Introduction

Ground subsidence, the gradual sinking or settling of the ground surface, poses a critical concern in geological/geotechnical engineering, impacting infrastructure integrity, environmental stability, and human safety. Its multifaceted causes, including excessive groundwater extraction, natural compaction of subsurface sediments, and dissolution of soluble rocks, weaken subsurface support, leading to surface deformation and settlement [23]. This subsidence phenomenon poses significant risks, damaging buildings, roads, pipelines, and other vital infrastructure, disrupting services, and posing safety threats. Moreover, it alters drainage patterns, heightening flood risks while inducing ground cracking, tilting, and differential settlement.

In urban settings, ground subsidence exacerbates hazards, increasing susceptibility to floods and structural damages in essential infrastructure like highways, bridges, levees, sea walls, and underground utility pipes [16]. Southeast Texas, commonly referred to as SETX, is very vulnerable to geological hazards such as ground subsidence, floods, and coastal erosion. This susceptibility can be attributed to the region's specific geographical location and underlying geology characteristics [19, 31]. The coastal region of SETX is geologically characterized by the convergence of three major fault zones, namely the Hockley–Conroe Fault System, the Long Point–Eureka Heights fault, and the Addicks Fault System [18]. The fault systems, described by Khan et al. [17], transverse deposits from the Holocene and Paleocene epochs. These fault systems display aseismic creep, leading to temporal changes. The act of extracting groundwater in excessive amounts and the process of sediment compaction are widely recognized as major factors contributing to subsidence [22, 32]. The lowered groundwater table, due to the groundwater pumping, increases the weight of overburden soils, resulting in a gradual ground settlement for clayey soils, which is a ‘consolidation’ mechanism [4, 26, 28, 34].

Addressing this geohazard has become imperative, given its financial burden and potential danger to urban infrastructure. Geo-engineers utilize various investigative techniques, from site characterization to soil and rock testing, coupled with groundwater monitoring, to evaluate subsidence potential and inform preventive and remedial actions [5]. In response to the significant nature of the issue, the Government of Texas and the United States Geological Survey (USGS) together formed the Harris-Galveston Subsidence District (HGSD) in 1975. This initiative was then followed by the establishment of subsidence districts in the neighboring counties. In order to effectively observe and address ground subsidence, HGSD and USGS have employed borehole extensometers and created a comprehensive network for monitoring groundwater [10]. Furthermore, in partnership with the National Geodetic Survey (NGS), the HGSD has successfully devised a leveling survey and established a Global Navigation Satellite System (GNSS) network.

Several prior studies have utilized Interferometric Synthetic Aperture Radar (InSAR) to assess ground subsidence in the SETX region, encompassing Houston and its neighboring areas. In their study, Qu et al. [24] utilized data obtained from the European Remote Sensing (ERS-1) and ERS-2 satellites to ascertain a subsidence rate of 53 mm/year throughout the time frame of 1993 to 2000. Additionally, the authors of the study utilized data gathered by Envisat and the Advance Land Observation Satellite (ALOS) to ascertain subsidence rates of up to 30 mm/year between the years 2004 and 2011. In a study undertaken by Miller and Shirzaei [19], measurements were performed using ALOS data from 2007 to 2011 and Sentinel-1A/B data from 2015 to 2017. The results with the ALOS and Sentinel-1 data demonstrated the subsidence rates of up to 49 mm/year and 34 mm/year, respectively. Furthermore, Khan et al. [17] conducted an analysis of data obtained from the Sentinel-1 satellite spanning the period from 2016 to 2020. Their study focused on quantifying the extent of subsidence, revealing measurements of up to 9 cm in magnitude.

In various global cities, the monitoring of ground subsidence through InSAR time series analysis, notably the Persistence Scatterer Interferometric Synthetic Aperture Radar (PS-InSAR) technique, has been extensively employed. However, the reliability and accuracy of InSAR measurements can be affected by multiple sources of noise and errors, including atmospheric propagation effects and temporal decorrelation [2]. The PS-InSAR method involves stacking SAR data over time, selecting a single reference acquisition date that considers temporal and spatial baselines to uphold high coherence among persistent scatterers across generated interferograms. Notably, urban areas with man-made structures tend to exhibit more persistent scatterers, ensuring phase stability and high amplitude over extended periods, thus rendering PS-InSAR highly reliable and accurate in these settings. Recent studies have applied the PS-InSAR technique to monitor and map ground subsidence in diverse urban landscapes, including Mexico City [8], Seoul, South Korea [6, 20, 21], Jakarta, Indonesia [11], Lahore, Pakistan [15], Beijing, China [1], and Karaman, Turkiye [9].

The main aim of this study is to quantitatively determine the magnitude and the rate of ground subsidence in SETX, with a particular focus on Harris County and Galveston County, as well as their adjacent territories; thus, engineers gain a comprehensive understanding of the potential consequences of flooding. In order to accomplish the objective, this research utilized the PS-InSAR technique for the spatio-temporal assessment of ground subsidence, making use of Sentinel-1 SAR satellite data obtained during descending observations from January 2020 to March 2022. Furthermore, the research compared the InSAR subsidence results with the global positioning system (GPS) data and aquifer water levels. The study emphasizes the importance of integrating InSAR, GPS, and groundwater data in the formulation of flood resilience measures by incorporating a range of measuring techniques.

Materials and methods

Study area

The study area lies at 95° 21′ 32″ W and 29° 45′ 55″ N on the Southeast of Texas (Fig. 1). This region is home to about 2.3 million people [29]. The study area covers the city of Houston, the largest city in the state of Texas and the fourth largest in the United States. Figure 1 also shows the location of GPS stations distributed in the study area and operated by HGSD. HGSD utilized a periodically measured (PAM) GPS station and an extensometer. The root mean square (RMS) accuracy of the GPS data provided is approximately 5–8 mm for the vertical direction or ellipsoidal height [33]. The study area is located over three main aquifers, including Evangeline, Chicot, and Jasper. The uppermost and youngest portion of the Chicot aquifer is composed of Pleistocene and Holocene-age sediment. Then, the lower Evangeline aquifer is composed of Miocene and Pliocene age sediments. These two aquifers are composed laterally discontinuous of clays, gravels, silts, and sands. Generally, sand is more dominant in the Chicot aquifer than in the Evangeline aquifer. Evangeline and Chicot are connected hydraulically [17, 30]

Fig. 1
figure 1

Map of study area showing the SAR coverage area and locations of GPS stations

Full size image

Datasets and data processing

The examination of ground subsidence areas and rates in the SETX region made use of Sentinel-1 SAR C-band data, which runs at a wavelength of 5.54 cm and is administered by the European Space Agency (ESA). The given diagram in Fig. 1 illustrates the Synthetic Aperture Radar (SAR) coverage of the designated study area, depicted by a red rectangle. Additionally, it displays the spatial distribution of GPS stations in the SETX region, denoted by red dots. The study area is located inside the designated path 143 and frame 492. In this study, a collection of 153 Sentinel-1 single-look complex (SLC) images was utilized. These images were obtained using the interferometric wide-swath (IW) mode and vertical transmission and vertical return (VV) polarization mode. The data was gathered during the descending observation period, which extended from January 2020 to March 2023 (Table 1). The datasets utilized in this study were obtained from the Alaska Satellite Facilities (ASF). Prior to analysis, each scene received preprocessing using the Sentinel Application Platform (SNAP) tool.

Table 1 Acquisition dates of Sentinel-1 datasets during descending observation
Full size table

A SAR image was chosen as a reference image for the purpose of coregistering another SAR image. The coregistration method was conducted with a high level of precision, achieving subpixel accuracy. This involved generating interferometric pairs and resampling the images to guarantee optimal alignment. Following that, the photos that were coregistered underwent a cropping process, resulting in images that only encompassed the particular region of interest. Subsequently, these images were employed for the purpose of generating interferograms, which effectively captured signal phase changes, or interferences, and yielded InSAR images. In order to get Differential Interferograms (DInSAR) images that accurately represent deformation phases, the interferograms were subjected to a subtraction process with the topographic phase. The topographic phase was produced using a Digital Elevation Model (DEM) with a spatial resolution of 30 m, obtained from Copernicus DEM.

PS-InSAR method

The image acquired on April 23, 2021 was selected as a reference image and subsequently employed to generate a cumulative count of 152 interferograms. These interferograms that represent DInSAR were subsequently imported into the Stanford Method for Persistent Scatterers (StaMPS) module for further processing. In the first step, an iterative methodology was utilized to calculate the phase noise value for each prospective pixel in all interferograms. Pixels were then selected based on their noise characteristics, which facilitated the determination of the proportion of random pixels that were not part of the primary source (non-PS) in the scene. Additionally, this approach enabled the computation of density per square kilometer [14]. Subsequently, a threshold value of 0.4 was implemented in order to eliminate pixels that had an impact on signals originating from adjacent ground resolution elements or exhibited excessive noise, as described by Hooper et al. [12]. The threshold value was principally taken into account in order to exclude areas over water and pixels that are significantly decorrelated in vegetation areas [27]. The spatially uncorrelated look angle error of the selected pixels was rectified by applying a DEM to correct the wrapped phase. Following this, a technique for 3D unwrapping was utilized in order to unwrap the phase accurately and consistently. For further information, refer to the study conducted by Hooper et al. [13]. The study involved the examination of unwrapped phases, with a specific focus on identifying and discarding any occurrences of unwrapping errors. The last stage of the process entailed the computation of the spatially correlated look angle (SCLA) error, which is predominantly attributed to spatially correlated errors in the DEM. Furthermore, the estimation of phase errors resulting from atmospheric and orbital components was conducted concurrently. A mean deformation picture was derived by subtracting the error image of the spatially varying phase from the unwrapped phase. In addition, a time-series analysis was conducted, which yielded line-of-sight (LOS) data aimed at capturing the temporal variations in surface deformation.

Based on the findings of Kearns et al. [16], it is postulated that an estimated horizontal displacement of roughly 2 mm exists each year, with no discernible consistent trend detected across most of the Houston region. As a result, under the assumption of little horizontal deformation, the predominant component of the line-of-sight (LOS) velocity can be attributed to vertical deformation. The equation can be used to describe the correlation between LOS deformation and subsidence, which refers to vertical deformation, given by

$$S= -\frac{d}{cos{\theta }_{i}}$$

where S represents the subsidence, d denotes the line-of-sight deformation, and θi indicates the incident angle (34 degrees) [21, 35]. It is important to acknowledge that the presence of a negative sign denotes the occurrence of ground subsidence at the specific location mentioned.

The collection, processing, and interpretation of groundwater level and GPS data fall under the purview of the Gulf Coast Program within the Texas Water Science Center of the USGS. The main goal of this study is to acquire a thorough comprehension of the effects of groundwater extraction on ground subsidence in the Houston–Galveston region of Texas. The initiative seeks to evaluate and track variations in groundwater levels and their association with ground subsidence phenomena within the region through the examination of various data sets.

Results and discussion

Time series InSAR deformation map

The utilization of the time-series InSAR methodology employing the StaMPS algorithm yielded a total of 4,320,006 measurement points (MP). Figure 2 presents the mean deformation map illustrating the patterns of subsidence observed in the SETX region. This image was generated using InSAR data collected between January 2020 and March 2023. The observed transition in color from blue to red is indicative of an escalating subsidence rate of up to 2 cm each year. In accordance with the anticipated results of the PS-InSAR technique, particularly the StaMPS algorithm employed in this investigation, urban areas exhibited a greater density of measurement points, whereas vegetated or agricultural areas had a scarcity of such points. Subsidence phenomena have been documented in the northern regions of Houston, including Spring and The Woodlands, as well as in western Houston, including Katy. These areas have exhibited notable rates of subsidence, reaching up to 2 cm per year. This sinking can be attributed to the excessive extraction of groundwater. Mont Belvieu, located east of Houston, has been recognized as a significant region experiencing notable subsidence rates due to the extensive extraction of local oil and natural gas deposits.

Fig. 2
figure 2

Mean vertical deformation map over SETX area using Sentinel-1 data from January 2020 to March 2023

Full size image

In order to compute the time series surface deformation, the average of the MPs located within a 100-m radius from a certain location was calculated. Subsequently, a comparison was made between the obtained results and the GPS measurements acquired from station P047 and station P075, as illustrated in Fig. 3a and b, respectively. The subsidence rate at station P047 was calculated to be 1.02 cm/year for the period between 2020 and 2021, as determined by GPS readings. By contrast, the SAR data revealed a subsidence rate of 0.96 cm per year over the identical time frame. The SAR observations indicate that the phenomenon of subsidence persisted throughout the year 2021, with an escalating rate of up to 1.12 cm per year. In contrast, the time series analysis of surface deformation at station P075 indicates a very consistent pattern, as evidenced by data collected from both the GPS station and SAR measurements.

Fig. 3
figure 3

Comparison between time-series GPS stations at (a) P047 and (b) P075 with StaMPS measurement

Full size image

Correlation between subsidence and groundwater level change

The Gulf Coast Aquifer System, which spans the entirety of the Texas coastline, is characterized as an intricate and multi-layered groundwater system [17]. The geological formation under consideration comprises a total thickness of approximately 3600 m, consisting of various sedimentary deposits such as clay, silt, sand, and gravel. These deposits encompass many aquifers and confining systems within the formation. The aquifer system comprises several distinct components, namely the Evangeline aquifer system, the Catahoula confining system, the Jasper aquifer, the Burkville confining system, and the Chicot aquifer. The Chicot and Evangeline aquifers serve as the principal water supplies for urban areas located in close proximity to the coastline. The phenomenon of ground subsidence in the region was initially documented during the early 1900s. The excessive extraction of groundwater from the Evangeline and Chicot aquifers has resulted in a notable reduction in water levels within specific areas of Houston, culminating in a substantial impact by the year 1977. By the year 1979, there was a recorded subsidence of around three meters in the terrain. The variations in water-level elevations above sea level within the Chicot and Evangeline aquifers from 1990 to 2021 are depicted in Fig. 4, as reported by [25].

Fig. 4
figure 4

Contours showing the changes in groundwater levels in Chicot and Evangeline aquifers at the end of 2021

Full size image

The fluctuations in water levels within the Chicot and Evangeline aquifers display diverse topographical patterns, encompassing a decrease of 80 m to an increase of 40 m. Both aquifers exhibit four unique zones of fast groundwater decrease, sometimes referred to as decline bowls. The aforementioned declining bowls are situated in Katy, Rose Hill, Egypt, and The Woodlands. Furthermore, a more extensive region seeing an increase in water levels can be observed in central Houston, Friendswood, and La Porte. It is noteworthy that locations experiencing significant subsidence rates exhibit a strong correlation with areas characterized by dropping groundwater levels, namely in the Katy, Spring, and The Woodlands regions. The observed pattern indicates a positive relationship between fluctuations in groundwater levels and the incidence of subsidence.

Figure 5 presents a time-series study of groundwater level variations at station P047 and station P075, offering significant insights into the correlation between water-level fluctuations and subsidence regions. Water-level measurements were made at observation wells in the study area by USGS personnel. The depth to water in each well was then measured by using a calibrated steel tape, airline, or electric groundwater-level tape in accordance with methods described in Cunningham and Schalk [3]. Figure 5a illustrates a noticeable decrease in the water level at station P047, commencing in April 2021. During this temporal interval, the water level experienced a decrease from a depth of 50.67 m below the surface to a depth of 61.52 m. In contrast, the water level at station P075 (Fig. 5b) exhibited a very consistent pattern, fluctuating within the range of 23 to 24 m. The observations are consistent with the time-series vertical deformation derived from the PS-InSAR measurements. These findings provide evidence that allows for the inference of a link between changes in water levels and temporal subsidence patterns. The observed decrease in water level at station P047 is consistent with subsidence activity, as evidenced by the PS-InSAR observations. In contrast, the water level at station P075 remains reasonably steady, indicating a limited occurrence of subsidence in the corresponding area. The aforementioned findings underscore the importance of monitoring groundwater levels as they offer vital insights into subsidence events and their temporal patterns.

Fig. 5
figure 5

Time series of groundwater level change at (a) station P047 and (b) station P075 from 2020 to 2023

Full size image

Figure 6 illustrates the National Flood Hazard Areas of the Federal Emergency Management Agency (FEMA), with a map displaying the mean vertical deformation over the SETX region. The maps incorporate a diverse array of data, encompassing hydrological variables such as the rate of flow, the slope of the topography, and the height, in addition to information pertaining to infrastructure, hydraulics, and land utilization (Federal Emergency Management [7]). The flood hazard map demarcates regions characterized by a flood hazard with a 1% annual probability or a recurrence interval of 100 years, as well as areas with a flood hazard of 0.2% annual probability, commonly referred to as a 500-year flood. Additionally, the map identifies locations where the risk of flooding is mitigated due to the presence of levee systems and regulating floodways. It was noted that expansive areas with elevated rates of subsidence exist beyond the demarcated flood hazard zones. According to a study conducted by Miller and Shirzaei [19], it was observed that approximately 85% of the flooded region resulting from Hurricane Harvey saw a subsidence rate of up to 5 mm/year. The study additionally demonstrated that the inundated regions exhibit signs of subsidence, as indicated by the results of the Chi-square test analysis. Hence, the ongoing phenomenon of ground subsidence in the examined region possesses the capacity to exacerbate flooding events through its detrimental effects on flood control infrastructure, modifications to floodplain delineations, and the drainage of base floods, as well as the submersion of wetland areas. Furthermore, the confluence of ground subsidence and the upward trend in sea levels have the potential to exacerbate the risks associated with flooding. A thorough comprehension of vertical ground motion can be achieved by the integration of PS-InSAR techniques with GPS data. In addition, the advancement of a hydrological model that integrates subsidence measurements holds the potential to aid coastal urban areas in reassessing areas prone to flooding and bolstering their ability to withstand flood events.

Fig. 6
figure 6

Flood risk zones provided by the National Flood Hazard Area (FEMA, 2022) displayed over the mean vertical deformation map

Full size image

Conclusions

This study provides an analysis of the ground subsidence observed in the Southeast Texas (SETX) region. The PS-InSAR measurements made between January 2020 and March 2023 have shown a significant subsidence rate of up to 2 cm/year in the northwest and eastern regions. The comparison with GPS and groundwater monitoring networks has validated accurate monitoring of InSAR ground subsidence. The findings suggest that fluctuations in groundwater levels have a significant role in the occurrence of ground subsidence in the SETX region. This observation is supported by the relationship between the changes in groundwater level and the geographical and temporal patterns of subsidence. Additionally, the study illustrates that subsidence rates display notable fluctuations over time and space, mostly influenced by the specific practices of local groundwater extraction. Moreover, it is important to acknowledge that the occurrence of floods and storm events, as demonstrated by the case of Hurricane Harvey, brings about the potential for flood-induced subsidence, thereby intensifying the frequency and magnitude of flooding. This study ultimately aims to make a contribution toward mitigating subsidence and flood dangers in the SETX region by promoting the utilization of InSAR, GPS, and groundwater data.