Search

In the field of multitemporal interferometric synthetic aperture radar (MT-InSAR), the selection of persistent scatterer (PS) is crucial for acquiring ground deformation product. To obtain precise ground deformation, pixels with as high signal-to-noise ratio (SNR) as possible should be selected, while pixels with low SNR should be avoided. To this end, we propose a novel framework, referred to as the PS feature-fusion network (PSFNet), for efficient PS selection. Specifically, we propose a data-driven two-branch network consisting of a ResUNet with spatial and channel attention, as well as a TANet with 3-D convolutional layers and a time-step attention block (T-Attention block), which can use not only spatial features of SAR image but also time-series phase features when selecting PS pixels. In particular, a time-step attention mechanism is proposed for accommodating to interferometric pairs with different SNRs to enhance the feature representation ability of the network. The proposed method was tested using the Sentinel-1 images, showing that it can select more PSs with higher quality compared with StaMPS. In addition, the prediction time of PSFNet requires only 0.26% of the running time of StaMPS, which greatly improves the efficiency of PSFNet for practical applications.

The analysis of kinematic characteristics and triggering factors of the potentially unstable slopes is of great significance for the slope protection and landslide reinforcing. Located in Mao Country, Sichuan Province in China, Wadi village is a typical mountainous area where geological disasters frequently occur due to the complex geological environment, intense tectonic activities, and concentrated seasonal rainfall. However, the slope stability and potential risk over this area are not fully evaluated yet. In this article, we investigate the unstable slope in Wadi village using time-series synthetic aperture radar interferometry (InSAR) technique. The deformation history retrieved from 132 C-band descending Sentinel-1 images between October 2014 and September 2019 demonstrates that the Wadi landslide presents a slow sliding trend with the averaged line-of-sight velocity of −70 mm/year and has a distinct seasonal velocity at the head and toe area. We further decompose the predominant and periodic velocities by a kinematic model to analyses spatial and temporal characteristics of Wadi landslide. The results reveal that the Wadi landslide is a hybrid type of push and pull with two source areas in the head and toe of the slope, respectively. The periodic velocity variations are highly correlated with seasonal rainfall. Our study also demonstrates the importance of correcting InSAR unwrapping error for interpreting the landslide kinematics and other small-scale geomorphological processes.

Very high requirement of co-registration accuracy better than 0.001 pixels for Sentinel-1 TOPS (Terrain Observation by Progressive Scan) mode presents a great challenge for application of SAR (synthetic aperture radar) interferometry in Earth Observation. The state-of-the-art techniques have demonstrated that the low coherence scenarios and abrupt loss of coherence between two arbitrary acquisitions are main sources of error to degrade the performance of TOPS time-series co-registration. In this paper, we present a modification to overcome both limitations through the coherence enhancement. The motive behind this is to improve the quality of observations before co-registration and meanwhile avoiding the coherence loss caused by fast decorrelation. To this end, principal components analysis based spatio-temporal filtering is first used to remove the artifacts in burst interferograms over strong noise areas. Rather than heuristically choosing a sub-set of interferograms as a small baseline technique does, we use Dijkstra's shortest path algorithm under graph theory framework to maximize the coherence of a sub-set interferograms. The performance of presented method against the state-of-the-art techniques is fully evaluated by synthetic data and a Sentinel-1A stack over a low coherence scene in Indonesia. Comprehensive comparisons demonstrate 9%-17% uncertainty reduction of time-series co-registration when applying our method.