Displacement Time Series Research Articles

Mapping Surface Deformation in Rwanda and Neighboring Areas Using SBAS-InSAR

Surface deformation poses significant risks to urban infrastructure, agriculture, and the environment in many regions worldwide, including Rwanda and the neighboring areas. This study focuses on surface deformation mapping and time series analysis in Rwanda and the neighboring areas from 2 July 2016 to 8 June 2023 using the Small Baseline Subset Interferometric Synthetic Aperture Radar (SBAS-InSAR). The influence of atmospheric delay error is effectively reduced by integrating the Generic Atmospheric Correction Online Service (GACOS), which provides precise atmospheric delay maps. Then the SBAS-InSAR method is utilized to generate surface deformation maps and displacement time series across the region. The results of this study indicated that the maximum deformation rate was −0.11 m/yr (subsidence) and +0.13 m/yr (uplift). Through time series analysis, we quantified subsidence and uplift areas and identified key drivers of surface deformation. Since subsidence or uplift varies across the region, we have summarized the different deformation patterns and briefly analyzed the factors that may lead to deformation. Finally, this study underscores the importance of SBAS-InSAR for tracking surface deformation in Rwanda and the neighboring areas, which offers valuable perspectives for sustainable land utilization strategizing and risk mitigation.

Kinematic GNSS inversion of the large afterslip (Mw 6.4) following the 2019 Mw 6.2 Hualien earthquake (Taiwan)

The postseismic deformation following the April 2019 Mw 6.2 Hualien earthquake presents an unique opportunity to investigate the mechanisms by which the northern section of the Longitudinal Valley accommodates lithospheric deformation. We apply a variational Bayesian independent component analysis approach to displacement time-series to infer a 6-month long afterslip. Kinematic inversion shows that displacements are well explained by widespread afterslip (∼60 km in the along-strike direction) with limited slip (≤0.1 m) surrounding the coseismic slip area. The total geodetic moment relieved by afterslip (M0 ∼ 4.6 × 1018 Nm, i. e., Mw ∼ 6.4) is twice as large as the mainshock seismic moment, which represents a rare exception of a moderate magnitude event for which the afterslip moment exceeds that of the seismic moment. Then, combining geodetic and seismological analysis, we infer that afterslip is the dominant mechanism of near-to intermediate-field postseismic deformation and also likely represents the driving force that controls aftershock productivity and the spatiotemporal migration of seismicity. Besides, the fault zone frictional stability parameter a-b of rate-and-state dependent friction (a-b ∼ 0.0067–0.02) is comparable with previous estimates in the Longitudinal Valley. Finally, the study demonstrates that the northern Longitudinal Valley region hosts complex seismogenic structures that display a variety of slip behaviors.

A GIS Plugin for the Assessment of Deformations in Existing Bridge Portfolios via MTInSAR Data

The paper presents a GIS plugin, named Bridge Assessment System via MTInSAR (BAS-MTInSAR), aimed at assessing deformations in existing simply supported concrete girder bridges through Multi-Temporal Interferometry Synthetic Aperture Radar (MTInSAR). Existing bridges require continuous maintenance to ensure functionality toward external effects undermining the safety of these structures, such as aging, material degradation, and environmental factors. Although effective and standardized methodologies exist (e.g., structural monitoring, periodic onsite inspections), new emerging technologies could be employed to provide time- and cost-effective information on the current state of structures and to drive prompt interventions to mitigate risk. One example is represented by MTInSAR data, which can provide near-continuous information about structural displacements over time. To easily manage these data, the paper presents BAS-MTInSAR. The tool allows users to insert information of the focused bridge (displacement time series, structural information, temperature data) and, through a user-friendly GUI, observe the occurrence of abnormal deformations. In addition, the tool implements a procedure of multisource data management and defines proper thresholds to assess bridge behavior against current code prescriptions. BAS-MTInSAR is fully described throughout the text and was tested on a real case study, showing the main potentialities of the tool in managing bridge portfolios.

Temporal gravimetry, campaign and permanent GNSS, and interferometric radar techniques: A comparative approach to quantifying land deformation rates in coastal Texas

Land subsidence and local sea-level rise are well-known on-going hazards that negatively impacting coastal regions, exacerbating coastal flooding, threatening infrastructure stability, accelerating erosion, and amplifying the risks of inundation and property damage. This study used four techniques to quantify land deformation rates in the Texas Coastal Bend region and to investigate the controlling factors on two different spatiotemporal scales: (1) local scale, where biweekly temporal gravity and campaign Global Navigation Satellite System (GNSS) measurements were acquired at six locations over 2 years (October 2020–September 2022); and (2) regional scale, where vertical displacement time series were extracted from the Interferometric Synthetic Aperture Radar (InSAR) and 15 permanent GNSS stations over 5 years (January 2017–November 2021). The observed inconsistency between land deformation rates derived from gravity (range: −33.32 ± 149.30 to +338.68 ± 107.93 mm/yr) and campaign GNSS (range: −17.91 ± 14.90 to +5.69 ± 9.93 mm/yr) surveys could be attributed to the short campaign period and the infrequency of data acquisition. The InSAR-derived deformation rates (range: −12.16 ± 0.12 to +11.70 ± 0.00 mm/yr) were consistent with the permanent GNSS-derived rates (range: −14.0 ± 0.6 to +9.0 ± 3.0 mm/yr); both indicated spatiotemporal variability in land deformation rates across the Texas Coastal Bend. A total of 4 coastal towns (−2.3 ± 3.3 mm/yr), 3 cities (−1.4 ± 2.6 mm/yr), 3 inland towns (−3.6 ± 4.2 mm/yr), and 7 industrial plants (−3.8 ± 5.8 mm/yr) were found to have significant land subsidence rates due to sediment compaction, growth faulting, oil/gas extraction, and groundwater extraction. Results of this study emphasize the advantages and limitations of four different techniques in quantifying land deformation rates; can enhance estimates of local sea-level rise; and offer valuable insights to support coastal communities in mitigating the impacts of land subsidence and improving their resilience.

Removing atmospheric noise from InSAR interferograms in mountainous regions with a convolutional neural network

Atmospheric noise in interferometric synthetic aperture radar (InSAR)-derived estimates of surface deformation often obscures real displacement signals, especially in mountainous regions. As climate change disproportionately impacts the mountain cryosphere, a reliable technique for atmospheric correction in high-relief terrain is increasingly important. We developed and implemented a statistical machine learning atmospheric correction approach that relies on the differing spatial and topographic characteristics of slow-moving periglacial features and atmospheric noise. Our correction is applied at the native spatial and temporal resolution of the InSAR data, does not require external atmospheric reanalysis data, and can correct both stratified and turbulent atmospheric noise.Using Sentinel-1 data from 2017 to 2022, we trained a convolutional neural network (CNN) on observed atmospheric noise from 330 short-baseline interferograms and observed displacement signals from time series inversion of 1322 interferograms. We applied our trained CNN to correct 251 additional interferograms over an out-of-region application area, which were inverted to create displacement time series. We used the Rocky Mountains in New Mexico, Colorado, and Wyoming as our training, validation, testing, and application areas. When applied to our testing dataset, our correction offered performance improvements of 131%, 208%, and 68% in structural similarity index measure over corrections using atmospheric reanalysis data, phase correlation with topography, and high-pass filtering, respectively. The CNN-corrected time series reveals previously obscured kinematic behavior of rock glaciers and other features in the application dataset. Our flexible, robust approach can be used to correct arbitrary InSAR data to analyze subtle surface deformation signals for a range of science and engineering applications.

A data-driven approach for analyzing contributions of individual loading factors to GNSS-measured bridge displacements

A bridge may displace due to various loadings (e.g., thermal (Xia et al. in Struct Control Health Monit 28(7):e2738, 2013), winds (Owen et al. in J Wind Eng Ind Aerodyn 206:104389, 2020), and vehicles (Xu et al. in J Struct Eng 133(1):3–11, 2007)) acting upon the bridge. Identifying the contributions of individual loading factors to the measured bridge displacements is important for understanding the structural health conditions of the bridge. There is however no effective method to quantify the contributions when multiple loadings act simultaneously on a bridge. We propose a new data-driven method, termed random forest (RF)-assisted variational mode decomposition (RF-AVMD), for more effective identification of dominant loading factors and for quantifying the contributions of individual loading factors to the measured bridge displacements. The proposed method is applicable to studying the displacements of any bridge structures and allows for the first time to separate the contributions of individual loadings. The effectiveness of the proposed method is validated using data from Tsing Ma Bridge (TMB), a large suspension bridge in Hong Kong recorded during two consecutive strong typhoons. The results reveal that the transverse displacements of TMB mid-span were controlled by the crosswinds, the longitudinal displacements were dominated by the temperature and winds along the bridge centerline, and the vertical displacements were mainly due to the winds along the bridge centerline, temperature, and traffic flows. Displacement time series that responded to each loading factor was derived. The proposed method provides important new insights into the impacts of individual loadings on the displacements of long-span bridges.

Activation of Ms 6.9 Milin Earthquake on Sedongpu Disaster Chain, China with Multi-Temporal Optical Images

In recent years, disaster chains caused by glacier movements have occurred frequently in the lower Yarlung Tsangpo River in southwest China. However, it is still unclear whether earthquakes significantly contribute to glacier movements and disaster chains. In addition, it is difficult to measure the high-frequency and large gradient displacement time series with optical remote sensing images due to cloud coverage. To this end, we take the Sedongpu disaster chain as an example, where the Milin earthquake, with an epicenter 11 km away, occurred on 18 November 2017. Firstly, to deal with the cloud coverage problem for single optical remote sensing analysis, we employed multiple platform optical images and conducted a cross-platform correlation technique to invert the two-dimensional displacement rate and the cumulative displacement time series of the Sedongpu glacier. To reveal the correlation between earthquakes and disaster chains, we divided the optical images into three classes according to the Milin earthquake event. Lastly, to increase the accuracy and reliability, we propose two strategies for displacement monitoring, that is, a four-quadrant block registration strategy and a multi-window fusion strategy. Results show that the RMSE reduction percentage of the proposed registration method reaches 80%, and the fusion method can retrieve the large magnitude displacements and complete displacement field. Secondly, the Milin earthquake accelerated the Sedongpu glacier movement, where the pre-seismic velocities were less than 0.5 m/day, the co-seismic velocities increased to 1 to 6 m/day, and the post-seismic velocities decreased to 0.5 to 3 m/day. Lastly, the earthquake had a triggering effect around 33 days on the Sedongpu disaster chain event on 21 December 2017. The failure pattern can be summarized as ice and rock collapse in the source area, large magnitude glacier displacement in the moraine area, and a large volume of sediment in the deposition area, causing a river blockage.

A New Algorithm for Predicting Dam Deformation Using Grey Wolf-Optimized Variational Mode Long Short-Term Neural Network

To solve the problems of difficult-to-model parameter selections, useful-signal extraction and improper-signal decomposition in nonlinear, non-stationary dam displacement time series prediction methods, we propose a new predictive model for grey wolf optimization and variational mode decomposition and long short-term memory (GVLSTM). Firstly, we used the grey wolf optimization (GWO) algorithm to optimize the parameters of variable mode decomposition (VMD), obtaining the optimal parameter combination. Secondly, we used multiscale permutation entropy (MPE) as a standard to select signal screening, determining and recon-structing the effective modal components. Finally, the long short-term memory neural network (LSTM) was used to learn the dam deformation characteristics. The result shows that the GVLSTM model can effectively reduce the estimation deviation of the prediction model. Compared with VMDGRU and VMDANN, the average RMSE and MAE value of each station is increased by 19.11%~28.58% and 27.66%~29.63%, respectively. We used determination (R2) coefficient to judge the performance of the prediction model, and the value of R2 was 0.95~0.97, indicating that our method has good performance in predicting dam deformation. The proposed method has outstanding advantages of high accuracy, reliability, and stability for dam deformation prediction.

PS‐InSAR post‐processing for assessing the spatio‐temporal differential kinematics of complex landslide systems: A case study of DeBeque Canyon Landslide (Colorado, USA)

AbstractThe complex superimposition of different kinematics and nested sectors within landslide systems amplifies the challenge of interpreting their heterogeneous displacement pattern and targeting effective mitigation solutions. As an example of such peculiar spatio‐temporal behaviour, the DeBeque Canyon Landslide (Colorado, USA) is emblematic of the application of interferometric post‐processing analysis for a detailed, remotely‐based investigation. We employed a multi‐geometry Persistent Scatterers (PS) InSAR dataset to provide continuous information on the spatio‐temporal scale and achieve a solid representation of the segmented kinematics and timings. Using an updated geomorphological map of the landslide system, we performed a two‐dimensional decomposition of the Persistent Scatterers (PS) dataset to determine the displacement orientation and inclination for each internal sector of the landslide system. We then conducted statistical analyses on the displacement vector characteristics and time series data. These analyses enabled us to spatially characterize the segmented activity patterns of the landslide system and identify abrupt changes in trends associated with preparatory and triggering factors. A clear differentiation of the rotational or translational kinematics within the landslide system was accomplished solely using surface displacement measures. Moreover, the application of a Bayesian model on the bi‐dimensional vector time series leads to the identification of significant differences in the deformational behaviour of each sector with respect to precipitation and temperature factors. Our approach represents a replicable method for local‐scale characterization and monitoring of landslides exhibiting complex spatio‐temporal displacement patterns and providing an effective, low‐cost solution for transportation agencies from a risk‐reduction perspective.

Effect of spatially varying earthquake ground motions on seismic response of a railway viaduct considering multiple site configurations

Currently, it is admitted that extended structures are subjected to spatially varying earthquake ground motions. It has been recognized that the causes of this variability are time delay at each support of the structure, coherency loss effect caused by the propagation of seismic waves and local site effect. The spatial variation of local characteristics of the soil profile defines the site effect, which affects the amplitude and the frequency content of seismic ground motion. The main objective of this work is to provide comparative results and evaluate the variation of the dynamic response of a bridge adopting varying characteristics of soil foundation. Based on the density spectral method, an efficient simulation method of spatially varying earthquake ground motions is developed. The simulation of spatially variable seismic ground motions is performed for different locations on the ground surface with varying site conditions. According to of soil classification described in seismic codes, four different soil configurations were considered for generating sixty-four displacement time series. Several dynamics analyses of a bridge to three cases of spatially variable seismic ground motions, besides the uniform case, are made. The results of this study indicate that depending on soil configurations beneath each support, the seismic response may vary significantly.

Landslide displacement prediction using time series InSAR with combined LSTM and TCN: application to the Xiao Andong landslide, Yunnan Province, China

AbstractResearch on landslide displacement prediction based on interferometric synthetic aperture radar (InSAR) deformation data involves two main issues. First, InSAR can provide only one-dimensional deformation data along the satellite’s line of sight (LOS), which cannot truly reflect the deformation of the landslide body in the downward direction along the slope. Second, the use of a single prediction model does not adequately account for both long-term and local changes in landslide displacement, affecting the accuracy of the predictions. To address this, in this study, Long Short-Term Memory networks (LSTM) and temporal convolutional network (TCN) models are combined to construct a method (LSTM-TCN) of landslide displacement prediction. This method can consider the long-term and localized changes in landslide displacement. The method is first based on InSAR technology to obtain surface deformation. The deformation of the landslide is subsequently computed in the downward direction along the slope to obtain the landslide displacement time series data. Next, the LSTM-TCN is used for landslide displacement prediction. Finally, the mean absolute error (MAE), root mean square error (RMSE) and coefficient of determination (R2) are used to evaluate the performance of the model. The experiment is conducted on the Xiao Andong landslide in Anshi village, Fengqing County, Lincang City, Yunnan Province, China. The LSTM-TCN model achieves an R2 of 0.75, an RMSE of 0.43 cm, and an MAE of 0.36 cm. Compared with the individual LSTM and TCN models, the LSTM-TCN model exhibits the highest prediction accuracy and the smallest prediction error, which is closer to the true result that in the other models. These results demonstrate that the combined LSTM-TCN model effectively captures the complex features and long-term trends in landslide displacement data, significantly enhancing the accuracy of predictions.

A Well-Tested Pump Card Classification and Identification Algorithm Based on Time Series Analysis Techniques

Abstract To accurately identify the operational conditions of oil pumping wells, promptly monitor the working status of oil wells, expedite responses to sudden pumping well accidents, and ensure orderly production at the oil extraction site, this paper introduces a time segment-based oil pumping well condition identification algorithm. This method no longer relies on drawn dynamometer cards for condition identification. Instead, it solely utilizes sensor-provided polished rod displacement and load time series, aligning more closely with the data’s intrinsic nature. The algorithm is sufficiently lightweight, highly responsive, and does not require GPU resources, making it suitable for edge deployment. Furthermore, this approach provides evidence of the time continuity and numerical continuity of time series data. Compared to traditional identification methods, this approach maintains a high level of accuracy, offers a streamlined model, and is interpretable, presenting a novel perspective for oil pumping well condition identification.

Battery Condition Monitoring of Quadrotor UAV Using Machine Learning Classification Algorithm

Unmanned aerial vehicle flight performance and efficiency rely on various factors. Flight instabilities can happen due to malfunctions inside the system and disturbances from the external environment. Battery status plays a significant role in healthy flight conditions. A weak battery will affect the performance of propellers and motors, and the presence of wind disturbance can contribute towards inefficient flying capabilities. Therefore, investigation of fault at the early stage is crucial to maintain the great performance of the UAV. This paper aims to investigate the best prediction system from the existing machine learning algorithm such as Decision Tree (DT), Linear Discriminant (LD), Naïve Bayes (NB), Support Vector Machine (SVM), K-Nearest Neighbors (KNN) and Neural Network (NN) to classify the battery condition of the quadrotor by extracting the features from the displacement time series dataset. By using recorded flight data, it will be statistically analyzed to extract the flying condition features. The extracted features are the Euclidian distance (ED), speed, acceleration, Periodogram Power spectral density (PSD) and Fast Fourier Transform (FFT) of the signal. The result shows that the two best classifier algorithms are the Decision Tree and Neural Network models with training accuracy of 98% and 93% in Set A and B, respectively.

Multidisciplinary assessment of seasonal ground displacements at the Hatfield Moors gas storage site in a peat bog landscape

The study aims to analyse ground displacement conditions observed over an Underground Gas Storage (UGS) site located at Hatfield Moors (United Kingdom), with a focus on understanding its implications for decarbonization efforts. The location serves as an active onshore storage site and was used as an analogy to assess ground motion implications around Carbon Capture and Storage (CCS) by the British Geological Survey (BGS) as part of the SENSE (Assuring integrity of CO2 storage sites through ground surface monitoring) project. Given the value of continuous and real-time monitoring of ground movements induced by gas storage activities, the study leverages satellite Interferometric Synthetic Aperture Radar (InSAR) data to assess the environmental impact of UGS operations. Using free and open-source Sentinel-1 satellite data, ground motion patterns over Hatfield Moors are analysed, highlighting displacements ranging from − 5.0 to -10.0 mm/year within the peat bog. In addition, the Time Series (TS) of ground displacement from January 2018 to December 2022 reveals a seasonality in ground motion, with uplift observed in late winter and subsidence in late summer, showing a periodicity of approximately 1 year and a magnitude of +/-10.0 mm. Through in-depth analysis, the study highlights the need to understand the underlying causes of ground fluctuations at gas storage sites. This paper shows that InSAR has the versatility to integrate seamlessly with different monitoring tools and methodologies, opening avenues for comprehensive and holistic analyses. Cross-correlation analyses further elucidate temporal relationships between different datasets by evaluating InSAR time series, UGS injection/withdrawal data and piezometric data. This involves decomposing the TS into distinct components, including trend, seasonality and residuals. The case of Hatfield Moors shows a significant discrepancy between the UGS data and the InSAR TS, while also demonstrating a clear correlation between the groundwater data and the InSAR TS. By integrating insights from geology, hydrology and remote sensing technologies, the study navigates the complexities inherent in areas of overlapping phenomena. Accurate interpretation is essential for informed decision making, particularly at sites such as Hatfield Moors, where the convergence of natural peat motion and storage operations highlights the need for interdisciplinary analysis to understand the underlying causes of ground fluctuation.

Integrating low-cost GNSS and MEMS accelerometer for precise dynamic displacement monitoring

We assess the performance of the self-developed receiver coupling dual-frequency low-cost GNSS chipset and IMU MEMS sensor. The system also comprises software dedicated to vibration monitoring based on relative kinematic positioning in a post-processing mode. The observation assessment reveals high accuracy of low-cost GNSS and MEMS accelerometer data. The Septentrio Mosaic X5 chipset exhibits competitive to high-grade receivers’ phase observation noise, outperformance in code noise level and observation auto-correlation at a level that may be neglected even in high-rate applications. The evaluation of the MEMS accelerometer data proves that the sensor accuracy is adequate for vibration monitoring. With the shake table experiment, we examine the performance of vibration recovery. An advantage of the coupled solution over a GNSS-only one is documented with a 15 % reduction of the amplitude error. The backward smoothing of coupled solutions drives the next 20 % error drop. Finally, the high performance of the coupled solution based on low-cost GNSS & accelerometer is confirmed by a mean amplitude error of 0.4 mm. The frequency analyses indicate that both coupled and single-sensor solutions precisely detect the vibration frequencies; however, they also confirm the outperformance of the former over the latter. The power spectra of the recovered displacement time series accentuate the benefits of GNSS + accelerometer fusion and backward filtering for reducing displacement noise.

Prognostic Properties of Instantaneous Amplitudes Maxima of Earth Surface Tremor.

A method is proposed for analyzing the tremor of the earth's surface, measured by GPS, in order to highlight prognostic effects. The method is applied to the analysis of daily time series of vertical displacements in Japan. The network of 1047 stations is divided into 15 clusters. The Huang Empirical Mode Decomposition (EMD) is applied to the time series of the principal components from the clusters, with subsequent calculation of instantaneous amplitudes using the Hilbert transform. To ensure the stability of estimates of the waveforms of the EMD decomposition, 1000 independent additive realizations of white noise of limited amplitude were averaged before the Hilbert transform. Using a parametric model of the intensities of point processes, we analyze the connections between the instants of sequences of times of the largest local maxima of instantaneous amplitudes, averaged over the number of clusters and the times of earthquakes in the vicinity of Japan with minimum magnitude thresholds of 5.5 for the time interval 2012-2023. It is shown that the sequence of the largest local maxima of instantaneous amplitudes significantly more often precedes the moments of time of earthquakes (roughly speaking, has an "influence") than the reverse "influence" of earthquakes on the maxima of amplitudes.

Toward long-term monitoring of regional permafrost thaw with satellite interferometric synthetic aperture radar

Abstract. Active-layer thickness (ALT) is estimated for a study area in northern Alaska's continuous-permafrost zone using satellite data from Sentinel-1 (radar) and ICESat-2 (lidar) for the period 2017 to 2022. Synthetic aperture radar (SAR) interferograms were generated using the Short Baseline Subset (SBAS) approach. Displacement time series over the thaw season (June–September) are fit well with a linear model (root mean square error (RMSE) scatter is less than 7 mm) and show maximum seasonal subsidence of 20–60 mm. ICESat-2 products were used to validate the interferometric synthetic aperture radar (InSAR) displacement time series. ALT was estimated from measured subsidence using a widely used model exploiting the volume difference between ice and water, reaching a maximum depth in our study area of 1.5 m. Estimated ALT is in good agreement with in situ and other remotely sensed data but is sensitive to assumed thaw season onset, indicating the need for reliable surface temperature data. Our results suggest the feasibility of long-term permafrost monitoring with satellite InSAR. However, the C-band (∼55 mm center wavelength) Sentinel radar is sensitive to vegetation cover and, in our studies, was not successful for similar monitoring in the heavily treed discontinuous-permafrost zone of central Alaska.

Prediction of the displacement in a foundation pit based on neural network model fusion error and variational modal decomposition methods

Prediction of foundation pit displacements is an important part of pit monitoring and early warning system, and accurate prediction of pit displacements can reduce pit disaster accidents. Herein, the Elman neural network model optimized by the mind evolutionary algorithm (MEA) is combined with the variational modal decomposition (VMD) and the fusion error method to accurately predict pit displacements at the top of the slope. Firstly, VMD is used to decompose the time series of pit displacements into trend, period, and random terms, and analyzed the rationality of the decomposition of the sub-displacement sequences; Secondly, the MEA-Elman model is used to predict the trend, period, and random terms and decomposition error in the decomposition of pit displacements, and finally the prediction results of the sub-displacements are superimposed to obtain the total displacement; The measured data of a pit in Qingdao, China, are used to verify the reliability of the model, and the calculated results show that the predictions are consistent with the measured data.

Land surface response to groundwater drawdown and recovery in Taiyuan city, Northern China, analyzed with a long-term elevation change measurements from leveling and multi-sensor InSAR

In recent years, a comprehensive program of groundwater management was adopted in Taiyuan city, Northern China, with the aim to prevent the lowering of groundwater level and manage land subsidence due to excessive groundwater pumping in the past decades. Groundwater recovery has been observed in this city following the termination of groundwater pumping. While the link between land subsidence and groundwater pumping is well documented, how the land surface responds to the recovery of groundwater has not been studied in detail. In this study, we analyzed the leveling data from 1956 to 2000 to investigate changes in ground elevation in response to groundwater drawdown prior to the termination of groundwater pumping and combined them with InSAR (Interferometric Synthetic Aperture Radar) displacement data during 2003–2020 to evaluate the effect of groundwater recharge after the termination of groundwater pumping. A method was proposed to merge InSAR time series of displacement derived from multiple satellites (ENVISAT, COSMO-SkyMed, TerraSAR-X, and Sentinel-1) to achieve a consistent, long-term continuous deformation. We found that previous groundwater depression cones in Taiyuan (Xizhang, Wanbailin, Wujiabao, and Xiayuan) have turned into groundwater recovery and the corresponding subsidence centers have switched to land uplift. The subsidence center of Xizhang, north of Taiyuan city, experienced a total subsidence of 749 mm by Aug 2003; it then turned into uplift with an amount of rebound of 9.2 % of previous subsidence by the end of 2020. In central area of Taiyuan, the cumulated subsidence is 1127 mm in Wanbailin, 1817 mm in Xiayuan, and 3343 mm in Wujiabao. Beginning in Aug 2010, they all turned into land uplift, with a rebound of 7.9 %, 4.8 % and 3.6 % of the previous subsidence in each center, respectively. An analysis of the correlation between groundwater level and ground displacement suggests that the ground uplift is related to the poro-elastic rebound mechanism in the porous aquifer system, which is caused by the rise in pore pressure with groundwater recovery, and the amount of subsidence/uplift is controlled by the stratigraphic properties. The adverse impacts on the built environment due to the rising groundwater level were discussed. The findings improve our understanding of anthropogenic ground deformation in complex urban aquifers influenced by groundwater pumping and recharge and provides essential information that can contribute to future groundwater management and groundwater-related hazard mitigation over the Taiyuan city.

Time Series Prediction of Reservoir Bank Slope Deformation Based on Informer and InSAR: A Case Study of Dawanzi Landslide in the Baihetan Reservoir Area, China

Reservoir impoundment significantly impacts the hydrogeological conditions of reservoir bank slopes, and bank slope deformation or destruction occurs frequently under cyclic impoundment conditions. Ground deformation prediction is crucial to the early warning system for slow-moving landslides. Deep learning methods have developed rapidly in recent years, but only a few studies are on combining deep learning and landslide warning. This paper proposes a slow-moving landslide displacement prediction method based on the Informer deep learning model. Firstly, the Sentinel-1 (S1) data are processed to obtain the cumulative displacement time-series image of the bank slope by the Small-BAseline Subset Interferometric Synthetic Aperture Radar (SBAS-InSAR) method. Then, combining data on rainfall, humidity, and horizontal and vertical distances of pixel points from the water table line, this study created a dataset with landslide displacement as the target feature. After that, this paper improves the Informer model to make it applicable to our dataset. This study chose the Dawanzi landslide in the Baihetan reservoir area, China, for validation. After training with 50-time series deformation data points, the model can predict the displacement results of 12-time series deformation data points using 12-time series multi-feature data, and compared with the monitoring values, its Mean Square Error (MSE) was 11.614. The results show that the multivariate dataset is better than the deformation univariate data in predicting the displacement in the large deformation zone of bank slopes, and our model has better complexity and prediction performance than other deep learning models. The prediction results show that among zones I–IV, where the Dawanzi Tunnel is located, significant deformation with the maximum deformation rate detected exceeding –100mm/year occurs in Zones I and III. In these two zones, the initiation of deformation relates to the drop in water level after water storage, with the deformation rate of Zone III exhibiting a stronger correlation with the change in water level. It is expected that deformation in Zone III will either remain slow or stop, while deformation in Zone I will continue at the same or a decreased rate. Our proposed method for slow-moving landslide displacement forecasting offers fast, intuitive, and economically feasible advantages. It can provide a feasible research idea for future deep learning and landslide warning research.