Subsidence Monitoring and Mechanism Analysis of Anju Airport in Suining Based on InSAR and Numerical Simulation

Abstract

The mountainous area of southwest China is characterized by significant topography and complex geological conditions, which pose great challenges to the airport’s site selection, construction, and safe operation. Suining Anju Airport, one of the key projects under construction in southwest China, is essential in alleviating and dredging the air passenger flow in Sichuan Province. Because the overlying quaternary strata’s physical and mechanical properties, thickness, and distribution range are fairly different in the longitudinal and transverse directions, the Anju Airport’s foundation in the hilly area has typical inhomogeneity. Large-scale excavation and filling pose a challenge to the ground stability of the airport. To comprehensively monitor Anju Airport’s uneven ground subsidence during the construction period, this paper selected SAR image data collected by the Sentinel-1A satellite from May 2018 to June 2021 to extract time-series ground subsidence measurements based on the SBAS-InSAR method. Furthermore, based on the simulation of roadbed filling in the airport’s parallel slide fill area, the dynamic evolution analysis of soil stress field and internal subsidence caused by roadbed filling activities was carried out to further reveal the occurrence mechanism of ground subsidence. The monitoring results show that the subsidence centers of Anju Airport are mainly distributed in the filling areas, and the average annual subsidence is −20~−75 mm/yr from May 2018 to June 2021. Comparative analysis with in situ data indicates that the RMSE of InSAR monitoring results was ±6.12 mm. The numerical simulation shows that the subsidence of the airport parallel slide is mainly caused by a load of subgrade filling body and the compression of its weight. The results of this study can provide reference methodology and data support for the construction and future safe operation of Suining Anju Airport.

1. Introduction

According to relevant data, mountainous areas in China account for more than 68% of the country’s land area, of which 48.8% are located in the southwest [1,2]. Because of the mountainous mountains, ravines, and complex terrain in southwest China, the land utilization rate in mountainous areas is shallow, and the infrastructure construction is complicated. In order to develop the urban economy in southwest China, it is urgent to put the infrastructure into construction, especially the site selection planning, reconstruction, and later operation of the airport. As a typical ground infrastructure, the airport carries a fast traffic function, and its subsidence has been a concern by the masses. Especially in the early stage of airport operation, the uneven subsidence of the airport foundation poses a huge security threat to the take-off, landing, and taxiing process of aircraft. Therefore, ground subsidence monitoring and risk control are of great significance to the safe operation of airports.

InSAR technology has been widely used in landslide monitoring [3,4,5], urban ground monitoring [6,7,8,9], and seismic subsidence [10,11,12] due to its all-weather, high-precision, and large-range monitoring advantages. However, the traditional differential InSAR (DInSAR) has limitations in its application to large-scale monitoring due to shortcomings such as spatio-temporal incoherence, atmospheric delay, and high phase noise [13]. The proposed time-series InSAR technology dramatically improves the monitoring accuracy of InSAR and can provide an effective detection method for the spatio-temporal evolution of regional land subsidence [14,15,16,17]. However, due to the dense vegetation and complex topography, airport construction and ground subsidence monitoring are faced with many difficulties, especially the core problems of high-fill foundation engineering, namely differential subsidence, and slope stability. Wu and Yang [18] obtained the 20-year subsidence time series of Hong Kong airport based on improved multi-temporal InSAR technology, established a subsidence model based on InSAR measurement data, and revealed the spatial-temporal variation law of land subsidence. Chen et al. [19] obtained the ground subsidence rate and sequential subsidence of Zhongchuan International Airport on the Loess Plateau based on the time-series InSAR technology. They evaluated the monitoring results by using the internal and external inspection methods. Zhang et al. [20] obtained the ground subsidence of Ankang Airport in the early stage of operation based on the time-series synthetic aperture radar interferometry technology of elevation correction and analyzed and discussed the subsidence mechanism of the airport and expansive soil slope in the high-fill area.

Kudryavtsev et al. [21] conducted a numerical simulation of the subgrade structure of permafrost and soft soil in the independent railway construction area of the Far East Federal Zone to obtain the subsidence information of the area. On this basis, he proposed the low cage embankment structure to reduce the vertical and horizontal subsidence of the embankment. Li et al. [22] used FLAC3D to conduct a numerical simulation of the construction process of a high-fill foundation in Jiuzhai Huanglong Airport and analyzed and evaluated the subsidence of a high-fill foundation. Based on the monitoring data of ground subsidence, Zhu et al. [1] made an in-depth analysis of the causes of uneven subsidence of lvliang Airport high-fill foundation from the aspects of numerical simulation of soil filling height, soil filling rate, and foundation compaction.

The original data and analysis methods used in the above research are relatively single, and the obtained research results have certain limitations. The results obtained by using the time-series InSAR technology combined with the numerical simulation of subgrade high fill are diverse and can monitor the complete ground and underground subsidence and help to analyze the causes of differential subsidence of airport foundations deeply. Based on 89 Sentinel-1A high-resolution SAR images from May 2018 to June 2021, this paper uses SBAS-InSAR technology to obtain the spatial distribution of ground subsidence at Anju Airport and analyze the temporal and spatial ground subsidence characteristics of the airport. The accuracy of SBAS-InSAR observation points is verified by the existing ground measured data. On this basis, FLAC 3D is used to simulate the backfilling process of the airport high-fill foundation from August 2018 to March 2019. The internal ground subsidence caused by the backfilling process is analyzed through the subsidence field obtained, which is convenient to provide data support for the subsequent monitoring of airport subsidence and provide a reference for the monitoring of similar high-fill airports’ subsidence in southwest China.

2. Study Area and Material

2.1. Study Area

Anju Airport in Sichuan Province is located at 30°22′40.40″~30°24′58.96″ north latitude and 105°24′44.21″~105°25′58.08″ east longitude. It is a typical high-fill airport in a mountainous area, with a linear distance of 17 Km from the center of Suining, as shown in Figure 1a. According to the Statistics Journal of Suining’s National Economic and Social Development in 2016 [23], Suining received 39.3176 million domestic tourists in the whole year, an increase of 26.8% year-on-year, and the number of tourists showed an upward trend. However, with the continuous expansion of the city scale, the area of the former Nanba Airport in Suining has a severe conflict with the urban development plan, and the perimeter of the airport runway passes through the central city, bringing significant security risks to urban residents. Suining municipal government is a famous tourist city in Sichuan Province and the secondary transportation hub of the province with a massive tourist flow. To this end, the Suining city government decided to develop air traffic infrastructure on the basis of the “8 highway lines in 1 ring network” and “7 railway lines in 3 directions”. The development of the transportation, industry, and culture belt between Chengdu and Chongqing will be promoted through comprehensive transportation system construction in Suining, including river, road, railway, and air transport infrastructure.

Anju Airport began construction of the pilot section in January 2014, and the main part of the project started construction on 18 January 2018, with a construction period of three years. The length of the airport runway is 2200 m, 45 m wide, and the size of the parallel slide is 2200 m, 18 m wide, covering a total area of 14.4 Km². Up to now, the project is in the phase of ancillary facilities construction. Figure 1b,c represent the comparison before and after the airport’s construction.

The original terrain of the study area is relatively complex, consisting of medium and low hills and intermountain gullies. The specific landform features are medium and shallow hills with box-shaped valley branches, and the gullies are gently u-shaped. The original foundation soil mainly consists of sandstones of Suining formation, argillaceous siltstone, silty mudstone, quaternary residual slope deposit (Q4el + DL), tillage soil, fill soil, block stone, and silty clay. At the same time, the original gullies were all seasonal gullies without perennial gullies and rivers. Atmospheric precipitation replenishes surface water. Part of atmospheric precipitation collects in the drainage area in gullies through surface runoff. Part is temporarily stored in ponds, paddy fields, and other water storage areas and slowly discharged to downstream gullies. Figure 1d shows the water distribution in Anju airport [24]. Large-scale and high-depth filling in silty clay areas with strong swelling and contraction greatly impacts engineering geology and hydrology conditions. It will inevitably cause a lot of post-construction land subsidence [25].

2.2. Data Source

In this study, 89 Sentinel-1A SAR images were collected from May 2018 to June 2021 to monitor the subsidence characteristics of the airport ground. At the same time, STRM data with 30 m resolution was used to remove the influence of the topographic phase. The imaging mode is area-wide amplitude (IW) mode, the polarization mode is VV + VH polarization, the spatial resolution is 2 m × 14 m, and the incidence angle is 39.2372°.

Table 1. SAR datasets.

Sensor	Path	Number of Images	Date Range
Sentinel-1A	164	89	May 2018–June 2021

shows the specific data. Meanwhile, we set the time baseline threshold at 48 d and the vertical limit at 186 m. Figure 2 shows the corresponding space-time baseline.

3. Study Methods

This paper uses SBAS-InSAR technology to extract the ground subsidence sequence of Anju Airport and uses numerical simulations to obtain the underground subsidence field caused by subgrade filling. On this basis, the layered subsidence model of filling the area is established, and the mechanism of land subsidence is revealed further. Figure 3 is the technical flowchart of this paper.

3.1. SBAS-InSAR Method

SBAS-InSAR came into being as D-InSAR technology evolved. SBAS-InSAR technology is based on the interference pairs of multiple master images, and the phase analysis of highly coherent targets is carried out to obtain the time-series subsidence information. By choosing appropriate space-time difference interference for the baseline threshold, one can use linear phase change models to select the coherence of target modeling and calculations and use space-time filtering to estimate and remove the atmospheric delay and reduce the differential interferometry synthetic aperture radar technology to deal with the loss of coherent effects, as well as the elevation error and atmospheric ground subsidence time-series information at the same time. The incoherence effect, elevation, and atmospheric errors cause problems in managing ground time-series subsidence information [14,15,25,26]. The basic idea is that N + 1 SAR images are arranged in time order (t₁, t₂…t_n), and one image is selected to register the main image. We set an appropriate spatio-temporal baseline threshold to form the baseline network. Finally, the combined groups of images are interfered with differentially to generate M differential interferograms. Where M needs to meet the following conditions:

$$\frac{N+1}{2}\le M\le \frac{N(N+1)}{2}$$

(1)

Suppose differential interference processing is performed on the image at the t_i, t_j (t_i < t_j) moment of pixel (m, n) to obtain the k-amplitude differential interference phase δφ_k (m, n), and the formula is as follows [25]:

$$\delta {\phi }_k\left(m,n\right)\approx {\phi }_k^{def}(m,n)+{\phi }_k^{topo}(m,n)+{\phi }_k^{atm}(m,n)+{\phi }_k^{noise}(m,n)$$

(2)

Among them, φ_k^def (m, n) represents the subsidence phase, φ_k^topo (m, n) represents the topographic phase at pixel point (m, n), φ_k^atm (m, n) represents the atmospheric phase, and φ_k^noise (m, n) represents noise phase.

For all the obtained interferograms, the linear equations between the differential interference phase and the shape variables of image acquisition time are constructed, and the matrix form is expressed as:

$$A\phi =\delta$$

(3)

where A represents the M × N bond number matrix, $\phi$ represents the parameter matrix composed of the unknown subsidence phase values corresponding to unknown pixels (M, N) at N times, and $\delta$ represents the matrix composed of the unwrapping phase values of M interference pairs.

If all interference pairs belong to the same small baseline subset, the estimation of cumulative shape variables can be acquired with the least square method:

$$\hat{\phi }=A^{+}\delta \phi ,A^{+}={(A^{T}A)}^{-1}{A}^T$$

(4)

If A contains multiple sub-baseline sets, Formula (3) must be solved by the singular value decomposition method, and then the cumulative shape variable can be obtained [14].

3.2. Iterative Solution Based on Mohr–Coulomb Model

When simulating the impact of subgrade filling on ground subsidence, it is essential to select an appropriate constitutive model to describe the mechanical response between loose or consolidated soil particles [27]. Mohr–Coulomb’s theory points out that the reason for material yielding is that the shear stress on a section reaches the strength limit [28]. The Mohr–Coulomb model is used as the constitutive model in this study. The main idea is to determine the stress–strain state of soil elements by yield function. If the material does not yield, the soil element’s strain increment and stress increment are obtained according to the elastic law. If yield occurs, the strain increment and stress increment are solved according to the flow rule constructed by the potential function. Finally, the total stress of the node is obtained by superposition of the original stress and stress increment. When the material does not yield, we can use Hooke’s law to calculate the increment of strain variable and principal stress as follows:

$$\{\begin{array}{l}\Delta {\sigma }_1={\alpha }_1\Delta e_1^e+{\alpha }_2(\Delta e_2^e+\Delta e_3^e)\\ \Delta {\sigma }_2={\alpha }_1\Delta e_2^e+{\alpha }_2(\Delta e_1^e+\Delta e_3^e)\\ \Delta {\sigma }_3={\alpha }_1\Delta e_3^e+{\alpha }_2(\Delta e_1^e+\Delta e_2^e)\end{array}$$

(5)

where in the increment of the principal stress variable can be decomposed into:

$$\Delta e_i=\Delta e_i^e+\Delta e_i^p,i=1,2,3$$

(6)

where $\Delta e_i$ represents the principal strain increment, $\Delta e_i^e$ represents the elastic increment, and $\Delta e_i^p$ represents the plastic increment, where $\Delta e_i^p=0$. ${\alpha }_1=K+\left(4/3\right)G$, ${\alpha }_2=K-\left(2/3\right)G$. K represents the volume modulus, G represents the shear modulus and $\Delta {\sigma }_i\left(i=1,2,3\right)$ represents the stress increment.

Based on the iterative solution of the Mohr–Coulomb model, the discrete hybrid method is used to discrete the finite element region into constant strain tetrahedral elements. According to the principle of virtual work, we derive the unbalanced force and use the node motion equation to solve the l node velocity component [29]. Then the strain rate component is solved according to Gauss law. Finally, the strain rate component is substituted into the constitutive equation to calculate the stress change rate. The constitutive equation generally has the following form:

$$\widetilde{[\sigma ]}=H_{ij}({\sigma }_{ij},{\xi }_{ij},k)$$

(7)

where $\widetilde{\left[\sigma \right]}$ is the stress change rate, ${\sigma }_{ij}$ is the stress component, H_ij represents a specific functional relationship, and k is the parameter related to the load history.

By integrating the stress in the time step domain, the total stress of the node at the present stage can be obtained. Based on the principle of virtual work, the node unbalance force of the next step is calculated. Finally, the stress increment of each time step is superimposed to obtain the total stress of the node, and the subsidence of each time step is superimposed to obtain the total subsidence of the node. The schematic diagram of the FLAC iterative calculation is shown in Figure 4 [29].

4. Subsidence Monitoring and Analysis

4.1. Analysis of Ground Subsidence Monitoring Results in the Construction Period

Figure 5 shows the annual subsidence rate of Anju Airport from May 2018 to June 2021, obtained by SBAS-InSAR technology. Red represents land subsidence, and green represents land uplift. Figure 5 shows differential subsidence in the airport, and the average annual subsidence rate from May 2018 to June 2021 is −75~−25 mm/yr. According to the overall subsidence rate diagram, Anju Airport mainly has four large subsidence areas, which are the eastern cliff area of Hujiawan (HJW), the gully area of YFW (YFW), the slope area of Tianjiawan (TJW), and Longjiawan (LJW). The main subsidence rates in the Hujiawan area range from −20 mm/yr to −65 mm/yr, and the maximum subsidence rate reaches −94 mm/yr. The main subsidence rate ranges from −20 mm/yr to −75 mm/yr, and the maximum subsidence rate reaches −108 mm/yr. The main subsidence rate ranges from −20 to −50 mm/yr in Tianjiawan and from −20 to −45 mm/yr in Longjiawan. Among them, the southern end of the runway where HJW is located and the central part of the parallel slide where YFW is located have large subsidence, while the northern part of the parallel slide where TJW is located has relatively small subsidence. It can be seen from the remote sensing image that the original landform of Youfangwan contains two ponds. According to the field investigation by existing scholars, the surface of the pond after drainage has a relatively deep degree of plastic and soft plastic silty clay. In addition, the silty clay in this area has strong swelling and contraction, producing massive subsidence under water absorption and dehydration. In addition, few or no subsidence observation points were detected on the southwest side of the HJW area and the east side of the YFW area. It is speculated to be caused by the main engineering construction of the airport, frequent engineering interference, and the incoherence of subsidence variables beyond the detection range of InSAR.

4.2. Analysis of Ground Subsidence Monitoring Results in the Late Construction Period after Landfill

In this paper, the subsequent time-series monitoring of the airport after the filling is carried out to analyze and discuss the airport’s ground subsidence after filling on a large scale, and the comparison with the measured leveling data is verified. Figure 6 shows the average annual subsidence rate from September 2020 to June 2021. We can find that after the excavation, filling, and slope protection work, the subsidence observation points detected by InSAR increased significantly, and significant subsidence occurred in the filling area. The average annual subsidence rate of the study area was 74~25 mm/yr. On the one hand, the subsidence of the filling area varies with the thickness of the filling body. On the other hand, due to the different physical properties of soil in different stratum in subgrade filling, large area differential subsidence occurs in the HJW, YFW, TJW, and LJW areas under a load of upper fill. However, with the completion of site leveling and runway paving, the subsidence rate of the eastern TJW and LJW zones of Anju Airport has been flat, while the subsidence range of the YFW and the HJW zones has been reduced, but there is still a large subsidence rate of −10~−30 mm/yr. These results suggest that we should pay close attention to the stability of the airport’s high-fill slope and roadbed and strengthen the monitoring of the airport fill area to avoid slope landslides and other phenomena caused by unstable slopes which can affect the regular operation of the airport in the later period.

4.3. Ground-Level Data Validation

In order to verify the reliability of the SBAS-InSAR processing results, we selected the ground leveling data from 2020 to 2021 in this study, and the SBAS-InSAR technology was used to comprehensively evaluate and analyze the ground subsidence accuracy in the study area. Figure 7a shows the cumulative subsidence time-series results of InSAR observation points selected in the HJW, YFW, and YJW subsidence areas. The selected subsidence monitoring points are P1 and P5 in HJW, P2 and P3 in YFW, and P4 in TJW. We can see that the regions where P1–P5 are located have experienced significant subsidence, the ground slightly uplifted in the later period, the maximum accumulated subsidence is up to −70 mm, and the overall characteristics of continuous subsidence and subsidence are undulating. Figure 7b represents the on-time measured ground leveling data from September 2020 to June 2021. T07, C04, and C13 correspond to subsidence monitoring points P1, P2, and P3, respectively. It can be seen from Figure 7b that the area where HJW and YFW are located has been slowly sinking.

Figure 8 shows the comparison of leveling data with SBAS-InSAR monitoring data. Figure 8a shows that InSAR monitoring point P1 is relatively consistent with ground-level data. Figure 8b,c show that ground points C04 and C13 show a slow subsidence trend from November 2020 to February 2021, while InSAR monitoring points P2 and P3 show an accelerated subsidence trend. It may be related to the accelerated ground subsidence caused by earth-moving vehicle transportation and the rapid subsidence caused by the self-weight of the fill body during the construction period. In general, the subsidence trend of InSAR monitoring points and ground leveling data is the same, and the selected verification points have a high correlation with the measured leveling points.

The accumulated subsidence of P1, P2, and P3 and the measured horizontal accumulated subsidence on the ground are shown in Figure 9. The correlation was 0.98, 0.90, and 0.96. Additionally, the RMSEs were 4.72, 7.17, and 6.23, respectively, as shown in

Table 2. Validation of temporal InSAR monitoring results.

Date	Accumulated Subsidence (mm)
	Point
	P1	T07	P2	C04	P3	C13
2020/09/14	0.00	0.00	0.00	0.00	0.00	0.00
2020/9/26	−5.56	−5.49	−7.58	−5.73	−7.69	−7.87
2020/10/08	−11.66	−13.90	−18.51	−12.02	−18.34	−14.54
2020/10/20	−18.11	−21.69	−24.10	−18.04	−25.39	−20.83
2020/11/01	−18.44	−29.00	−25.30	−24.51	−28.40	−27.22
2020/11/13	−29.36	−34.93	−32.50	−26.00	−38.88	−31.10
2020/11/25	−38.07	−39.10	−35.75	−29.36	−40.63	−35.18
2020/12/07	−45.72	−43.14	−39.82	−32.26	−46.53	−38.06
2020/12/19	−50.37	−46.99	−44.32	−34.44	−46.71	−40.86
2020/12/31	−55.59	−50.17	−45.70	−36.74	−52.70	−43.13
2021/01/12	−53.94	−52.02	−46.43	−39.37	−52.13	−46.11
2021/01/24	−54.39	−53.68	−49.17	−40.22	−54.31	−47.50
2021/02/05	−63.76	−55.73	−50.98	−41.86	−59.37	−49.09
2021/03/01	−59.85	−58.94	−47.66	−44.65	−56.48	−53.38
2021/03/13	−62.03	−60.84	−47.77	−46.24	−60.10	−54.24
2021/03/25	−60.13	−62.23	−50.32	−46.61	−61.18	−56.14
2021/04/06	−61.55	−63.67	−48.21	−47.88	−61.07	−57.60
2021/04/18	−70.46	−65.44	−52.34	−48.87	−64.05	−58.43
2021/04/30	−62.58	−67.46	−39.32	−49.72	−51.72	−59.28
2021/05/12	−66.91	−68.47	−42.17	−50.92	−55.75	−61.02
2021/05/24	−59.00	−70.28	−36.45	−52.07	−52.01	−62.20
RMSE	±4.72 (in P1)		±7.17 (in P2)		±6.23 (in P3)
RMSE	±6.12

. An RMSE of ±6.12 mm was obtained by comparing all validation data. It shows that the subsidence rate of leveling is in good agreement with that of InSAR, and the ground subsidence information extracted by SBAS-InSAR is accurate and reliable.

5. Embankment Filling Deformation Model

5.1. Filling Areas Model

In this study, the YFW filling area was taken as an example to simulate the ground subsidence caused by the subgrade filling process. According to the contour data, it can be seen that the topographic fluctuation of the YFW region is slight, so the bedrock is regarded as the ideal plane. The model is divided into three layers to facilitate calculation and analysis: filling body, plastic silty clay, and bedrock. Then the model was meshed and imported into FLAC3D 6.0. The original formation model was 120 m long, 2 m deep, and 30 m high. According to the geological survey data, the filling area’s height is 27 m, and the slope ratio is 1:2. Therefore, the bottom length of the filling body is 120 m, the top size is 50 m, the height is 27 m, and the depth is 2 m. Figure 9a shows the generalized soil model. The first layer is the bedrock, the second layer is silty clay, and the third layer is the fill (silty mudstone). The soil physical parameters used in this paper refer to the published research results in this region as shown in

Table 3. Physical parameters of each layer.

Parameter	Value
	Bedrock	Silty Clay	Silty Mudstone
Modulus of compression, Es (Mpa)	500	4.0	45.0
Poisson’s ratio, v	0.2	0.35	0.25
Cohesion, C(Kpa)	40	36	70
Internal friction angle, $\phi$ (°)	40	15	35
Nature bulk density (KN/m3)	24.1	19.8	21.6

[28]. Figure 9b represents subsidence boundary conditions. Blue, green, and red represent initial subsidence in the X, Y, and Z directions, respectively.

Figure 9. (a) Formation information; (b) boundary condition setting (set the initial velocity of X, Y, and Z).

Figure 9. (a) Formation information; (b) boundary condition setting (set the initial velocity of X, Y, and Z).

According to the experiment, the elastic modulus can be obtained by repeatedly checking the compression modulus 2–5 times. The value of the volume modulus (K) and shear modulus (G) can be calculated according to the following formula:

$$K=\frac{E}{3\left(1-2v\right)}$$

(8)

$$G=\frac{E}{2(1+v)}$$

(9)

where E is the elastic modulus and v is Poisson’s ratio.

5.2. Subgrade Backfilling Simulation

Figure 10a is the initial stress field, and the left scale is the stress magnitude. A negative value represents compressive stress, and a positive value represents tensile stress. The initial pressure is compressive due to gravity acting on the soil and bedrock. Figure 10b shows the force in the z direction after subgrade filling. We can see that the formation stress caused by backfilling increases to about 20 times the initial stress, indicating that the original formation stress has changed significantly due to the filling activity. Considering the construction design scheme and site construction backfill, this numerical model adopts the method of hierarchical filling. It took eight months to fill the model in 27 stages in batches, each with a height of 1 m. Figure 10c shows the horizontal subsidence diagram at the end of subgrade filling. It is not difficult to find that the horizontal subsidence of the shallow subgrade presents a “barbell type” distribution, and the subsidence of the shoulder on both sides of the subgrade is significant. In contrast, the subsidence of the center of the subgrade is slight. Figure 10d is the subsidence cloud diagram of the in situ ground after foundation backfilling. It is easy to find that the vertical subsidence of the entire subgrade presents an “oval” distribution because with the deepening of subgrade backfilling depth, the upper filling body load borne by the original ground increases, and the subsidence gradually increases.

5.3. Influence of Fill Thickness

In the filling engineering involved in airport construction, the maximum filling thickness is up to 50 m, and the distribution and thickness of the filling body affect the distribution and size of land subsidence. Figure 11a shows the curve of the initially accumulated ground subsidence from the start of the main body of the building to the end of the filling project. It can be seen that with the deepening of the thickness of the filling body, the ground subsidence caused by layered filling also increases linearly. According to the analysis of the existing stratified monitoring data, part of the ground subsidence of the airport is caused by the dead weight of the filling body, and the other part is caused by the loading of the filling body accumulated in the original foundation soil. To analyze the relationship between land subsidence and fill thickness, ground monitoring points and fill thickness at their locations were numerically simulated by subgrade fills and analyzed (

Table 4. Subsidence of center point when filling height changes.

Key Point	Subsidence (m)
	Depth of Fill (m)
	0	5	10	15	20	25	27
Real point value	0	−0.081	−0.226	−0.243	−0.292	−0.368	−0.393
Point prediction value	0	−0.073	−0.145	−0.218	−0.291	−0.363	−0.392

). As can be seen from the subsidence of the center point, when the filling height changes, with the significant increase in the thickness of the filling body, the subsidence amount of the monitoring point also increases significantly. When the filling reaches 27 m, the original ground subsidence is as high as 0.392 m.

Figure 11b shows that the simulated maximum subsidence of the central node is 0.392 m, slightly smaller than the measured value of 0.393 m. The reason is that the actual measured results include rapid subsidence during construction and creep subsidence after construction. However, in general, there is a good correlation between the predicted value and the measured value of the center node during subgrade backfilling, and the correlation is as high as 0.98.

6. Discussion

During the airport’s construction, the surface changes and backfill lead to a loss of coherence in SAR images, making it challenging to extract ground information, especially in the landfill area. Therefore, the change of soil displacement field before and after subgrade backfilling is obtained by the numerical simulation method, and the original foundation subsidence diagram further analyzes the subsidence mechanism of the filled area. Subsequently, through the SBAS-InSAR technique for the subsidence of airport construction in the late information, we analyzed the initial consolidation subsidence of the airport after landfill. Given the deficiency of InSAR monitoring points in the construction area due to frequent engineering disturbances, this paper finally realized the monitoring and analysis of airport subsidence during the whole construction period by combining InSAR and numerical simulation technology. The relevant results directly reflect the accumulated subsidence from the original ground to the ground surface before and after the airport filling.

Anju Airport is a newly built airport in southwest China. Due to the short monitoring time of the temporal InSAR technology for ground subsidence, the existing temporal InSAR monitoring results are insufficient for predicting the subsidence in the study area. However, it can be seen that the southern end of the runway and the parallel taxiway are still in the sinking stage and may continue to slowly sink in the next decade, as can be seen from the monitoring data and subsidence trend of the whole time series. It suggests that the relevant departments should strengthen the subsidence monitoring of YFW and HJW in the later stage of airport operation and pay close attention to the subsidence of HJW and YFW.

7. Conclusions

Because the overlying quaternary strata’s physical and mechanical properties, thickness, and distribution range are fairly different in longitudinal and transverse directions, the Anju Airport foundation in the hilly area has typical inhomogeneity. Large-scale excavation and filling pose a challenge to the ground stability of the airport. This paper uses the SBAS-InSAR technology to extract the ground subsidence characteristic information from the 89Sentinel-1A images of views of the landing orbit, covering Anju Airport in Suining to obtain the distribution characteristics of land subsidence before and after the construction of Anju Airport. At the same time, the Mohr–Coulomb model was used to simulate the change of displacement field caused by subgrade backfilling and further analyze the internal subsidence mechanism of the filled area. The main results are as follows:

• After the completion of the main body of Anju Airport, uneven land subsidence occurred, and its subsidence center is mainly distributed in YFW and HJW fill areas. The maximum subsidence rate of YFW is −108 mm/yr, and the maximum subsidence rate of HJW is −94 mm/yr. Compared with the leveling observations of three in situ points, the precision and reliability of the accumulated ground subsidence detected by InSAR were validated. The RMSEs of the InSAR observations in three points are ±4.72, ±7.17, and ±6.23 mm, respectively. Additionally the R²s are 0.98, 0.90, and 0.96, respectively. Statistic analysis with all in situ data shows the RMSE is ±6.12 mm.
• Through numerical simulation of the backfilling of subgrade in YFW, the accumulated subsidence of the original ground with different backfilling heights is obtained. When the backfilling height reaches 15 m, the accumulated subsidence of the original foundation is 0.218 m. When filled to 27 m, the maximum accumulated subsidence of the original foundation is 0.392 m. The RMSE is ±0.03 m, and the correlation is 0.98 when comparing the numerical simulation results with the measured data. The numerical simulation results reflect the internal subsidence of soil in detail, provide an analytical basis for ground monitoring data detected by InSAR, and make up for the shortcomings of InSAR monitoring points caused by frequent engineering disturbances.
• This paper finally realized the monitoring and analysis of airport subsidence during construction by combining InSAR and numerical simulation methods. It provides technical means for studying high-fill airports’ differential ground subsidence and slope stability. Related results are essential for further monitoring, early warning, and scientific prevention and control.