Ground infrastructure monitoring in coastal areas using time-series inSAR technology: the case study of Pudong International Airport, Shanghai

ABSTRACT Shanghai Pudong International Airport (PDIA), with its east side built along the coast with weak geological conditions, is prone to uneven foundation settlement due to the consolidation and compression of soil and erosion of coastal tides, affecting the safe operation of the airport. Therefore, it is crucial to conduct dynamic subsidence monitoring within the airport, especially in the runway area. 29 scenes of ascending track Sentinel-1A radar images from August 2016 to June 2018 are selected to perform surface deformation inversion based on PS-InSAR and improved SBAS-InSAR for PDIA and its around coastal area. Through cross-validation, the reliability of the time-series InSAR technique for dynamic monitoring of surface deformation of coastal zone infrastructures is confirmed. The results show severely uneven settlement. By combining the monitoring results with the local geological and hydrological dataset, the driving factors of differential deformation of the infrastructures are analyzed, including stratigraphic geological conditions, ground loadings, foundation treatment methods, water erosion, and groundwater level changes. Finally, the time-series deformation characteristics and the causes of PDIA's runway are emphasized based on the PS deformation results. This case provides a reference for the safety management of critical infrastructure in coastal areas using advanced InSAR technique.


Introduction
With global warming, the global mean sea level continues to rise, which poses a serious challenge to the survival and development of human society (Wang et al. 2021).In 2020, the Department of Marine Early Warning and Monitoring of the Ministry of Natural Resources of China released the 2019 China Sea Level Bulletin.The report states that from 1980-2019, the rate of sea level rise along the coast of China is 3.4 mm/year, which is higher than the global average for the same period.In the past 40 years, the coastal sea level in China has been rising at an accelerated rate.The rising sea level has led to an increase in nearshore wave and tidal energy, and intensified coastal erosion and beach erosion, causing subsidence of varying degrees on the coastal zone ground infrastructure, resulting in serious hazards.And ground subsidence in coastal areas causes relative sea level rise, forming a disaster cycle and increasing the risk of coastal zone hazards.
Since the end of the last century, some coastal cities in China, Japan, and the Netherlands have faced problems of expansion and a shortage of traffic roads to meet economic development needs (Hoeksema 2007).As a result, land reclamation from the sea has been carried out in most coastal areas (Yan et al. 2013).The forming land space is suitable for establishing industrial parks, ports, and airports due to their wide and flat characteristics.However, coastal erosion is increasing with the sea level rising, coupled with the effects of ground loading, geological conditions, groundwater extraction, and others; these reclaimed land are highly susceptible to uneven deformation, posing a potential threat to infrastructure facilities.
To solve these potential problems, traditional methods of land subsidence monitoring are mainly GNSS measurements (Benoit et al. 2015;Ito et al. 2016;Komac et al. 2015), level measurements (Stiros 2004;Vasco 2005), and extrapolation meters.These methods are mostly limited to point measures in a small spatial area, costly, time-consuming, and lacking in timeliness.With the increasing development of Interferometric Synthetic Aperture Radar (InSAR) technology, scholars from various countries have used differential interference, time-series InSAR, Pixel Offset tracking, and other techniques for seismic inversion (Ding et al. 2004;Hooper et al. 2012;Lindsey et al. 2015), landslides (Wasowski and Bovenga 2014;Intrieri et al. 2018;Rosi et al. 2018), glacier movement (Rignot et al. 2000;Luckman, Padman, and Jansen 2010), ground subsidence (Ding et al. 2004;Du et al. 2018;Wang et al. 2022), and other geohazard monitoring.The development of InSAR technology has an incomparable advantage compared with traditional ground-based methods in wide-area deformation monitoring.With the launch of multi-platform radar remote sensing satellites, the temporal resolution of the monitoring becomes increasingly higher, significantly improving the efficiency and reducing the monitoring cost.
The Shanghai Pudong International Airport (PDIA), Xiamen New Airport and Hong Kong International Airport (HKIA) are all typical coastal airports in China.As these infrastructures are built on marine sediments, the reclaimed foundations are highly susceptible to differential settlement or uplift, which can easily cause severe damage to the airport runways and underground facilities structures, affecting the safe operation of the infrastructures.Therefore, it is crucial to conduct land subsidence monitoring within the airport area, especially in the runway area.Domestic and international scholars have carried out some research on airport surface deformation in various environments.Take some inland airports, for example, Iqaluit Airport, Baffin Island, Canada (Short et al. 2014); the Kuala Lumpur International Airport (Marshall et al. 2018); Leonardo da Vinci International Airport, Italy (Ciampoli et al. 2020); and the Beijing International Airport (Gao et al. 2019;Dai et al. 2020), and so on.
Representatively, Zhao et al. (2011) used PS-InSAR to monitor the settlement of the reclaimed foundations of HKIA by modifying with ground truth data; this study confirmed the continuous influence of ocean current motion on the stability of foundations.Cavalie, Sladen, and Kelner (2015) used NSBAS-InSAR to measure ground deformation in the Var delta and the NCA airport, combining the ascending and descending data; the deformation was decomposed into east, west, and vertical components found near vertical deformation and extended over the entire delta.Liu et al. (2019) investigated the land subsidence zone of the reclamation land and the surface rebound uplift zone of the landfill materials using the SBAS-InSAR, revealing the surface deformation pattern and spatial and temporal evolution characteristics of the new Xiamen airport after land reclamation.Wu et al. (2020) used multiple SAR sensors and an improved MT-InSAR method to analyze the historical subsidence processes at Hong Kong International Airport.Miao et al. (2021) mapped the spatial and temporal deformation of Shenzhen Baoan International Airport using ENVISAT ASAR data and Sentinel-1 data under ascending and descending orbits, reflecting the uneven settlement of the reclaimed airport is concentrated on the reclaimed land.
Previous studies primarily focused on airport ground settlement and potential risks; thus, there is a lack of investigation into the multiple driving factors.PDIA, one of the typical coastal airports in China, is partially built on reclaimed land prone to uneven settlement due to various reasons such as soil consolidation and compression, groundwater level changes and coastal erosion, which threaten the safe operation of the airport.The ground settlement, especially in the runway area, directly affects the safety of aircraft take-off and landing.In the absence of ground-based measurements, InSAR technology, with its high resolution in time and space, can bridge this gap and provide a valuable reference for the safe operation of coastal ground infrastructure.In this study, we aim to use InSAR technology to monitor critical infrastructure in and around coastal airports over long periods and investigate settlement dynamics in these infrastructures.
As a first-tier city in the Yangtze River Delta, Shanghai has had many studies on urban ground subsidence since the last century (Yao et al. 2019).Dong et al. (2014) used ALOS-1 data (2007-2010) to obtain the ground subsidence in Shanghai and the driving factors of some essential facilities in the urban area.Ma et al. (2017) estimated the cumulative land subsidence trend using multi-platform SAR and altimetry data, and analyzed the risk of land subsidence and sea level rise in the Shanghai area.Yao et al. (2019) studied the spatial distribution of surface deformation and geological features along a typical profile in Shanghai.Yu et al. (2020Yu et al. ( , 2021) ) quantified the relationship between compressive soils and slow settlement rates on a macroscopic scale, describing the micro-scale pore and structural characteristics of compressive soils.Previous studies have focused on ground settlement behavior in urban areas of Shanghai.In addition, previous studies have found that surface settlement in the main urban areas of Shanghai is highly correlated with groundwater extraction (Xiong and Zhu 2007).In this study, we have monitored more essential infrastructures such as airports, seawalls and highway roads in coastal areas and explored the driving factors that induce subsidence in conjunction with basic geological and hydrological information.
By using the PS-InSAR technique and the enhanced SBAS-InSAR technique, we inverted the surface deformation field of the coastal shore of Shanghai Pudong New Area from August 2016 to June 2018.The multiple drivers of the differential settlement are discussed concerning actual geological data, hydrological data and historical optical imagery, focusing on the deformation of the runway area and its causes.The results show that the time-series InSAR technique can monitor coastal infrastructure on a large scale and reflect the evolution of coastal surface subsidence in coastal areas.Long-term deformation monitoring can be implemented to operate critical infrastructure such as coastal airports.This case study also provides some reference for preventing and controlling ground settlement hazards in coastal areas.

Study area overview
Pudong International Airport is situated in the coastal area, west of Huangpu River, east of the Yangtze River mouth, south of Minhang District and Fengxian District, and north of Chongming Island across the Yangtze River.Pudong New Area covers an area of 1210 km 2 , with 12 streets and 24 towns.The area's topography is high in the southeast and low in the northwest, with an average elevation of 3.87 m.The stratum is the alluvium of the Yangtze River, and the shape of the land is triangular.Its coastline is 115 km long and has an oceanic climate with four distinct seasons, abundant precipitation and sunshine.As of 2021, Pudong International Airport has five runways, two satellite halls and two terminals.Figure 1 shows the location and distribution of facilities at Pudong International Airport.
Since its establishment, Pudong International Airport has undergone long-term manual ground monitoring.Shortly after opening the western part of the site, some settlement occurred due to aircraft loading and poor geological conditions.However, after years of settlement consolidation, it is now more stable.However, the eastern part of the site was built on reclaimed land, and settlement continues to be severe.For example, before the construction of Runway 4 in the east part of the airport, a stacking test project was carried out.Several settlement monitoring points were set up, revealing extremely heavy settlements.These results suggest that more severe settlement can occur on freshly-passed runways.The factors contributing to this phenomenon are often diverse, and in this study, we hope to explore the possible driving factors through the use of time-series InSAR techniques.

Technical principles and processing
This study used Sentinel-1A radar images to monitor the Shanghai coastal area.PS points were selected using the amplitude departure method combined with the coherence coefficient method.Processing was carried out by the PSI algorithm (Ferretti, Prati, and Rocca 2001).However, as the coastal area is far from the main urban area, some water bodies, grasses, shrubs, and other vegetation cover reduce the density of PS points.The SBAS algorithm has a better baseline combination than the PS algorithm and obtains a higher point density, making it more suitable for monitoring the region.
However, during SBAS processing, ground control points (GCPs) must be manually selected to correct orbital errors and remove residual flat-earth phases to avoid misestimates introduced by selecting GCPs blindly.In this study, We use the invariant and highly coherent PS points obtained from PS-InSAR processing as GCPs for the SBAS processing instead of the conventional strategy to eliminate orbital phase errors and residual flat-earth phases.The results of the residuals show that this has greatly improved the accuracy of the GCPs.
Figure 2 shows the main flow of this experimental treatment.In this study, we selected 29 ascending orbit Sentinel-1A radar satellite images of Shanghai coastal area between August 2016 and June 2018, and the detailed parameters are shown in Table 1 We use PS-InSAR and improved SBAS-InSAR technique to monitor the dynamic time series of surface deformation of such coastal airports with PDIA.

Permanent scatterer-InSAR
Since the PS-InSAR technique was proposed (Ferretti, Prati, and Rocca 2001), the problem of spatio-temporal decoherence in interferometry has been weakened, and it is also possible to isolate atmospheric delay and other error terms to derive a millimeter-scale surface deformation monitoring results (Colesanti et al. 2003;Ferretti et al. 2007).
Taking N + 1 SAR images as an example, one of the images is selected as the master image and interfered with the other N secondary images to obtain N interferometric pairs, and the phase of PS point M on the ith differential interferometric pair can be expressed as Eq.1.
Where w d (M, t i ) is the phase change caused by surface deformation; w t (M, t i ) represents the phase of externally introduced DEM error; w a (M, t i ) accounts for the phase of atmospheric delay difference between images; w o (M, t i ) is the phase of orbital error; and w n (M, t i ) is the noise phase.
w n (M, t i ) has little effect for the highly coherent PS points (Jiang et al. 2016).We use a linear model to estimate the DEM residual phase and linear deformation rate, then perform phase deconvolution, use high and low pass filtering to eliminate the residual phase and atmospheric delay phase, and finally obtain the deformation phase and convert it to the deformation in the satellite line-of-sight direction.
The core basis of PS-InSAR technology is the selection of PS.The geographic location, resource environment, radar images and other factors of this experimental object are considered comprehensively.In this experiment, the amplitude discrepancy method combined with the coherence coefficient method is chosen to improve the screening method of PS.
In a fixed window, the coherence coefficient of an image element is estimated based on the neighboring image elements in the vicinity of that image element.For the generated 28 time-series interferometric image pairs, the time series of correlation coefficients of image resolution units can be obtained from Eq. 2: g 1 , g 2 , g 3 . . .g N .The average value of the time-series correlation coefficient  a is calculated pixel by pixel, and g threshold g t is set, if g .g t , the pixel can be used as a PS point.
Where X and Y denote the blocks of pixels contained in the inner window region of an arbitrary interference image pair and * denotes the conjugate multiplication.
The amplitude deviation threshold method uses the stability of the image element amplitude to select the PS point, expressed by the amplitude deviation index D as: Where m denotes the mean value of the temporal amplitude; and s denotes the standard deviation of the temporal amplitude.Suppose D t is the amplitude departure threshold, and if the image element amplitude departure index D satisfies D 1 , D t , it can be used as a PS point.The results show that the number of PS points obtained is sufficient and evenly distributed within the study area.Most of them avoided water bodies, crop planting areas, and transition grasses of airport runways and taxiways, which shows that this double-threshold approach can provide high-quality PS points in complex environments.(see Figure 5a).

Short baseline subsets-InSAR (SBAS-InSAR)
SBAS-InSAR is a temporal InSAR method proposed by (Berardino et al. 2002) that is different from the PS-InSAR strategy.Suppose that N scenes of SLC images and M interferograms are generated according to a certain spatio-temporal baseline threshold, and M satisfies Eq. 4.
If there are two images acquired at t 1 and t 2 , the phase of PS point m on the j-th differential interferogram can be expressed as: Where d 1 and d 2 denote the deformation in the radar satellite line-of-sight direction relative to the reference moment.Dw denote the DEM error phase, atmospheric delay phase, orbital error phase, and noise phase, respectively.Without considering the effect of these four phase differences, Eq. 5 can be written as: Expressing the phase in Eq. 6 as an expression for rate and time yields: Expressing the phase of the interferogram as the result of the deformation rate at each time is: Here, the problem of rank loss in matrix A is solved by singular value decomposition (SVD) of matrix A. Finally, the deformation rate is obtained by the Singular Value Decomposition according to Eq. 8. Finally, the integration of the rates is done according to each time period to calculate the form variables in the radar satellite line of sight direction.According to previous studies, the Shanghai area is flat, and the deformation in the vertical direction dominates (Wang et al. 2022).Therefore, we consider the primary settlement in the vertical direction.For a single track, the deformation in the LOS direction can be converted to the vertical by Eq. 9.
When performing PS-InSAR processing, images from September 19, 2017 was selected as the master image and the other 28 images were used as slave images, and 28 sets of interferometric image pairs were generated (Figure 3a).For SBAS-InSAR processing, we set the image spatial baseline threshold to 2% of the maximum baseline and the temporal baseline threshold to 100 days, and a total of 99 interferometric image pairs were generated (Figure 3b).The spatio-temporal baseline distribution of the generated interferometric image pairs by the two methods are shown in Figure 3.

Improved short baseline subsets InSAR
In SBAS-InSAR processing, after the interferometric preprocessing and phase unwrapping, the residual topographic phase and phase ramps usually require recalculation and removal.Sufficient GCPs selected with high quality, i.e. with high coherence showing almost no deformation, are applied for orbit refinement and re-flattening.
However, in such areas of coastal cities with broad interference fringes and high coherence, the artificial selection of ground control points is highly subjective.To reduce the errors caused by artificial selection, we introduce the PSs screened in PS-InSAR processing as the GCPs.We select control points in such a way that the interval is every 5 km × 5 km with 30% overlap and find an optimal control point conforming to the principle of high coherence and far from the deformation zone for orbit refinement re-flattening in SABS-InSAR.
Figure 4 shows the geographic distribution of 36 ground control points filtered from the PS-InSAR results, with some control points shared by overlapping areas.The ground control points introduced in PS-InSAR are processed by orbit refinement and re-flattening, and the obtained results are shown in Table 2, where those with the same color are shared ground control points, and all the obtained GCPs have coherence coefficient values above 0.99.
Table 2 shows that the maximum residual obtained after re-flattening is 1.054, the minimum residual is 0.062, and the average residual is 0.479.The results show that introducing PS points as GCPs to eliminate the orbital phase and residual flat-earth phase results in higher accuracy.Compared to artificially selected control points, the results are somewhat subjective without a priori knowledge, while the artificial introduced error is greater, the residuals of chosen artificially ground control points are between 2 and 10, and the influence is greater in mountainous areas with significant topographic relief.As seen from Figure 5, the distribution of the vertical deformation derived by two time-series InSAR is more consistent, and both show unevenly distributed deformations within PDIA and its surrounding area.Figure 6 shows the histogram results of the distribution of vertical shape variables for each coherent target in the two time-series InSAR monitoring results.It can be seen that both vertical deformation variables are concentrated between −50 and 10 mm, and are consistent.In addition, the number of DS (Distributed Scatterer) points was much higher than the number of PS points; the main reason is that the study area is located in a suburban area where some grasslands and low vegetation reduce the number of PS points.However, the trend of vertical deformation distribution is consistent for both time-series InSAR monitoring.

Coastal infrastructure settlement analysis
Figure 7 shows the coastal area's vertical deformation field, including the airport, at 3-month intervals monitored by the improved SBAS-InSAR technique.We can see the surface deformation process of the coastal infrastructures, demonstrating the time-series InSAR technology can effectively monitor the deformation of coastal areas, which is important for protecting critical coastal infrastructures, such as airports, docks, ports, etc.   Figure 8 shows the spatial distribution of vertical deformation and the schematic diagram of surface facilities obtained by the SBAS-InSAR technique.The monitoring results reveal the deformation process of important infrastructures in the coastal area, such as highways, magnetic levitation tracks and coastal seawalls.The deformation evolution of several linear features can be seen in Figure 7.The Shanghai bypass expressway and magnetic levitation train sections in this area are relatively stable, with settlement rates between −2 and 2 mm/a.However, the Shanghai-Lu Expressway section produced a more serious settlement, especially on the loop section bordering the elevated Disney Resort, where the settlement amount reached more than −50 mm.Meanwhile, the uneven settlement also occurred in the Shenjiahu Expressway section, with settlement coming more than −40 mm at the section entering the airport satellite hall.
As the primary safeguard to resist coastal geological hazards such as storm surge, seawater intrusion and coastal erosion, the coastal seawall project is a critical barrier to ensure the safety of coastal cities.Most seawall roads and outer slopes are made of rigid concrete structures, and the inner side is planted with slope protection vegetation.Research showed that from 2007 to 2015, the overall coastal seawall settlement in Shanghai showed intensified-slowed-exacerbated volatility, especially in the reclaimed area outside the PDIA, where settlement was the most severe, with annual settlement up to −162.1 mm and the percentage of settlement up to 60.5% (Pei, Liao, and Wang 2013;Chen et al. 2016).From our monitoring results, the coastal seawall continued to maintain a more serious settlement trend from 2016 to 2018, with the settlement reaching −40 mm to −80 mm in two years.The factors affecting seawall settlement are diverse, including the type of stratigraphy, compressibility, and lithology that control the spatial pattern of seawall settlement, while water erosion becomes the main driving force.An airport polder river surrounds the inner side of the seawall, and the seawall has experienced more severe settlement under the combined erosion of the paddock river and the East China sea.The deformation of seawalls will weaken the flood resistance of coastal areas and make the drainage capacity of cities more difficult, resulting in serious water accumulation during flooding.SLR also makes the seawall standard higher; although the settlement of the seawall and the rising sea level are two opposite vectors, the coastal cities are superimposed on the harm.This study also provides an important reference to relevant authorities in disaster prevention and control.

Analysis of the causes of differential settlement
The above results show that the study area as a whole produced some differential subsidence from August 2016 to June 2018.While the causes of this phenomenon are diverse, we analyze the driving factors of deformation in combination with relevant information from the field.

Geological conditions
The experimental area is located at the front edge of the Yangtze River Delta, which constitutes a basic Quaternary landform with about 300 m of Quaternary strata.The upper part is marine deposits and the lower is continental deposits.The various infrastructures are built on thick layers of highly compressible loose sediments with complex sediments alternating between sand and clay.Below 70 m of the surface, the distribution of the soil layers is relatively uniform.In contrast, above 70 m, the spatial distribution of the soil layers varies due to geological changes such as scouring by underground rivers (Gong 1995).Previous studies have shown that the elevation gradually decreases from west to east within the site area.The earth's surface down to a depth of 40 m is mainly composed of mucky clay, sandy silt and silty fine sands with horizontal stratification, and the distribution of stratigraphy is shown in Table 3.In addition, from west to east of the airport, the thickness of the soft soil layer increases, the thickness of the mucky clay layer increases, the moisture content and natural void ratio increases, and the compressibility of the soil layer is greater, resulting in a greater settlement deformation of the mucky clay soil (Liu 2018).
As shown in Figure 5, the airport was divided into two parts, east and west, by the black dashed line (AA') located in the middle of the Runway 1 and Runway 2. The east part was a shoal earlier before the coastal reclamation project reshaped it as land.Moreover, the geological conditions were fragile because there were still a lot of silt deposits under the blowing sand.Compared with the east part, the west has an earlier land formation time and flatter terrain.From the soil profile of the four runways, the distribution of soil layers in the central layer and the area below is consistent, but the distribution varies more in plan and space in the shallow layer.Figure 9 shows the vertical displacement of typical feature points in the eastern and western fields, with D14 at the northern end of the taxiway and PS*1-5 at the north end of Runway 1 (see Figure 12 for detailed distribution of points).And from the time-series InSAR monitoring results, it happens to be AA' the dividing line, producing severe uneven settlement on the east and west sides of the site.At the same time, the T1 and T2 terminals on the dividing line also have settlement variability: the T1 terminal in the west has a settlement of about −15 mm at its southern location, while the T2 terminal in the east shows settlement on both sides, with about −25 mm of settlement.The previous study also showed cracks of 2∼5 cm on the transition zone in front of the two terminals were generated (Liu 2018).It can be seen that the difference in geological conditions within the site area becomes a key factor for the uneven settlement of the airport.

Ground loads
With the economic development of the Pudong New Area, the amount of ground load (including dynamic and static load) is increasing, and the sediment is consolidated and compressed under pressure, thus causing surface settlement.For example, the cargo center at Sanjia Port near the airport, where many cargo containers are stacked, has a large ground load and is built on reclaimed land, with a settlement of up to −50 mm or more (see Figure 10d).In addition, the apron around the terminal building and the apron around the runway were subjected to long-term aircraft loading, which also produced a certain amount of settlement, about −20 mm, and the settlement in the cargo stacking area of the airport was also more serious (see Figure 10c).In addition, the settlement in the Disney resort area reached more than −50 mm.Disneyland has seen 11 million visitors in the first year since it opened in 2016, which continues to increase year after year; the dynamic loads from more visitors also impact the surface settlement at Disneyland (Dai et al. 2020) (see Figure 10b).

Foundation treatment method and ancient river
In the airport area, the foundation area is a soft soil layer due to weak geological conditions.The foundation treatment determines the good or bad decision of control of runway settlement, and the way of foundation treatment is different for different runway areas.It has been shown (Liu 2018) that Runway 1, treated by the 'strong ramming' method, had a uniform settlement but a large settlement since its completion in 1999, and the deformation was only evident in the area where the ancient river flowed through.For Runway 2, deep surcharge preloading and shallow dynamic compaction' were used for foundation treatment (Zhang and Huang 2007), which reduced the deformation of the foundation to a certain extent.Compared with other runways, the geological conditions of Runway 3 are better.The foundation treatment was only done by impact rolling' in the shallow layer (Wang et al. 2007), while 'vacuum pre-pressing' was used where the ancient river flowed through.The time-series InSAR monitoring results show that the deformation of the area treated with deep vacuum pre-pressure is significantly smaller than that of the area treated with shallow layers only (see Section 5.4 for details).It can be speculated that different foundation treatments directly influence the deformation of the runway area.

Stream erosion
The rivers and culverts around PDIA are crossed, and the soft soil layer of the foundation of this site may produce some loosening under the erosion effect of flowing water, causing surface deformation.On the coastal seawall east of the airport, the inner and outer sides are subject to the erosive action of the flowing water from the Weichang River and the Yangtze River, respectively, which causes the seawall to settle more severely.Moreover, the different flow velocities of the two rivers have different magnitudes of erosion effects on the coastal seawall, thus causing the difference in settlement rates between the inner and outer sides.The settlement rate of the inner side is −15 mm/ a to −20 mm/a, while the settlement rate of the outer side is more than −35 mm/a.At the same time, we found that the deformation was different in the area near the north and south ends of Runway 4, and the deformation at the north end was more extensive than that at the south end.It can be speculated that the closer the area is to the river, the greater the effect of flowing water erosion on the surface deformation.

Groundwater withdrawal and engineering construction activities
Groundwater extraction is an important dynamic condition that causes surface deformation.The groundwater resources in the study area are abundant and mainly buried in the Quaternary aquifer system.This aquifer system can be divided into one submerged aquifer and five pressurized  and ground subsidence, and the lower the groundwater level, the more serious the ground subsidence.From historical data, the past large-scale exploitation of groundwater in pressurized aquifers has caused a significant reduction in groundwater levels and a series of geological hazards of ground subsidence, which poses a potential threat to the operation of various infrastructures near the coast.
For this purpose, we counted the variations of groundwater levels in each aquifer (A1 to A5) from June 2002 to June 2020 at the groundwater-level tube well monitoring facilities in our study area near the coast.The locations of the monitoring facilities are shown in Figure 1 It can be seen from Figure 11 that the groundwater level of each aquifer kept rising during the study period.It is because, after 2000, groundwater exploitation was restricted under the constraints of relevant policies while the relevant departments opened the groundwater recharge project.Therefore, there is no noticeable subsidence leakage due to groundwater exploitation in this coastal area from the time-series InSAR monitoring results.However, some infrastructure in the study area, such as coastal seawalls, harbors, and offshore runways, still experienced more severe subsidence under continuous groundwater rebound.Therefore, it is reasonable to infer that the change in groundwater levels was not one of the driving factors for the deformation of infrastructure in this coastal area during this period.
At the same time, several areas underwent severe settlement in the study area (circled in red in Figure 8), and we have compared the monitoring results with historical optical images.The results show that most of these settlement areas, except for the cargo stacking area and the logistics center, are construction complexes that have just been completed or are under construction (see Figure 10a).The most typical site is Shanghai Disneyland, completed in 2016.Although not built on land enclosed by the sea, it has also produced a certain amount of subsidence after it has been open for two years.Large areas of subsidence have occurred, particularly in the rides.It is not only because the park's visitor traffic has increased in recent years, but also the constant operation of new rides that have caused some deformation of the entire surface of the park.It can be seen that construction activities become another factor driving the subsidence when groundwater extraction and utilization are in balance.Therefore, it can be inferred that Runway 5, under construction, is affected by construction activities to a certain extent.Meanwhile, due to the proximity to the coastline, the erosion of the Yangtze River and the Weichang River has led to a linear settlement zone on the west side of Runway 5 (see Figure 10e).

Displacement deformation analysis of airport runway
In airport deformation monitoring, the settlement changes in the runway area directly impact the aircraft take-off and landing process, so it is necessary to focus on the analysis of the deformation and causes of the runway area.
For PS-InSAR processing, to screen the PS points, we use the amplitude dispersion index threshold combined with the coherence coefficient threshold, and the obtained PS points are uniformly distributed on the runway area.Each PS point is independent, making its deformation monitoring accuracy for a single target point more accurate.Therefore, this experiment analyzes the runway displacement based on the monitoring results of each PS point.Figure 12 shows the LOS deformation rate of the PS points of the runway area, and it can be found that the deformation rate mainly ranges from −15 mm/a to 10 mm/a, with more severe settlement in the local area.Based on the airport-wide PS-InSAR deformation rate map, some typical PS points on the four runways are selected for analyzing the runway deformation.Figure 13 shows each PS point's cumulative vertical displacement variation on the runway area and its fitted curve.
Figure 13a shows five PS points on Runway 1 with a maximum uplift of 11.22 mm and a maximum deformation of −11.16 mm.The accumulated vertical displacements of PS points on the whole runway are concentrated between −8 and 8 mm, showing that Runway 1 has a relatively stable trend after more than ten years of consolidation and settlement since its completion in 1999.
Compared with Runway 1, the geological condition of Runway 2 is poorer, formed by the land blowing of Yangtze River sediment.Therefore, a deep-shallow layered approach was adopted in the foundation construction, with surcharge preloading in the deep layer and ramming in the shallow layer (Zhang and Huang 2007).From the historical settlement observation data (Liu 2018), since its completion in 2005, Runway 2 has produced uneven settlement along the runway midline longitudinally, with more severe settlement variation at the north end than the south end, while the deformation variation is more stable in the middle.Figure 13c and d shows the vertical deformation of the PS points at the north and south ends of Runway 2. The results show that uneven deformation maintains on the whole runway.The maximum deformation at the north end of the runway reaches −43.12 mm and the accumulated deformation is generally above −15 mm, while at the south end of the runway, the maximum deformation is only −18.50 mm and the overall deformation is above −5 mm, while in the middle section of Runway 2, the displacement rate is between −5 and 5 mm/a.Runway 3 is closer to inland and has relatively good geological conditions.Therefore, impact rolling was performed only in the shallow layer during foundation treatment (Wang et al. 2007), while vacuum preloading was used in the deep layer where the ancient river channel flowed.The past settlement observation results of Runway 3 showed a trend of south section > middle section ≈ north section.The Figure 13e and f shows the vertical displacements of the south and middlenorth sections of Runway 3. In the southern area of Runway 3, the maximum deformation is −33.18 mm, and the general deformation reaches −11 mm or more.The maximum deformation in the middle and north sections of the runway lifts 20.49 mm, and the minimum deformation is −16.57mm.The overall vertical displacement is concentrated between −10 and 10 mm, and the runway's middle and north sections are more stable than the southern area.This tendency is consistent with the changes in settlement observation data in previous periods.The ancient river flows underneath the middle of Runway 3, and vacuum preloading is adopted as the foundation treatment.It is presumed that this treatment effectively suppressed the deformation in the middle of the runway.
Figure 13b shows the vertical displacement time series of Runway 4. From the deformation results, the settlement trend due to surface consolidation and compression is pronounced because Runway 4 was established later than the other runways, and because of the poor geological conditions.The maximum deformation reaches −39.20 mm, and the average accumulated vertical displacement reaches more than −18 mm.
The monitoring results also show that severe settlement occurred at the north and south ends of the taxiway between Runway 2 and Runway 4. According to the vertical deformation results of points D1∼D14 in the Figure 13g and h, the maximum deformation reaches −73.62 mm, and the overall average deformation is more than −35 mm.It can also be found that the closer the monitoring points are to the southern end, the more severe the accumulated settlement displacement is, which also indicates that the closer the coastline is, the more obvious the erosion effect of flowing water is.At the northern end of the taxiway, the overall deformation is severe, with the maximum settlement amount reaching −78.81 mm and the average settlement amount reaching more than −45 mm.Due to the deplorable geological conditions in this area, the site was blown into the land by Yangtze River sediment, and the short construction time of Runway 4, the taxiway part produced more serious settlement under the influence of the passing aircraft load.

Conclusion
In this study, Sentinel-1A data acquired from August 2016 to June 2018 covering the Shanghai area were used to inverse the surface deformation field using both the PS-InSAR and the enhanced SBAS-InSAR techniques.Results have confirmed the feasibility of the time-series InSAR technique in monitoring the surface deformation of coastal infrastructure zone dynamically.The following main conclusions can be obtained: (1) By combining the geological and hydrological data, the factors inducing infrastructures' differential settlement can be summarized as the amount of ground loading, the way the foundations are being treated, the erosion of the flowing water and the construction activities.The factors that induce settlement vary with the location of the infrastructure, but the geological conditions dominate, while others play a predisposing role.Furthermore, with the rebound of groundwater in Shanghai, groundwater extraction is no longer a key source of ground settlement.(2) In the PDIA, severely uneven settlement appears in the east and west part, divided by the black dashed line (AA') of the reclaimed land; with the west side being more stable, the east area is experiencing a severe settlement.The deformation of Runway 1 has stabilized because it has been consolidated for many years.Variable displacement appears along Runway 2's midline longitudinally, with its middle being more stable and both the north and south ends experiencing more severe deformation.The deformation of Runway 3 is relatively stable, although an ancient river flows beneath it.The most severe deformation is found on Runway 4 because of its last completion time and the poorest geological conditions.
In coastal areas, in particular, many cities have undertaken land reclamation projects to address the shortage of land resources.However, these reclaimed lands are prone to surface subsidence, a slow and long-lasting process that is not easily noticed in the early stages.Once subsidence continues to develop, some critical infrastructure within the city is highly susceptible to damage.This paper verifies the feasibility of the time-series InSAR technique for high-precision monitoring of surface deformation in the coastal zone using essential infrastructures along the Shanghai area as experimental objects.The driving factors causing the uneven settlement of each infrastructure are also analyzed in detail.
With the increase of multi-temporal and multi-source SAR data, monitoring the coastal zone's ground settlement is no longer limited to traditional ground-based monitoring methods.Follow-up work can further explore integrating multi-platform, multi-track datasets for surface deformation monitoring in coastal cities, which has important implications.This case study also references ground settlement control in other coastal cities.

Figure 1 .
Figure 1.Map of Pudong International Airport Location and Facilities Distribution.

Figure 2 .
Figure 2. The main process of experimental processing.

Figure 5
Figure 5 gives the cumulative vertical deformation comparison graph of the PDIA retrieved by two time-series InSAR methods.As seen from Figure5, the distribution of the vertical deformation derived by two time-series InSAR is more consistent, and both show unevenly distributed deformations within PDIA and its surrounding area.Figure6shows the histogram results of the distribution of vertical shape variables for each coherent target in the two time-series InSAR monitoring results.It can be seen that both vertical deformation variables are concentrated between −50 and 10 mm, and are consistent.In addition, the number of DS (Distributed Scatterer) points was much higher than the number of PS points; the main reason is that the study area is located in a suburban area where some grasslands and low vegetation reduce the number of PS points.However, the trend of vertical deformation distribution is consistent for both time-series InSAR monitoring.

Figure 4 .
Figure 4. Schematic diagram of ground control points.

Figure 7 .
Figure 7. SBAS-InSAR monitoring results of time series deformation in coastal areas.

Figure 8 .
Figure 8. Distribution of vertical deformation variables on the surface monitored by SBAS-InSAR.

Figure 9 .
Figure 9.Comparison of the time-series deformation of characteristic points in the eastern and western parts of the airport.
aquifers, from top to bottom: submerged layer A0 (2∼25 m thick), second pressurized aquifer (20∼30 m thick), third pressurized aquifer (20∼30 m thick), fourth pressurized aquifer (30∼50 m thick), and fifth pressurized aquifer (10∼80 m thick).The A0 and A1 aquifers are poorly water-rich and easily contaminated by shallow burial; thus, A2∼A5 have become the main groundwater extraction layers in the coastal area of Shanghai.From 1990 to 2000, groundwater extraction in Shanghai's pressurized aquifers increased continuously, and the extraction caused changes in the groundwater flow field.Groundwater extraction monitoring based on groundwater level indicators revealed a positive correlation between groundwater extraction

Figure 10 .
Figure 10.Cumulative vertical deformation map of critical infrastructure in the study area.(a) group of buildings under construction, (b) Disneyland resort, (c) airport cargo loading area, (d) Sanjia port cargo center, (e) west side of runway 5 at PDIA.
. Since 2002, groundwater levels in each aquifer in the near-coastal area have increased substantially.The water level of the A1 aquifer increased from −8.64 m to −1.0 m, the water level of the A2 aquifer increased from −8.42 m to −0.86 m, the water level of the A3 aquifer increased from −8.66 m to −0.11 m, the water level of the A4 aquifer increased from −24.14 m to −1.38 m, and the water level of the A5 aquifer increased from −22.7 m to −3.7 m. Figure 11 shows the groundwater level trend in each aquifer from March 2016 to October 2018.

Figure 11 .
Figure 11.Groundwater level variation from March 2016 to October 2018.

Figure 12 .
Figure12.LOS directional deformation rate of PS points in the runway area.

Figure 13 .
Figure 13.Cumulative vertical displacement of PS points in the runway area.
Time Direction Orbit Wavelength Angle of Incidence Imaging mode Polarization Mode August 2016∼June 2018 Ascending Path: 171 Frame: 96 5.6 cm/C Band 39.27°IW VV

Table 2 .
Ground control point processing results.