Landslide displacement prediction is crucial for the early warning of slope failure but remains a challenging task due to its spatiotemporal complexity. Although temporal dependency has been well studied and discussed, spatial dependence is relatively less explored due to its significant variations of the spatial structure of landslides. In this study, a novel graph convolutional incorporating GRU network (GC-GRU-N) is proposed and applied to landslide displacement forecasts. The model conducts attribute-augmented graph convolution (GC) operations on GNSS displacement data with weighted adjacency matrices and an attribute-augmented unit to combine features, including the displacements, the distance, and other external influence factors to capture spatial dependence. The output of multi-weight graph convolution is then applied to the gated recurrent unit (GRU) network to learn temporal dependencies. The related optimal hyper-parameters are determined by comparison experiments. When applied to two typical landslide sites in the Three Gorge Reservoir (TGR), China, GC-GRU-N outperformed the comparative models in both cases. The ablation experiment results also show that the attribute augmentation, which considers external factors of landslide displacement, can further improve the model’s prediction performance. We conclude that the GC-GRU-N model can provide robust landslide displacement forecasting with high efficiency.

Many coastal cities have undertaken reclamation projects to satisfy the land demands of rapid urbanization. However, the foundations of reclaimed land are susceptible to settlement and can have undesirable environmental impacts that could adversely affect these dense, populated areas. In the case of international airports built on reclaimed areas especially, regional-scale deformation must be monitored to ensure operational security for public safety. Persistent Scatterer SAR Interferometry (PSI) technology has proven to be an effective tool to detect ground deformation in urban areas. However, it is still a challenge to apply PSI to effectively monitor settlement at airports built on newly developed coastal reclamation areas because of the scarcity of identifiable targets. Moreover, additional issues arise as the complicated deformation patterns associated with the underlying geological conditions make it difficult to interpret InSAR-derived results. In this study, a time-series analysis of a high-resolution TerraSAR-X satellite image stack acquired from September 2011 to October 2012 was performed by employing a modified PSI technique to retrieve the mean deformation velocity and time series of surface deformation at Pudong International Airport. Qualitative evaluation of spatial distribution and temporal evolution of deformation was conducted by joint analyses of deformation measurements and local geological data. Detailed analysis of various driving forces for deformation patterns confirmed that the results of deformation monitoring obtained by PSI are reliable and consistent with that of local geological surveys. Since the factors responsible for the subsidence within the airport are still at play, ongoing and routine deformation monitoring is warranted.

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

Time-series SAR/InSAR techniques have proven to be effective tools for measuring landslide movements over large regions. Prior studies of these techniques, however, have focused primarily on technical innovation and applications, leaving coupling analysis of slope displacements and trigging factors as an unexplored area of research. Linking potential landslide inducing factors such as hydrology to SAR/InSAR derived displacements is of crucial importance for understanding landslide deformation mechanisms and could support the development of early-warning systems for disaster mitigation and management. In this study, a sequential data assimilation method named the Ensemble Kalman filter (EnKF), is adopted to explore the response mechanisms of the Shuping landslide movement in relation to hydrological factors. Previous research on the Shuping landslide area shows that the reservoir water level and rainfall are the two main triggering factors in slope failures. To extract the time-series deformations for the Shuping landslide area, Pixel Offset Tracking (POT) technique with corner reflectors was adopted to process the TerraSAR-X StripMap (SM) and High-resolution Spotlight (HS) images. Considering that these triggering factors are the primary causes of displacement fluctuations in periodic displacement, time-series decomposition was carried out to extract the periodic displacement from the POT measurements. The correlations between the periodic displacement and the inducing factors were qualitatively estimated through a grey relational analysis. Based on this analysis, the EnKF method was adopted to explore the response relationships between the displacements and triggering factors. Preliminary results demonstrate the effectiveness of EnKF in studying deformation response mechanisms and understanding landslide development processes.

A mega project, Mountain Excavation and City Construction (MECC), was launched in the hilly and gully region of the Chinese Loess Plateau in 2012, in order to address the shortage of available land and create new flat land for urban construction. However, large-scale land creation and urban expansion significantly alters the local geological environment, leading to severe ground deformation. This study investigated the topographic changes, ground deformation, and their interactions due to the MECC project in the Yan’an New District (YND). First, new surface elevations were generated using ZiYuan-3 (ZY-3) stereo images acquired after the construction in order to map the local topographic changes and the fill thickness associated with the MECC project. Then, the interferometric synthetic aperture radar (InSAR) time series and 32 Sentinel-1A images were used to assess the spatial patterns of the ground deformation in the YND during the postconstruction period (2017–2018). By combining the InSAR-derived results and topographic change features, the relationship between the ground deformation and large-scale land creation was further analyzed. The results indicated that the MECC project in the YND has created over 22 km² of flat land, including 10.8 km² of filled area, with a maximum fill thickness of ~110 m. Significant uneven ground deformation was detected in the land-creation area, with a maximum subsidence rate of approximately 121 mm/year, which was consistent with the field survey. The strong correlation between the observed subsidence patterns and the land creation project suggested that this recorded uneven subsidence was primarily related to the spatial distribution of the filling works, along with the changes in the thickness and geotechnical properties of the filled loess; moreover, rapid urbanization, such as road construction, can accelerate the subsidence process. These findings can guide improvements in urban planning and the mitigation of geohazards in regions experiencing large-scale land construction.

Reservoir water and rainfall, leading to fluctuations groundwater levels, are the main triggering factors that induce landslides in the Three Gorges Reservoir area. This study investigates the response mechanism of landslide deformation under reservoir water and rainfall variations through long-time on-site observations. To address the non-stationary characteristics of the time-series records, joint time-frequency analysis (JTFA) is first introduced into our landslide prediction model. This model employs optimal variational mode decomposition (VMD) to obtain specific signal components with clear physical meaning, such as trend component and periodic components. Then, multi-scale response analysis between the displacement and external factors three wavelet methods was conducted. The analysis results show a 1 year primary cycle of the time series associated with the landslide evolution. The reservoir water level and rainfall show anti-phase fluctuations. The periodic displacement correlates significantly with rainfall, lagging by about two months. The reservoir water is anti-phase with the landslide displacement, preceding it by approximately three months (−51 ± 8° phase difference). For landslide displacement prediction, the gated recurrent units (GRU) neural network model is integrated into the deep learning forecasting architecture. The model takes into account the correlation and hysteresis effect of input variables. Through six experiments, we investigate the effect of data volume on model predictions to determine the optimal model. The results demonstrate that our proposed model ensures high performance in landslide prediction. Moreover, a comparison with six other intelligent algorithms shows the advantages of our model in terms of time-effectiveness and long-sequence forecasting.