Although ground-based synthetic aperture radar (GB-SAR) interferometry has a very high precision with respect to deformation monitoring, it is difficult to match the fan-shaped grid coordinates with the local topography in the geographical space because of the slant range projection imaging mode of the radar. To accurately identify the deformation target and its position, high-accuracy geocoding of the GB-SAR images must be performed to transform them from the two-dimensional plane coordinate system to the three-dimensional (3D) local coordinate system. To overcome difficulties of traditional methods with respect to the selection of control points in GB-SAR images in a complex scattering environment, a high-resolution digital surface model obtained by unmanned aerial vehicle (UAV) aerial photogrammetry was used to establish a high-accuracy GB-SAR coordinate transformation model. An accurate GB-SAR image geocoding method based on solution space search was proposed. Based on this method, three modules are used for geocoding: framework for the unification of coordinate elements, transformation model, and solution space search of the minimum Euclidean distance. By applying this method to the Laoguanjingtai landslide monitoring experiment on Hailuogou Glacier, a subpixel geocoding accuracy was realized. The effectiveness and accuracy of the proposed method were verified by contrastive analysis and error assessment. The method proposed in this study can be applied for accurate 3D interpretation and analysis of the spatiotemporal characteristic in GB-SAR deformation monitoring and should be popularized.

Landslide is the second most frequent geological disaster after earthquake, which causes a large number of casualties and economic losses every year. China frequently experiences devastating landslides in mountainous areas. Interferometric Synthetic Aperture Radar (InSAR) technology has great potential for detecting potentially unstable landslides across wide areas and can monitor surface displacement of a single landslide. However traditional time series InSAR technology such as persistent scatterer interferometry (PSI) and small-baseline subset (SBAS) cannot identify enough points in mountainous areas because of dense vegetation and steep terrain. In order to improve the accuracy of landslide hazard detection and the reliability of landslide deformation monitoring in areas lacking high coherence stability point targets, this study proposes an adaptive distributed scatterer interferometric synthetic aperture radar (ADS-InSAR) method based on the spatiotemporal coherence of the distributed scatterer (DS), which automatically adjusts its detection threshold to improve the spatial distribution density and reliability of DS detection in the landslide area. After time series network modeling and deformation calculation of the ADS target, the displacement deformation of the landslide area can be accurately extracted. Shuibuya Town in Enshi Prefecture, Hubei Province, China, was used as a case study, along with 18 Sentinal-1A images acquired from March 2016 to April 2017. The ADS-InSAR method was used to obtain regional deformation data. The deformation time series was combined with hydrometeorological and related data to analyze landslide deformation. The results show that the ADS-InSAR method can effectively improve the density of DS distribution, successfully detect existing ancient landslide groups and determine multiple potential landslide areas, enabling early warning for landslide hazards. This study verifies the reliability and accuracy of ADS-InSAR for landslide disaster prevention and mitigation.

Remote sensing-based ground fissure extraction techniques (e.g., image classification, image segmentation, feature extraction) are widely used to monitor geological hazards and large-scale artificial engineering projects such as bridges, dams, highways, and tunnels. However, conventional technologies cannot be applied in loess areas due to their complex terrain, diverse textural information, and diffuse ground target boundaries, leading to the extraction of many false ground fissure targets. To rapidly and accurately acquire ground fissures in the loess areas, this study proposes a data processing scheme to detect loess ground fissure spatial distributions using unmanned aerial vehicle (UAV) images. Firstly, the matched filter (MF) algorithm and the first-order derivative of the Gaussian (FDOG) algorithm were used for image convolution. A new method was then developed to generate the response matrices of the convolution with normalization, instead of the sensitivity correction parameter, which can effectively extract initial ground fissure candidates. Directions, the number of MF/FDOG templates, and the efficiency of the algorithm are comprehensively considerate to conclude the suitable scheme of parameters. The random forest (RF) algorithm was employed for the step of the image classification to create mask files for removing non-ground-fissure features. In the next step, the hit-or-miss transform algorithm and filtering algorithm in mathematical morphology is used to connect discontinuous ground fissures and remove pixel sets with areas much smaller than those of the ground fissures, resulting in a final binary ground fissure image. The experimental results demonstrate that the proposed scheme can adequately address the inability of conventional methods to accurately extract ground fissures due to plentiful edge information and diverse textures, thereby obtaining precise results of small ground fissures from high-resolution images of loess areas.

Combining observation well data and Interferometric Synthetic Aperture Radar (InSAR) deformation has shown great promise in groundwater levels monitoring and aquifer parameters estimation, which is crucial for groundwater management and related natural hazards anticipating. However, insufficient observation wells bring a drawback of this means in obtaining high spatial resolution results. Here, we employ a machine learning (ML) approach, the Optimizable Gaussian Regression Learner (OGPR), to integrate these two datasets to derive pixel-scale groundwater levels across Los Angeles and Orange County from June 2015 to November 2021. The results are validated with groundwater levels revealed from relative changes in seismic velocity (dv/v) measured at a dense network. With the innovative results, we found an interesting phenomenon that the groundwater levels show an overall rising trend during the study period, which is prominent in the San Gabriel Valley (SGV) and the Coastal Plain of Orange County (CPOC) basins with mean rate of 0.78 ± 0.04 and 0.31 ± 0.05 m/yr, respectively. Besides, the strongest seasonal groundwater level changes are observed in the CPOC with mean annual amplitude of 0.97 ± 0.16 mm, which results in seasonal ground deformation for 3.58 ± 0.51 mm. Moreover, by analyzing the long-term hydraulic head records in these two basins, we identify that the rising trends should be attributed to ∼ 6.5-year interannual variations. Our study underlines the importance of integrating multitemporal InSAR data with hydraulic head measurements to accurately quantify changes in groundwater levels, determine aquifer parameters, and assess variations in groundwater storage.

It is becoming increasingly vital to assess the risks of geological disasters associated with mining activities. However, most existing methods solely evaluate the deformation of individual points in mining areas without accounting for the overall deformation of an entire working face; in addition, if decorrelation occurs, it can degrade the interferometric time-series InSAR measurements of existing models. To address these disadvantages, we introduce a novel velocity–neighborhood spatial algorithm to accurately predict mining deformation trends and build upon the precise extraction of surface deformations from mining areas. Initially, surface deformation data from the Yuka mining area in Qinghai Province were obtained using the liquid cloud and surface based on small-baseline subset (LiCSBAS) method and transformed into a graphical structure. Subsequently, the temporal graph convolutional network (T-GCN) model was used to predict deformation trends, and the results show that it improved upon the accuracy of traditional recurrent neural networks and machine learning algorithms by 4% with a 15% reduction in root mean square error, thus proving its substantive advantages in spatiotemporal deformation trend predictions. The proposed method provides a more accurate and comprehensive scientific basis for monitoring and controlling geological disasters in mining areas while showcasing the vast application prospects of the T-GCN model.