-temporal InSAR methods have been developed to monitor the progression of deformations. However, persistent scatterer InSAR (PSI) and Small BAseline Subset (SBAS) are the most widely used methods in the literature. While SBAS is more effective in rural and bare areas lacking stable reflectors, ²¹ the PSI method has been widely used for artificial surfaces and engineering structures due to their strongly reflective elements such as light poles, house roofs, road guardrails, and so on, and can detect deformations in time series up to millimetric level (i.e. 0.1–2 mm/year ²² ) depending on the spatial resolution of SAR image.^23–25 Toward this end, the PSI method has been successfully applied to various SAR satellite data, including TerraSAR-X/TanDEM-X, COSMOSkyMed, ERS, and ENVISAT.^22,26,27 However, in the multi-temporal analysis of SAR satellite images, two key parameters gain importance: revisit time and spatial coverage. ²⁸ In addition to offering short revisit time and wide spatial coverage, Sentinel-1 satellite images have been freely provided by the European Space Agency (ESA) since 2014. As a result, it has gained significant attention in time-series deformation analyses and has been frequently used for monitoring engineering structures using the PSI method.^29–32

The literature indicates that multi-temporal InSAR (MT-InSAR)-based SHM adopts different approaches depending on the type of structure under investigation. For example, Macchiarulo et al. ³³ integrated time series InSAR and geographical information systems to perform a detailed stability assessment of infrastructure networks. Miano et al. ³⁴ classified bridges according to risk levels based on interferometric analysis results, while Mele et al. ³⁵ used time series InSAR data to identify active deformation zones in urban areas, particularly focusing on buildings. In most previous studies, individual structural types were examined in detail.^36–38 In contrast, this study provides a comprehensive assessment by analyzing multiple types of engineering structures individually within a single region affected simultaneously by landslides, earthquakes, and environmental factors.

Within the scope of the study, deformation analysis of the region located between the Büyükçekmece and Küçükçekmece Lakes was evaluated by utilizing MT-InSAR analysis with Sentinel-1 satellite images. In previous studies, deformations occurring in the region were measured using GNSS and leveling measurements.^39–42 In addition to traditional surveying measurements, InSAR analyses were also carried out for different purposes. Akarvardar et al. ⁴³ investigated the damage that occurred after the Izmit Earthquake of 1999 for the study region using E?uropean Remote-Sensing Satellite (ERS) data acquired between 1992 and 1999. Diao et al. ⁴⁴ examined secondary fault activity in the region with the help of the PSI method and Environmental Satellite (ENVISAT) images acquired between 2002 and 2010. Aslan et al. ⁴⁵ investigated long-term ground subsidence motions that occurred in Istanbul during the period between 1992 and 2017 by combining multi-track/sensor SAR datasets. Bayik et al. ⁴¹ evaluated landslide behaviors in the region using Sentinel-1 and ALOS-2 satellite data acquired between 2015 and 2020 with the PSI technique. Previous studies in the region mainly focused on regional surface movements related to landslides and earthquakes; however, the deformation of individual engineering structures in this region has not been systematically investigated. This study, therefore, provides a novel contribution by evaluating the structural movements of various engineering structures (building, viaduct, harbor, and pier) between 2018 and 2021 in both ascending and descending orbits using PSI. The findings provide critical insights for urban renewal and disaster risk reduction in Istanbul, supporting decision-makers in strengthening infrastructure and ensuring resilience against the anticipated major earthquake.

Study area

Istanbul, Türkiye’s most populous city of more than 15 million people, has an important geographical position as the city that connects two continents, Asia and Europe. The study area is located on the European side of the city, which comprises 39 districts, within the borders of four districts: Büyükçekmece, Avcılar, Esenyurt, and Beylikdüzü, as shown in Figure 1(a). The city, which is the economic, cultural, and historical center of Türkiye, has been experiencing serious population growth driven by migration. This situation leads to an increase in urbanization, and the districts in question, especially Beylikdüzü and Esenyurt, are among the most affected districts. ⁴¹

Figure 1.

(a) The location of the study area and (b) its geological map (modified from the studies by Alp et al. ⁴⁶ and Akbas et al. ⁴⁷ ). IMM: Istanbul Metropolitan Municipality.

The study area is gaining importance in many aspects. One of the primary reasons is its high susceptibility to landslides. The region between Büyükçekmece and Küçükçekmece Lagoon Lakes is well known for landslides and has been the focus of numerous geological and geodetic studies.^40,41,48 According to the landslide awareness booklet prepared by the Istanbul Metropolitan Municipality (IMM), this area contains numerous old, active, and passive landslide zones, primarily of the flow and sliding types.^49,50 The reason for the occurrence of so many landslides in this region is attributed to meteorological factors and soil structure. ⁵¹ The region exhibits the dominant climate characteristics of the Marmara region, with the coastline having a temperate climate. Rainfall is higher in winter than in summer, with an annual average temperature of approximately 15.3°C and a monthly total rainfall average of 662.5 mm—both exceeding the national average. ⁵² It is known that landslides generally occur during rainy seasons due to infiltration of water into the upper permeable soil layer, which weakens the soil structure. ⁴³

When the soil structure of this region is examined, different lithological formations are present as shown in Figure 1(b). Geologically, the area is composed of the Danişmen (Oligocene–Lower Miocene), Çekmece (Upper Miocene), and Ceylan (Oligocene) formations, along with the Lower Miocene Kıraç Member and alluvial deposits in lake and stream beds. The Danişmen Formation (Gürpınar Member), which is located around Büyükçekmece Lake, consists of claystone-shale-type fine crumbs interbedded with sandstone and milestone. The Çekmece formation (Bakırköy member), situated between the Büyükçekmece and Küçükçekmece lakes, is dominated by limestone and clay-marl. The Kıraç member consists of gravel, sand, and silt, characterized by a permeable to semipermeable soil structure. The region generally has a semi-impermeable structure. ⁵³ It has been stated that the area is prone to landslide formation due to permeable sandstone layers and impermeable claystone and siltstone layers. ⁵⁴

Another significant aspect of the study area is its proximity to the NAF line. Although Istanbul has significant potential in terms of social and economic aspects, it is also one of the regions with the highest risk in terms of earthquakes. ⁴⁹ This is primarily because a substantial portion of the NAF Zone (NAFZ) lies within the Marmara Sea, with the study area located approximately 10 km away from it (Figure 1(a)).^55,56 It is known that seismic activities not only trigger landslides but also have effects on the structures in the region.

Materials and method

SAR technology, which can actively provide data in all weather conditions and has the capacity to detect both day and night, plays an irreplaceable role in the fields of remote sensing and time series deformation analysis. In the study, the time series deformation analysis was performed using Sentinel-1 SAR satellite data acquired in interferometric wide mode. Sentinel-1 consists of two satellites: Sentinel-1A and 1B, launched in 2014 and 2016, respectively, and are provided free of charge by the ESA. Sentinel-1 satellites have medium resolution C band imagery with 5 × 20 m spatial resolution in range and azimuth. However, the Sentinel-1B mission ended in 2022 due to a malfunction in its instrument. For this reason, the monitoring period was determined between 2018 and 2021. Time series deformation analysis was performed with 240 single look complex (SLC) satellite images in both descending and ascending orbit directions. The Sentinel-1 SLC images have both vertical-vertical (VV) and vertical-horizontal polarizations; however, the deformation analysis was conducted using the VV polarization. The specifications of Sentinel-1 satellite data used are given in Table 1.

Table 1.

Specifications of Sentinel-1 SAR satellite data used.

Satellite	Orbit	Spatial resolution	Track	Incidence angle (°)	Number of images	Monitoring period	Master date
Sentinel-1 C band	Ascending	5 × 20 m (range × azimuth)	58	39.3	240	2018–2021	30 December 2019
	Descending		36	33.7	240	2018–2021	23 November 2019

SAR: synthetic aperture radar.

To perform time series deformation analysis, the PSI method was applied using the Sentinel Application Platform (SNAP) software ⁵⁷ and StaMPS application package. ⁵⁸ In the PSI technique, interferograms are generated from a single master image, and permanent scatterer (PS) points—targets that maintain consistent reflective properties across all interferograms—are identified. The satellite image resolution, as well as the temporal and perpendicular baselines, influence the density of PS points. Higher-resolution imagery typically allows the identification of a larger number of PS points, enhancing the precision of deformation analyses. ⁵⁹ Although the intensive use of C-band satellite data is due to the free availability of Sentinel-1 satellite data, it is also indicated that medium spatial resolution data can be sufficient to determine and monitor deformations in structures.^31,60,61 In the PSI technique, construction sites, road maintenance works, snow coverage, agricultural lands, forests, and water surfaces may prevent the detection of PS points. In contrast, PSs are often abundant in urban areas containing buildings, monuments, antennas, masts, conductors, exposed rocks, or top surfaces. ⁵⁹ In this study, time series deformation analysis was performed using the PSI method, as shown in Figure 2, with the SNAP and StaMPS softwares.

Figure 2.

The workflow of the study.

In the PSI analysis, interferograms were generated based on the master image using SNAP software, which is designed for processing and analyzing Earth observation datasets, especially Sentinel satellite products. The temporal and perpendicular baselines based on the master images are shown in Figure 3 for both the ascending and descending datasets. All Sentinel-1 SAR images were processed with default multi-looking factors (5 × 1 pixels). In the topographical phase removal step, the Shuttle Radar Topography Mission (SRTM) 1-arc second digital elevation model (30 m) was used. The produced interferograms were processed using the StaMPS application package with an amplitude dispersion threshold of 0.4 and the default parameters of StaMPS. The atmospheric correction step was applied using the toolbox for reducing atmospheric InSAR noise ⁶² to eliminate tropospheric effects utilizing a linear atmospheric model widely used for tropospheric corrections.^63,64 The obtained deformation information was evaluated and visualized using open-source QGIS software.

Figure 3.

Perpendicular baseline graph of the SAR images used for the study area in (a) ascending and (b) descending orbit direction. SAR: synthetic aperture radar.

The temporal coherence of PS points varies between 0 and 1. Values close to 1 indicate high quality and more reliable results, as they reflect the stability of the frame.^16,65,66 The histograms of temporal coherence values of PS points obtained in both ascending and descending orbits are given in Figure 4. It was observed that coherence values were concentrated above 0.60 in both orbit directions. Accordingly, among the PS points, those with a coherence value above 0.60 were taken into consideration to increase the reliability of the results.

Figure 4.

Histogram of the temporal coherence for (a) ascending and (b) descending orbit direction.

When analyzing the structure’s deformation time series, the time series deformation data were decomposed into three components—trend, seasonality, and residual (noise)—using the additive decomposition method. The additive decomposition method assumes that a time series is the sum of these three components and separates them accordingly. For the structure-based time series analysis, only the trend component was considered.

Since the deformation velocities were produced in the satellite’s line-of-sight (LOS) direction, a decomposition was performed to derive the horizontal (east–west) and vertical (up–down) components of movement. The north–south deformation component was neglected due to the low sensitivity of LOS measurements in that direction. The vertical and horizontal components were calculated as follows:

$$\left(\begin{array}{c}{d}_{\mathrm{asc}}\hfill \\ \multicolumn{1}{c}{d_{\mathrm{dsc}}}\end{array}\right)=\left(\begin{array}{cc}\cos {\theta }_{\mathrm{asc}}\hfill & -\cos {\alpha }_{\mathrm{asc}}\sin {\theta }_{\mathrm{asc}}\hfill \\ \multicolumn{1}{c}{\cos {\theta }_{\mathrm{dsc}}}& -\cos {\alpha }_{\mathrm{dsc}}\sin {\theta }_{\mathrm{dsc}}\hfill \end{array}\right)\left(\begin{array}{c}{d}_{\mathrm{ver}}\hfill \\ \multicolumn{1}{c}{d_{\mathrm{hor}}}\end{array}\right)$$

(1)

where θ and α represent the incidence and satellite heading angles for the ascending and descending directions, respectively. To enable the intersection of PS points from both orbits, a grid network with an approximate resolution of 200 × 200 m was established. Different grid sizes have been adopted in previous PSI studies, depending on the study objectives and the spatial characteristics of the area.^41,67,68 In this study, the grid size was chosen as 200 m to balance spatial resolution and PS density, ensuring a sufficient number of coherent PS points within each cell and enabling reliable overlap between the two orbit geometries. This choice is particularly relevant for heterogeneous environments that include dense urban fabric, viaducts, and coastal infrastructures, where PS distribution is highly variable due to differences in construction materials, orientation, and SAR viewing geometry. In densely built urban areas, PS density ranges between approximately 80 and 300 points per cell, providing statistically stable estimates of deformation rates. While smaller grid sizes can increase spatial resolution for individual structural elements, they may lead to sparse PS coverage and unstable displacement estimates, particularly when points from both orbits intersect. Following the gridding process, the vertical (up–down) and horizontal (east–west) displacement components were calculated using Equation (1).

Results and discussions

The deformation information of the study area was extracted from the aforementioned Sentinel-1 satellite images for both orbit directions (ascending and descending) for the years between 2018 and 2021. Deformation velocities derived from the interferograms generated using the PSI method are shown in Figure 5. The results obtained represent the LOS directions for both ascending and descending orbital directions. The subsidence regions are gradually shown with red and orange points, the stable regions with yellow points, and the uplift regions with blue and green points. The gaps seen in the deformation maps indicate areas without construction or regions where construction is still in progress.

Figure 5.

LOS deformation velocity results obtained for (a) ascending and (b) descending orbit direction. LOS: line-of-sight.

According to Figure 5, the deformation velocities vary between −9.5 and +7.3 mm/year in ascending and −9.9 and +6.6 mm/year in descending direction. The negative (−) sign represents the movement away from the satellite, while the positive (+) sign represents the movement through the satellite. Movements away from the satellite were observed in the regions located west of the Büyükçekmece Lake and southeast of the study area in both orbit directions, whereas movements through the satellite were detected in the regions located south of the Büyükçekmece Lake in both orbit directions. The results of this study reveal a distribution that aligns with the findings reported by Bayik et al. ⁴¹ and Aslan et al., ⁴⁵ although their analyses were conducted during different periods, leading to variations in deformation magnitudes. Both studies identified similar movements in the same regions using different SAR datasets (ERS-1/2, ENVISAT, ALOS-2, and Sentinel-1). While Bayik et al. ⁴¹ reported that deformation rates in the region ranged from −10 to 6 mm/year between 2015 and 2020, the present study, covering 2018–2021, similarly determined deformation rates between −10 and 7.3 mm/year. Notably, Aslan et al. ⁴⁵ demonstrated that subsidence in the area has persisted since 1992. However, neither study offered a detailed analysis of the deformation time series.

The relationship between LOS velocities and their standard deviations is given in Figure 6. It is observed that most of the PS points have low standard deviations concentrated around 0.2 mm/year. Similarly, according to the distribution of the LOS velocity values in Figure 6, the LOS velocities of the PS points are mostly distributed between −2.5 and +2.5 mm/year in both orbital directions.

Figure 6.

Distribution of velocity and standard deviation for ascending and descending orbit direction.

The LOS deformation velocities were decomposed using Equation (1) and results are given in Figure 7. According to the results, vertical movement is identified in areas where both orbits indicate motion in the same direction, whereas horizontal (east–west) movement is observed in regions where the directions of motion differ. In Figure 7(a), red indicates subsidence, while blue represents uplift. Similarly, red denotes movement toward the west, whereas blue indicates movement toward the east in Figure 7(b). Subsidence is observed in the areas west of both Büyükçekmece Lake and Küçükçekmece Lake. Horizontal movements are also detected locally in the region between the two lakes.

Figure 7.

(a) Vertical and (b) horizontal velocity (mm/year) distribution for the study area. Red arrows indicate the direction of the velocity.

In the study area, there are various engineering structures, including settlements, harbors/piers, and bridges/viaducts. Six different case regions were selected to investigate the deformation behaviors of the structures. Figure 8 shows the distribution of the case regions over the study area. The first three regions (A, B, and C) comprise settlements with distinct characteristics. Regions D and E represent the harbor and piers, respectively, while region F shows the viaduct located in the region. While vertical and horizontal deformation velocities were analyzed to characterize regional deformation patterns, the interpretation of individual structures relied primarily on LOS displacement time series, which are less affected by geometric uncertainties and provide a more robust representation of structural deformation behavior. Therefore, each region will be examined in detail with the deformation time series in the following sections.

Figure 8.

The evaluated structure types: settlements (A, B, and C), harbor and piers (D and E), and viaduct (F) shown with black boxes on ascending LOS deformation results. LOS: line-of-sight.

Deformation analysis of the settlements

For the evaluation of settlements’ deformation analysis, three different case regions were selected. The first region (region A) is located west of the Büyükçekmece Lake, as shown in Figure 8. The ascending and descending deformation results of region A are given in Figure 9(a) and (b). The movement away from the satellite is predominant in both directions in the region, similar to the results of Aslan et al. ⁴⁵ Moreover, the deformation time series of two points representing the same region are shown in Figure 9(c), along with their velocities, coherence, and standard deviation values.

Figure 9.

LOS deformation maps of region A in (a) ascending and (b) descending orbits. (c) LOS Time series deformations of two PS points (1 and 2) shown as black dots on LOS deformation maps. The dashed lines indicate the linear trend of time series deformation for selected PS points. LOS: line-of-sight; PS: permanent scatterer.

According to the results obtained, it is observed that the movement away from the satellite, which includes seasonal effects, is dominant in region A. It is found that the movements in the ascending and descending directions were consistent with each other, and the correlation coefficient of the ascending and descending deformation time series shown in Figure 9(c) was found high (0.96). The coherence values of the selected PS points for ascending and descending orbits exceed 0.85, and the standard deviations are around 0.25 mm/year, indicating that the results are trustworthy. There is movement away from the satellite with a seasonal effect in the LOS direction in both orbits, and the velocities of the movements are −5.6 and −6.7 mm/year, respectively. At the end of the 3 years, the cumulative deformation in the LOS direction reached 20 mm.

While a movement through the satellite is observed in rainy months (winter and spring seasons), a movement away from the satellite is observed in seasons when the temperature increases (summer and early autumn). This situation may be due to the characteristics of the soil structure. In clay soils, volume increases and decreases according to the water content. The water content of such soil increases during the rainy seasons due to the infiltration of precipitation accumulated on the surface into the ground and decreases due to evaporation in the hot seasons. ⁶⁹ The fact that the soil structure of the region is semipermeable ⁵³ and the swelling potential of the Gürpınar formation, which constitutes the geological structure of the region, except for sandy silt, is at a critical limit ⁷⁰ also supports this situation.

The second settlement region, region B, is located west of the Küçükçekmece Lake. Deformation maps of the region obtained in both orbits are given in Figure 10(a) and (b). Moreover, the deformation time series of two points representing the same region are shown in Figure 10(c), along with their velocities, coherence, and standard deviation values.

Figure 10.

LOS deformation maps of region B in (a) ascending and (b) descending orbits. (c) LOS time series deformations of two PS points (1 and 2) shown as black dots on LOS deformation maps. The dashed lines indicate the linear trend of time series deformation for selected PS points. LOS: line-of-sight; PS: permanent scatterer.

Considering Figure 10, the general deformation distribution in the region is similar for both orbits, except for the eastern part of the region. The LOS deformation time series of two PS points taken from the same locations in both deformation maps were found to be highly correlated, with a correlation value of 0.94. It is observed that the results obtained from PS points with coherence values above 0.80 and low standard deviation values are reliable. The velocities of the points are −5.94 and −6.64 mm/year, respectively. The cumulative deformations of the points in the LOS directions of the satellites were found to be approximately 25 mm of subsidence.

While the eastern part of the region showed movement through the satellite in the ascending orbit, it showed movement away from the satellite in the descending orbit. Such different movements seen in the orbits, which are similar to the results obtained by Bayik et al., ⁴¹ indicate that the movement in the region is in the horizontal direction. There are three types of landslides in the region: active, passive, and fossil, and it is known that sliding movement dominates in all of these landslide types. ⁵⁰ Avcılar district, where region B is located, is prone to landslides due to its permeable sandstone layers and impermeable claystone, siltstone, or mudstone layers.^54,71 Aslan et al. ⁴⁵ noted that the region has a history of slow-moving landslides, and as observed in this study, there was both horizontal and vertical movement in the region.

The central part of the region, where the time series was extracted (Figure 10), includes industrial facilities, while the surrounding areas are primarily residential. The similar motion observed in both orbits over the industrial zone indicates vertical movement, whereas discrepancies in the residential areas point to horizontal movement. This distinction is vital for identifying buildings in residential zones that require urban renewal and for assessing potential risks in disaster preparedness, particularly in the context of earthquakes.

Another region showing similar movements to the east side of region B is region C. Region C is located in the north of the study area and east of Büyükçekmece Lake. Deformation maps of the region obtained in both orbits are given in Figure 11(a) and (b), and the deformation time series of two PS points representing the same region are shown in Figure 11(c) with their velocities, coherence, and standard deviation values.

Figure 11.

LOS deformation maps of region C in (a) ascending and (b) descending orbits. (c) LOS time series deformations of two PS points (1 and 2) shown as black dots on LOS deformation maps. The dashed lines indicate the linear trend of time series deformation for selected PS points. LOS: line-of-sight; PS: permanent scatterer.

When the LOS deformation maps of the region are examined, it is found that the results obtained in the ascending and descending orbits are opposite to each other. The time series of two PS points taken from the region also shows movements in different directions. The first point shows a movement through the satellite in the ascending orbit, while the second point shows a movement away from the satellite in the descending orbit. The correlation value of time series deformations was found to be −0.82. The coherence values of the points are approximately 0.80, standard deviations are 0.20 mm/year, and velocities are +4.19 and −5.91 mm/year, respectively.

According to IMM, ⁴⁹ there are old landslides in the region, and Bayik et al. ⁴¹ also stated that old landslides were reactivated in their MT-InSAR analysis for the years 2015–2020. Considering that the opposite directional movements obtained in the ascending and descending results given in Figure 11(a) and (b) represent horizontal movement, it is observed that similar results were obtained in this study. However, another significant point to note is that the urbanization process continues in the region. In some cases, old (inactive) landslide areas can become active after a while due to natural and/or anthropogenic factors. ⁷² The fault lines in the region and NAFZ trigger both liquefaction and landslides, ⁷³ and some of the active landslides today are formed as a result of the loading of the crown parts of old landslides. ⁷² Furthermore, buildings that were under construction between 2015 and 2020, and whose movement could not previously be detected, were captured in this study for the period 2018–2021. Given the presence of landslides and ongoing construction in the region, it is essential to continue monitoring with a longer time series.

Deformation analysis of the harbor and piers

Monitoring deformations in coastal structures such as harbors and piers, as well as in settlement areas, is possible with MT-InSAR analysis. There are multiple studies in the literature that successfully monitor and extract the deformations of the harbors, piers, and breakwaters.^74,75 In this regard, the structural movements of the harbors and piers in the study region were evaluated. Regions D and E are located on the coast of the Marmara Sea. Region D represents the Gürpınar Pier. The deformations obtained for ascending and descending orbits are given in Figure 12(a) and (b), respectively, and the time series movements of two PS points are presented in Figure 12(c).

Figure 12.

LOS deformation maps of region D in (a) ascending and (b) descending orbits. (c) LOS time series deformation trends of two PS points (1 and 2) on LOS deformation maps, and the red vertical dashed line shows the earthquake in the Marmara Sea on September 26, 2019. The horizontal dashed lines indicate the linear trend of time series deformation for selected PS points. LOS: line-of-sight; PS: permanent scatterer.

The fish market, breakwater, and Gürpınar Pier were built on the area created by sea filling in 2014 and became operational in 2015. According to the deformation results, it is observed that there are subsidence movements in the eastern part of the fish market and the southern part of the pier. The deformation characteristics change from stable to movements away from the satellite toward the ends of the breakwater. To investigate time series characteristics of the movements for both orbital directions, time series deformations of the two PS points are given in Figure 12(c). The coherence values are higher than 0.85, and the standard deviations of the points are around 0.30 mm/year. Both points showed movement away from the satellite at velocities of −5.34 and −3.60 for ascending and descending orbits, respectively. A total movement of approximately 20 mm was detected in the breakwater in both orbits over 4 years. Besides the movement away from the satellite, it is observed that the PS points also presented seasonal movement. Since the presence of the breakwater significantly absorbs the waves, a stable movement prevails in the fish market located behind the breakwaters, similar to the results of Li et al. ⁷⁵ Additionally, at the PS points on the breakwater, the presence of a slight movement away from the satellite, as well as the subsidence trend, is also remarkable (Figure 12(c)). The main reason for this slight movement may be the earthquake with a magnitude of 5.7 that occurred in the Marmara Sea off the coast of Silivri on October 26, 2019. ⁷⁶

The other selected coastal structures are located in region E, which is called Ambarlı, consisting of a harbor and a pier. The deformations obtained for ascending and descending orbits for region E are given in Figure 13(a) and (b), respectively, and the time series movements of three P points are presented in Figure 13(c) to (d).

Figure 13.

LOS deformation maps of region E in (a) ascending and (b) descending orbits. (c) and (d) LOS time series deformation trends of three PS points (1, 2, and 3) on LOS deformation maps, and the red vertical dashed line shows the earthquake that occurred in the Marmara Sea on September 26, 2019. The horizontal dashed lines indicate the linear trend of time series deformation for selected PS points. LOS: line-of-sight; PS: permanent scatterer.

Considering Figure 13, the region showed movement through the satellite in both orbit directions; however, the breakwater located in Ambarlı Marina showed movement away from the satellite in both directions. The three different PS points were selected to present various movement characteristics of the pier and the harbor. All points have high coherence values above 0.60. The first and second points are located in the breakwater of Ambarlı Marina. At the first point, movement away from the satellite was observed with a rate of −5.94 ± 0.29 mm/year in the ascending direction and −3.54 ± 0.23 mm/year in the descending direction. However, the second point showed different movements in the two orbits, indicating that the movement was in the horizontal direction. It showed movement through the satellite with a rate of +3.74 ± 0.32 mm/year in ascending direction and movement away from the satellite with a rate of −4.80 ± 0.23 mm/year in descending direction. The third point is located in the Ambarlı Harbor, which was put into service in 1994 and is the first private harbor of Istanbul, and showed movements through the satellite in both directions with velocities of +2.71 ± 0.27 mm/year and +2.03 ± 0.19 mm/year, respectively.

According to the deformation time series of the points, it is observed that there are seasonal effects on the coastal structures. The movements away from the satellite were observed in both orbits on the left part of the marina’s breakwater. The deformation of point 1, located in the left part of the marina, reaches up to approximately 20 mm due to the seasonal subsidence movement. Unlike the left part, the right part of the marina showed different deformation characteristics in both orbits. The deformation time series of point 2 has seasonal uplift movement in the ascending direction and seasonal subsidence movement in the descending direction. Deformations reached up to approximately 20 mm in both directions. Breakwaters are constructed to ensure the safety of coastal structures and protect them against risks such as storm waves and seawater intrusion. ⁷⁷ For this reason, testing and control of the stability of the breakwater becomes important as it undergoes deformation over time. ⁷⁸

Point 3 is located in the Ambarlı Harbor. In the harbor, since there are continuous logistical activities in the coastal elements, the point density is concentrated on the mainland. Activities such as maintenance, repair, and extension of the pier, which were reported in the media ⁷⁹ and started in 2019, may also be associated with the low number of points on the piers. It is noteworthy that, unlike the marina, the movement through the satellite is dominant in the harbor and is similar in both orbits. The presence of all three types of landslides (active, passive, and old) in the region ⁴⁹ where the port is located, is likely to be among the factors triggering the movement.

When the deformation time series of the three points is examined, the sudden changes that occurred in September 2019 are noteworthy. It is observed that this sudden change is compatible with the date of the 5.7 magnitude earthquake that occurred off the coast of Silivri in the Marmara Sea on September 26, 2019. ⁷⁶ It was found that there were sudden movements away from the satellite in the coastal structures (marina and harbor) in the LOS direction in both orbits due to the earthquake. Due to its proximity to the NAFZ through the Marmara Sea, monitoring and control of structures bordering the Marmara Sea are gaining importance.

The results obtained show that MT-InSAR methods exhibit successful performance in monitoring the effects of earthquakes on structures. These techniques are particularly effective in detecting small-scale surface deformations and subtle structural responses following seismic events. However, large-scale and abrupt deformations caused by earthquakes may be challenging to capture due to coherence loss and phase unwrapping limitations. ⁸⁰ Despite these constraints, the capability of MT-InSAR to reliably identify small-magnitude displacements and postseismic deformation patterns over wide areas highlights its undeniable potential for regional-scale seismic impact assessment. Therefore, MT-InSAR represents a powerful and complementary tool for earthquake-related deformation monitoring, particularly when integrated with in situ geodetic and seismological observations.

Deformation analysis of the viaduct

Bridges and viaducts play an important role in highway transportation. These structures are frequently used, especially in metropolitan cities, to relieve traffic loads. However, deformations may occur over time due to external loads and natural disasters. Therefore, the use of MT-InSAR has become widespread in monitoring deformation formation in structures spanning large areas.^33,81,82

In the study, region F consists of the Karasu Viaduct, which forms part of the Trans European Motorway. The first 1000 m of the viaduct, which is 2160 m long, passes over Büyükçekmece Lake, ⁸³ and this area was examined within the scope of the study. The results obtained by MT-InSAR analysis are shown in Figure 14.

Figure 14.

LOS deformation maps of region F in (a) ascending and (b) descending orbits. (c) and (d) LOS time series deformations of two PS points (1 and 2) on LOS deformation maps. The dashed lines indicate the linear trend of time series deformation for selected PS points. LOS: line-of-sight; PS: permanent scatterer.

According to the results given in Figure 14, yellow points represent stable movement, while green points represent slight movement through the satellite. In this direction, it is observed that the movement on the viaduct is dominated by stable and uplifted movement in some places. It was determined that the coherence values of the PS points on the viaduct were high, which indicates the compatibility of the PSI method in the deformation analysis of the viaduct.

While similar results are obtained in both orbits, the deformation time series of points taken from the same region exhibit similar trends, with seasonal variations and slight movement through the satellite. The velocities of the points are 0.70 ± 0.26 mm/year and 1.08 ± 0.22 mm/year in ascending and descending directions, respectively. Considering the precision of the Sentinel-1 C band (5.6 cm), which is defined as 1.1 mm by Sabater et al., ²³ movements lower than 2 mm/year can be considered as insignificant. ⁸⁴ However, seasonal movements may still be observed through time series deformation analysis. Since the warming of the weather in spring and summer causes expansion in concrete and steel structure materials, ⁸¹ temperature triggers seasonal movements in such structures. ¹¹ To investigate the relationship between LOS deformation and temperature, temperature above 2 m was extracted from the ERA5-Land dataset using Google Earth Engine. The time series of LOS deformation in the ascending direction and air temperature are given in Figure 15. It was observed that LOS deformation exhibited behavior compatible with temperature, and the correlation between deformation and temperature was determined to be 0.80. This reveals the relationship between the structure and the temperature.

Figure 15.

LOS deformation in the ascending direction with air temperature collected from the ERA5-Land dataset. The orange bar and green points indicate the air temperature above 2 m and the LOS deformation for the selected PS point in the ascending direction, respectively. LOS: line-of-sight; PS: permanent scatterer.

Conclusion

Türkiye has faced numerous natural disasters throughout history, especially earthquakes. The recent earthquakes around Istanbul have once again highlighted the critical importance of building safety. In this city, where construction density is the highest, it has become increasingly important to assess whether the structures are ready for the anticipated major earthquake. While remote sensing technologies provide effective solutions for monitoring structural health, they also enable the assessment of geological, meteorological, and environmental impacts on buildings.

This research presents the spatial and temporal distribution of MT-InSAR-based deformation through a case study in Istanbul, utilizing Sentinel-1 satellite images and the PSI method. The results indicate that LOS deformation velocities range between −9.5 and +7.3 mm/year in the ascending direction and between −9.9 and +6.6 mm/year in the descending direction. In most parts of the region, similar LOS movements were observed for both orbit directions, with ascending and descending results generally found to be consistent and compatible.

Deformation analysis was conducted not only at the regional scale but also on a structure-specific basis, evaluating different engineering structures in six distinct areas. The results indicate that seasonal temperature variations appear to cause expansion during summer and movement away from the satellite in winter, particularly in viaducts, whereas the opposite behavior was observed in clay-rich geological regions. In these areas, movement away from the satellite occurred during the dry summer season, while movement through the satellite was observed in the wet winter season. As a result, variations in geological composition and environmental conditions lead to distinct deformation responses in different parts of the study area in the LOS direction.

The study demonstrated that the PSI-based MT-InSAR approach effectively captures structural movement characteristics and small-scale deformation signals. Furthermore, external factors such as landslides, earthquakes, and temperature fluctuations contribute to structural deformations, with their effects varying depending on structure type and geological conditions. Deformations in residential areas, harbors, piers, and viaducts were analyzed while considering these external influences. The information obtained from MT-InSAR facilitates disaster prevention and enables the implementation of precautionary measures without the need for costly and time-consuming field surveys. Nevertheless, the integration of MT-InSAR products with in-situ observations, such as structural sensors, would further enhance the reliability and physical interpretation of time series deformation signals. Such a combined monitoring framework would support the establishment of a more sustainable, multiscale, and long-term deformation monitoring system.

This study was conducted using freely available C-band Sentinel-1 satellite imagery. While detailed structure-specific analyses may benefit from X-band SAR data (i.e., COSMO-SkyMed, TerraSAR-X), which offer higher spatial resolution, improved geolocation accuracy, and the ability to detect more PS points, ³³ Sentinel-1 imagery provides a favorable balance between temporal resolution, spatial coverage, and data accessibility. Its high revisit frequency and wide coverage enable the reliable detection of surface deformation over large areas and across various infrastructure types and structures. Therefore, although Sentinel-1 may not capture sufficient PSs on the small structural components, it is well suited and preferable for regional-scale and structural monitoring, offering a practical and cost-efficient solution for long-term deformation assessment.

Given the presence of active landslides and the proximity to the NAFZ, as well as the ongoing urban expansion in the region between the Büyükçekmece and Küçükçekmece Lakes, continuous monitoring with SAR satellite techniques is recommended, especially for municipal planning and infrastructure investments. Establishing long-term monitoring systems is crucial for improving urban resilience and preparedness for the expected major earthquake in Istanbul. In this context, MT-InSAR-based monitoring frameworks, when applied with an awareness of their limitations and complemented by ground-based data, can provide valuable guidance for urban renewal and structural reinforcement decisions.