Deformation Detection and Attribution Analysis of Urban Areas near Dianchi Lake in Kunming Using the Time-Series InSAR Technique

Abstract

The main city of Kunming is located on the north bank of Dianchi Lake. The complex geological environment, large-scale construction, and expansion of the city in recent years have caused uneven land surface subsidence and threatened public safety. In this study, Sentinel-1 ascending and descending orbit datasets were collected for the period of February 2018 to May 2021. The characteristics of surface displacement in the Kunming downtown area were monitored using the time-series interferometric synthetic aperture radar (InSAR) technique, and attribution analysis was performed. It was found that areas with more severe surface settlement were concentrated in the International Exhibition Center area and the large, newly built communities near Dianchi Lake and the Xiaobanqiao Region. The multifactor attribution analysis results demonstrated that the subsidence areas are concentrated in urban villages and engineered, construction-intensive areas in the lakeside sedimentary layer area, with the maximum displacement rate reaching −23.12 mm/a in the line-of-sight direction of the Sentinel-1 ascending dataset. The reliability of the InSAR results was cross-validated with ascending and descending results. This study provides a scientific reference for urban development planning and potential geological disaster detection in Kunming.

1. Introduction

Surface subsidence is a type of geological disaster caused by a slow decline in the ground surface due to the action of natural or human factors. Uneven urban land subsidence severely threatens the safety of buildings and infrastructure, and affects the health and sustainable development of cities. Kunming is one of the cities in western China exhibiting relatively obvious surface subsidence. Historical leveling results indicate that the subsidence reached −227.5 mm from 1985 to 1998. The mean subsidence rate was approximately −17 mm/a, and the maximum subsidence rate reached −30 mm/a in the area surrounding Xiaobanqiao [1]. In recent years, large-scale subway line construction and intensive real estate development in the deposition area near Dianchi Lake has led surface settlement to become increasingly serious, causing cracking damage to houses, road surface damage and other damage phenomena [2,3]. Therefore, monitoring and analyzing the spatiotemporal variation in the land surface subsidence activity in the main urban area of Kunming on the north bank of Dianchi Lake is essential for disaster prevention and mitigation and for healthy urban development [4].

Compared to the traditional leveling method, InSAR technology can measure surface deformation throughout the day, under all weather conditions, over a large range and with a high precision. InSAR technology has gradually become one of the main technical means to monitor the health of urban infrastructure; it has been widely used to monitor surface deformation worldwide [5,6]. This technique can effectively identify and monitor deformation signals indicating the endangerment of urban safety, such as subsidence funnels attributed to groundwater exploitation [7,8,9], uneven deformation caused by fault activity [10,11,12], and surface subsidence due to infrastructure construction [13,14,15].

Previous studies have analyzed the surface displacement in Kunming. Zhu et al. retrieved the surface displacement of Kunming from 2007 to 2016 using Advanced Land-Observing Satellite-1 (ALOS-1) data from 2007 to 2011, COSMO-Skymed data from 2011 to 2016 and Sentinel-1A data from 2015 to 2016. Two serious subsidence areas have occurred in Kunming city, concentrated in Guandu District and Xishan District. The cumulative displacements reached −210 and −98 mm, respectively, during the period of 2007–2016, and the maximum deformation rate was −35 mm/a [16]. Shao et al. used Sentinel-1A data from 2014 to 2016 to obtain land subsidence time series for Kunming city. The results revealed that there were four subsidence areas in Kunming city, including Xiaobanqiao, Dongju Xincun-Yulong Village, Dianchi International Exhibition Center and Zongshuying area. The area with the most serious surface subsidence was Xiaobanqiao, where the deformation rate reached −30 mm/a [17]. Guo et al. employed the Small Baseline Subset InSAR (SBAS-InSAR) technique to monitor the surface deformation of the north shore of Dianchi Lake in Kunming using Sentinel-1A data from 2018 to 2019. The results showed that the mean displacement rate of the north shore of Dianchi Lake could reach −30 mm/a [18].

To study land subsidence in the main urban area of Kunming in recent years, in this paper, two Sentinel-1 datasets from ascending and descending orbits covering Kunming from February 2018 to May 2021 were collected. The spatiotemporal evolution of large-scale surface subsidence of the north shore of Dianchi Lake in Kunming was obtained by using time-series InSAR technology. The reliability of our InSAR results was cross-validated by comparing the results from both ascending and descending orbits. The possible reasons for subsidence were also discussed.

2. Study Area and Data

2.1. Study Area

Kunming, the capital of Yunan, is located in the middle of the Yunnan-Guizhou Plateau. Kunming exhibits an average altitude of approximately 1891 m and is surrounded by mountains on three sides. It is a typical plateau basin area. The topography can be divided into three grades: mountains and hilly forests, lakeside impact plains, and Dianchi Lake waters. The majority of built-up areas in Kunming are located on the north shore of Dianchi Lake, which belongs to the lakeside alluvial plain. The soil structure is dominated by silty and soft clay [18,19]. A general overview of the study area is shown in Figure 1.

2.2. Data

In this paper, 99 images from the Sentinel-1 ascending orbit and 91 images from the descending orbit covering the Kunming area from February 2018 to May 2021 were selected as experimental datasets, as shown in

Table 1. Synthetic aperture radar (SAR) image-specific data parameters.

No.	Satellite	Orbit Direction	Azimuth Angle (°)	Incidence Angle (°)	Number of SAR Images	Data Period
1	Sentinel-1A	ascending	347.57	34.16	99	12/02/2018–27/05/2021
2	Sentinel-1A	descending	192.40	34.25	91	22/03/2018–17/05/2021

. The spatial resolution of the Sentinel-1 images is approximately 5 m × 20 m (range × azimuth direction). During data processing, the Shuttle Radar Topography Mission (SRTM) digital elevation model (DEM) with a resolution of 30 m provided by the National Aeronautics and Space Administration (NASA) was used to remove the terrain phase. Precision orbit data provided by the European Space Agency (ESA) were used for orbit correction to improve the registration and baseline estimation accuracy [20].

3. Materials and Methods

3.1. PS-InSAR Principle

The traditional differential InSAR (D-InSAR) method is greatly affected by various phase noises, including decorrelation and atmospheric disturbance. The PS-InSAR technique can overcome these problems by selecting point-like targets (also referred to as persistent scatterers) that can maintain high coherence over a long period [19,20]. First, a master image is selected; the other images are then used as the slave images to generate differential interferograms. Next, the persistent scatterers are selected according to the stability of the phase information in the time series. Following spatiotemporal filtering, phase unwrapping, phase noise estimation, etc., the displacements are finally calculated [21]. The processing flow of PS-InSAR technology is shown in Figure 2. The calculation of time-series deformation on persistent scatterers greatly improves the accuracy of InSAR results.

First, one image was selected as the master image by optimizing the spatiotemporal baseline and the Doppler centroid difference distribution. Second, other slave images were co-registered with the master image to generate n − 1 interferograms [22]. The phase at a pixel in the j^th interferogram generated (1 ≤ j ≤ n − 1) can be expressed as:

$$\phi ={\phi }_{atm}+{\phi }_{topo}+{\phi }_{flat}+{\phi }_{defo}+{\phi }_{noise}$$

(1)

In Equation (1) above, the interference phase φ comprises multiple components, including the atmospheric noise phase φ_atm, the topographic relief phase φ_topo, the flat earth phase φ_flat, the phase change caused by two observation targets moving along the line-of-sight (LOS) φ_defo, and the interference phase attributed to various noise components φ_noise. Then, PS candidates were extracted by setting a reasonable amplitude deviation threshold. An adjacent PS point phase difference model was established to obtain surface deformation information. Nonlinear deformation and the atmospheric delay phase were removed through high-pass filtering in the time domain and low-pass filtering in the space domain [23,24,25]. Finally, the time-series displacement of the study area was obtained.

3.2. Data Processing

In this study, the PS-InSAR method in StaMPS software was used to process the Sentinel-1 radar images from 2018 to 2021. Due to the strict orbit control technique used in Sentinel-1, the effect of spatial decorrelation is small. Therefore, we selected the image at the middle time in the dataset as the master image. The longest spatial baselines were 194.5 m and 169.2 m in all interferograms generated by ascending and descending images, respectively. The spatial baseline distribution is shown in Figure 3. We used StaMPS-v4.1 software for data processing. The processing flow included PS point selection, phase unwrapping, estimation and removal of phase noise errors [26]. The surface displacements on the north bank of Dianchi Lake were finally extracted.

4. Results and Verification

4.1. Spatial Distribution of Surface Subsidence in the Main Urban Area of Kunming on the North Shore of Dianchi Lake

In this study, PS-InSAR technology was used to extract surface displacements from Sentinel-1 radar images covering Kunming. The annual mean surface deformation rates along the LOS direction of both ascending and descending Sentinel-1 datasets were obtained and are shown in Figure 4. Red (negative values) indicates that the surface moves away from the satellite, while blue (positive values) indicates that the surface moves toward the satellite.

As shown in Figure 4, obvious deformation has occurred in Xishan District and Guandu District on the north shore of Dianchi Lake in Kunming city, with the maximum deformation rate reaching −66.98 mm from 2018 to 2021. These areas are distributed near the Dianchi International Exhibition Center area (S₁ in Figure 4) and Xiaobanqiao in Guandu District (S₂ in Figure 4), respectively. The deformation areas include the Dianchi International Exhibition Center, Yejia village to the north of the International Exhibition Center, an urban village near Erji Road in the east, and two communities named Yongjing Bay and Shanhaihui in the south and west. The maximum deformation rate has reached −20.1 mm/a. The deformation rate in the S₂ settlement area is approximately −7 mm/a, mainly concentrated in the five communities around Xiaobanqiao and Vanke Community. The maximum deformation rate reached −23.12 mm/a.

4.2. Cross-Validation of the Results of the Ascending and Descending Orbit Datasets

Since ground leveling data and Global Navigation Satellite System (GNSS) measurements were unavailable in the study area, this paper analyzed the reliability of the results via cross-validation of the Sentinel-1 ascending and descending orbit datasets [6]. Surface deformation in urban areas is mainly caused by vertical subsidence. No obvious horizontal deformation has been reported in the main urban area of Kunming [27]. Therefore, it can be considered that the deformation along the LOS direction is caused by vertical surface subsidence. Here, we use incidence angle information to convert the LOS displacements of the ascending and descending datasets into corresponding vertical displacements, as shown in Figure 5.

Figure 5 shows the annual mean deformation rate maps plotted with the vertical deformation rate results of the ascending and descending datasets. The land surface settlement area in the study area remained constant. The maximum vertical deformation rates are approximately −18.7 mm/a and −18.3 mm/a for the ascending and descending datasets, respectively. To better compare the results of the ascending and descending datasets, we plotted probability density histograms of the results from the two datasets in Figure 6. A total of 168,516 and 247,287 coherent points were extracted in the ascending and descending datasets, respectively. The provability distributions of the ascending and descending results are consistent. The vertical deformation rate of most points is near zero; that is, it remains stable. The standard deviation (STD) is 0.36 for the ascending dataset and 0.54 for the descending dataset. The results in descending orbit are noisier.

In this study, profiles of the ascending and descending results are generated along the profile line P-P’ in Figure 5 and are compared in Figure 7. This profile crosses the S₁ and S₂ areas with the most serious deformation on the north shore of Dianchi Lake in Kunming. The figure shows that the vertical deformation results of the two datasets suitably agreed, with a Pearson correlation coefficient (R²) of approximately 0.98 and a root mean square error (RMSE) of approximately 1.26 mm/a [15]. The results indicate that in the absence of ground leveling data and GNSS measurement data, InSAR results could be cross-validated with results from different orbits or different satellites.

4.3. InSAR Experimental Results and Field Cross-Validation

To verify the InSAR results, we conducted a field investigation of the deformed region. Figure 8 shows in situ photographs and the cumulative time-series deformation at corresponding positions. Points b and c in Figure 8a are located in Yejia village, which is sandwiched between two rivers leading to Dianchi Lake. Obvious tensile cracks have occurred in part of the ground and building walls, as shown in Figure 8b,c. Point d in Figure 8a is located in a newly built community along Dianchi Lake. While the foundation of the building itself has remained stable, the area has exhibited settlement. Obvious tensile cracks have occurred at the bottom of the building edge in Figure 8d. The cumulative displacements in the vertical direction of points b, c and d from February 2018 to May 2021 are shown in Figure 8e. The displacement at these three points shows a linear deformation trend.

In addition, we selected six characteristic points A₁–A₆ in regions S₁ and S₂, and analyzed the time-series cumulative displacements in the deformed regions. All these points are marked with black triangles in Figure 9. Points A₁ to A₃ are located around the Dianchi International Exhibition Center. Among them, A₁ and A₂ are two newly built communities located on the peninsula. The construction of large-scale communities on lakeside sediments has greatly increased ground loads and caused surface subsidence. A₃ is an urban village named Yejia Village, which is sandwiched between two rivers leading to Dianchi Lake. Its subsidence is caused by geological factors and the quality issues related to the self-built houses. Compared with the controllable deformation of new communities A₁ and A₂, the deformation signal of Yejia Village is more dangerous.

A₃–A₆ in region S₂ surrounds the large community, Vanke Community. The area is located in the vicinity of Xiaobanqiao, which has historically been subject to severe subsidence. Figure 10 shows the time-series cumulative vertical displacements at points A₁–A₆ from 2018 to 2021. The vertical annual mean deformation rate is approximately −13 mm/a for the points of A₁–A₃. The maximum cumulative deformation reached −45.36 mm over three years. The vertical annual mean deformation rate for A₄–A₆ near the Vanke Community is approximately −12.2 mm/a, with a maximum cumulative displacement of −42.72 mm.

From the time-series results, we find that, overall, the displacements on these points show a linear trend. The displacement on point A₁ exhibits some jitter, probably due to the atmospheric disturbances that are not completely removed. The displacements of points A₄ and A₅ show a gradual slowing trend. Points A₄ and A₅ deformed by approximately 20 mm from February 2018 to May 2019; however, after May 2019, these points took more than 20 months to settle by 20 mm. This is consistent with the deformation characteristics of soft ground consolidation under load after the construction of a large number of high-rise buildings.

5. Analysis and Discussion

Kunming is located on the north bank of Dianchi. It is a typical lake sedimentary basin belonging to the ancient Dianchi sedimentary area [28]. Tertiary semi-diagenetic rocks and Quaternary loose strata are widely distributed and thick. They are formed by the accumulation of lakes and swamps and alluvial rivers. Soft soil, expansive soil and other special and unfavorable soil layers are common. Dianchi is a continuously degraded lake. The changes in the shoreline of Dianchi lake in different periods can be clearly seen in Figure 11a [16]. The shore of Dianchi Lake is mainly dominated by sedimentary layers, which consist of a large amount of soft soil and saturated powder and gravel layers interspersed with loose sediments such as silt. These sediments formed the strata in a short period of time, with low consolidation and high water content. With the continuous development of the city, the engineering construction around Dianchi has gradually expanded. The construction of buildings and roads has led to the ground settlement phenomenon as the underground soil is pressurized by the buildings, causing the water in the soft soil, silt and other sediments to be squeezed out and slowly consolidated under the action of underground pressure and temperature. These events indicate that geological conditions are the main factor of surface subsidence, and human activities and engineering construction are the triggering factors.

5.1. Analysis of Land Surface Settlement in the Engineering Construction Area

Extensive construction in the lakeside sedimentary layer area has led to surface subsidence, threatening urban safety. To further study the relationship between subsidence and engineering construction, this paper focuses on the effect of engineering construction on surface deformation in the S₁ and S₂ study areas. Figure 12 shows the vertical deformation rate plot of the settlement area based on the Sentinel-1 dataset. Figure 13 shows the change in engineering construction in the study area from 2016 to 2021.

5.1.1. Impact of Subway Construction

Subway construction is an important symbol of urban economic development and mutually benefits city populations and logistics [29]. The construction and operation of urban subway tunnels causes land surface subsidence. The surface subsidence affects the surrounding buildings and underground pipelines to different degrees. Surface settlement occurs in urban subway tunnels due to excavation surface settlement, erosion settlement, consolidation settlement of the soil, etc. [30].

Studies are conducted on the related settlement deformation between urban infrastructure health and metro tunnel construction to prevent surface subsidence and other disasters. Based on the current operation of the Kunming metro and metro under construction, Figure 14a shows the distribution of the Kunming subway superimposed on the regional deformation rate map. As can be seen in Figure 14b, there is significant surface settlement in the middle section of the extension of Metro Line 2, which is under construction (Yejia village–Liujia interval). The settlement area shows linear characteristics; the linear settlement area is highly consistent with the driving route of the extension line of Metro Line 2. The deformation rate in the settlement area is approximately −9.5 mm/a. In this paper, three points along the settlement area of the subway are selected in the middle and at both ends for the analysis of the accumulated deformation. Figure 14c shows the accumulated vertical deformation at the positions of points P₁, P₂ and P₃. The maximum accumulated displacement reaches −40.98 mm.

It is concluded that the consolidation degree of underground soil is low. Surface settlement is caused by inward movement of soil during excavation. In the case of complex geological conditions (e.g., the cascade terrain) even ground subsidence may be induced [31]. Therefore, continuous monitoring is required during the construction cycle and after completion [6].

5.1.2. Impact of the Development Complex

The monitoring results in Figure 12a show that areas C3 and C4 are located in the west and south of the Dianchi International Exhibition Center, respectively, near Dianchi Lake. The vertical deformation rates in the two areas are approximately −13 mm/a and −7 mm/a. Figure 13a shows that residential building construction in areas C3 and C4 began after 2016 and was completed in 2018. A check of the information shows that the completion dates of the district (Shanhaihui and Yongjing Bay) were 31 December 2017, and 31 December 2018, respectively. The central coherence points of the C4 area are sparse due to the incoherence phenomenon caused by construction in 2018. According to the monitoring data in this paper, the settlement of the two plots began after the completion of construction. It is inferred that the sediment formed the stratum in a short time, with a low degree of consolidation and large water content. Building construction leads to an increase in surface load, and strata consolidation causes the surface to settle.

The C5 area is located in the eastern part of the Dianchi International Exhibition Center. The settlement zone in this region is relatively scattered, and the vertical deformation rate is approximately −9.5 mm/a. The buildings in the area were built before 2016, and their density is high. Statistics show that there is some correlation between surface deformation and building density. If other metrics are ignored and building density is assumed to be the only criterion that represents building loads [16], then it can be inferred that high building density and thus excessive building loads eventually lead to surface settlement.

From the monitoring results in Figure 12b, it can be seen that three main areas in the S₂ study area experienced settlement. The main subsidence area is located in the C6 region. The main vertical deformation rate is approximately −11 mm/a, and the deformation reaches −23 mm/a at the maximum. The area is mainly used for residential construction (consisting of five subdivisions), as shown in the subdivision distribution map in Figure 15. As seen in Figure 13b, the communities were completed by 2016, except for the Vanke Community, which was still under construction. The monitoring data in this paper were continued after its completion. These data show that settlement started to occur after the buildings were completed. It is possible that the ground settlement occurred due to the increased surface load caused by their construction. The rate of surface subsidence in the region decreases significantly from 2007 to 2021, as shown in

Table 2. Surface subsidence rate in the S₂ study area of Kunming during the different periods (mm/a).

Settlement Center	Interval/Year
	2007–2008	2008–2010	2016–2017	2018–2021
S2 study area	−21.6	−20.2	−20.6	−10.81

below. The deformation rate of the soil consolidation effect slows down with time. Some of the missing correlation points in the lower right corner of the C6 area are due to the construction of the sixth phase of Vanke Community. It is understood that the sixth phase of Vanke Community is under construction from 2018 until 2020. This construction occurs within the monitoring time frame of this paper; hence, there is a loss of coherence.

5.2. Analysis of Land Surface Settlement in Urban Villages

The main reason for surface subsidence in urban villages is the poor foundation of the self-built houses themselves. Excessive load leads to ground settlement, triggering house tilting, wall cracking, and even the collapse of self-built houses. For example, 29 people were killed and 28 injured due to the collapse of rural self-built houses in Xiangfen County, Shanxi Province, on 29 August 2020. On 19 June 2021, a self-built house collapsed in Luyang town, Rucheng County, Chenzhou city, Hunan Province, killing five people and injuring seven others. On 29 April 2022, a self-built house collapsed in Changsha, Hunan Province, killing 53 people. From the monitoring results in Figure 12, it can be seen that C1, C2, C7, and C8 all have large surface settlements. The main deformation rates are approximately −8 mm/a, −8.5 mm/a, −6 mm/a, and −6.6 mm/a, respectively. After fieldwork, it was found that all four areas comprised urban villages. Figure 16 is a photo of the tilting site of the self-built house in the urban village. The red lines and arrows in the picture mark the direction of the tilted houses, and traces of tilting can be clearly seen.

The surface settlement in these four areas is mainly caused by foundation issues common to self-built houses in urban villages. These self-built houses generally have problems such as dense construction, poor foundations, and additional floors without permission. High-intensity load on the ground with soft soil will lead to accelerated surface consolidation effect and produce significant surface settlement. In addition, the deformation caused by the quality problems of the buildings themselves will also be detected in the time-series InSAR results. Due to the insufficient resolution of Sentinel-1 imagery, it is necessary to collect high-resolution SAR data and other auxiliary data to distinguish the type and cause of deformation

5.3. Building Deformation Analysis Considering Thermal Expansion and Cold Contraction

From the results in Figure 12a, it can be seen that the right side of the top of the Dianchi International Exhibition Center itself is deforming. The Dianchi International Exhibition Center was completed and put into use in 2015. This paper monitored data from February 2018 to May 2021. This shows that the Dianchi International Exhibition Center began to deform slowly after it was put into use, and the deformation rate was approximately −5 mm/a. Field inspection and data analysis, as shown in Figure 17a,b, indicates that the deformation of the top right side of the Dianchi International Exhibition Center is caused by the structure of the building itself. The top of the building is a steel structure, which is subject to seasonal changes. Figure 17c shows the cumulative displacement variation at the top of the Dianchi International Exhibition Center. The upper two curves are the cumulative displacements of the coherent points located at the left side of the top of the Dianchi International Exhibition Center, and the lower two curves are the cumulative displacements of the coherent points located at the right side.

The region shows a cyclical variation, as seen in the figure. The Dianchi International Exhibition Center is semicircular in shape; the top of the left side faces the ascending satellite orbit and the top of the right side has its back to the ascending satellite orbit, a situation that results in opposite periodic changes. To study the causes of the cyclical variation, this paper obtained daily average temperature data from February 2018 to May 2021 in Kunming city through Kunming meteorological stations (https://rp5.ru/world_weather (accessed on 9 March 2022)). The daily average temperature data were also superimposed on the cumulative displacement of the top feature points of the Dianchi International Exhibition Center. The results show that displacement is due to the cyclic variation of the structure itself, which is caused by thermal expansion and contraction due to temperature changes.

5.4. Attribution Analysis of the Influencing Factors of Land Subsidence

Based on the influence factors of surface displacement introduced above, a null hypothesis experiment can be performed for each influence factor H_i. H_i0: the factor has no influence on surface settlement. H_i1: the factor influences the surface settlement. Analysis of n coherent point targets for surface settlement in the study area is as follows:

$$\left[\begin{array}{ccc}{P}_{11}& \cdots & P_{14}\\ ⋮& \ddots & ⋮\\ P_{n4}& \cdots & P_{n4}\end{array}\right]\left[\begin{array}{c}{H}_1\\ ⋮\\ H_4\end{array}\right]+\left[\begin{array}{c}{\epsilon }_1\\ ⋮\\ {\epsilon }_4\end{array}\right]=X$$

(2)

In Equation (2) above, when H_i0 cannot be rejected, P_ni = 0. Conversely, when H_i1 cannot be rejected, P_ni = 1. Moreover, ε_i denotes the other influencing factors not analyzed, and Y is the real settlement at the n target points.

For any coherent point target in the study area, the amount of subsidence is v_i. First, according to the distance L between the point target and the subway line, the relationship between the settlement of point target v_i and distance L is compared point by point. When L < 100 m and v_i < −5 mm/a, P_n1 = 1; otherwise, the value is 0. If the point target is in the building development area, when the point target is in the building complex development area, P_n2 = 1; otherwise, the value is 0. If the point target is in the urban village area, when the point target is in the urban village area, P_n3 = 1; otherwise, the value is 0. Finally, If the trend of point target deformation has seasonal periodicity, P_n4 = 1; otherwise, the value is 0. Therefore, whether the surface subsidence is influenced by single or multiple influencing factors can be analyzed based on the above information [32].

In Figure 18, the green points represent the impact on surface deformation due to the construction of the subway, accounting for 5.5% of the PS points. The blue dots represent the surface deformation problem due to the development of building clusters, accounting for 29% of the PS points. The red points represent surface deformation due to the unsound foundations of self-built houses in urban villages, accounting for 53.5% of the PS points. The orange point represents the surface settlement caused by the parallel action of two driving factors. The yellow point represents the Dianchi International Exhibition Center due to its periodic deformation caused by the structure of the building itself, accounting for 12% of the PS points. The surface settlement caused by the unstable foundation of the self-built house in the urban village is the dominant factor.

6. Conclusions

A healthy infrastructure is the foundation for city operation and economic development. Both natural and human-made factors can lead to the occurrence of geological hazards such as ground subsidence, which can further affect the health of the infrastructure. In this paper, PS-InSAR technology was used to process 99 ascending and 91 descending Sentinel-1 images covering Kunming on the north bank of Dianchi Lake from February 2018 to May 2021. The retrieved displacements were validated by field investigation and cross-comparison of results from different orbits. The reliability of the results was verified. Two deformation areas located near Dianchi and Xiaobanqiao were selected for our study and analysis. The main factors causing surface subsidence in different areas have also been discussed.

The results show that the deformation areas are mainly concentrated around the Dianchi International Exhibition Center and Xiaobanqiao in the vicinity of large new communities and urban villages. The maximum subsidence rate is approximately −18 mm/a for both the ascending and descending datasets. We further conducted an attribution analysis and found that approximately 5.5% of the settlement was related to the subway. Approximately 29% of the subsidence targets were related to the newly built communities. Approximately 53.5% of the displacement points were located in the urban village area. Approximately 12% of the displacement points were caused by the thermal expansion and contraction of the building structure itself.

In summary, the surface deformation of urban villages near Dianchi Lake needs to be addressed to avoid greater disasters. PS-InSAR technology provides effective monitoring of regional subsidence and urban infrastructure health in Kunming. This work provides a scientific basis for the construction and development of Kunming. Collecting more geological information and ground measurement data will help to better interpret the surface subsidence in Kunming.