This study introduces the S1S2-Water dataset – a global reference dataset for training, validation and testing of convolutional neural networks for semantic segmentation of surface water bodies in publicly available Sentinel-1 and Sentinel-2 satellite images. The dataset consists of 65 triplets of Sentinel-1 and Sentinel-2 images with quality checked binary water mask. Samples are drawn globally on the basis of the Sentinel-2 tile-grid (100 x 100 km) under consideration of predominant landcover and availability of water bodies. Each sample is complemented with metadata and Digital Elevation Model (DEM) raster from the Copernicus DEM. On the basis of this dataset we carry out performance evaluation of convolutional neural network architectures to segment surface water bodies from Sentinel-1 and Sentinel-2 images. We specifically evaluate the influence of image bands, elevation features and data augmentation on the segmentation performance and identify best-performing baseline-models. The model for Sentinel-1 achieves an Intersection Over Union of 0.85, Precision of 0.93 and Recall of 0.90 on the test data. For Sentinel-2 the best model produces an Intersection Over Union of 0.96, Precision of 0.99 and Recall of 0.97 respectively. We also evaluate the performance impact when a model is trained on permanent water data and applied to independent test scenes of floods.

On 17 June 2020, an ancient landslide was partially reactivated close to the Aniangzhai village of Danba County in Sichuan Province of Southwest China. It was initiated by the erosion of the slope toe from the overflow of a dammed lake that was created due to heavy rainfall and the resulting debris flows coming from Meilonggou Gully to the Xiaojinchuan River. In this study, we report investigations on precursory and post-failure slope stability analysis exploiting optical and radar satellite remote sensing data. Using sub-pixel cross-correlation of optical data from Planet and Sentinel-2, we first derive the direction and magnitude of the main landslide failure. Advanced multi-temporal InSAR (MT-InSAR) analysis using Sentinel-1 and TerraSAR-X SAR data are then exploited to investigate the landslide kinematics before and after the big failure. Moreover, we report our experience on using a newly designed artificial corner reflector (CR), which is a half-round dihedral corner reflector (hr-DCR), for monitoring slope inability in this region using both ascending and descending SAR data. The CRs are quite useful auxiliaries for InSAR analysis as they could be recognized as stable targets during radar acquisitions, especially in the vegetated, semi-vegetated, or agricultural areas, where the widespread loss of coherence between consecutive image acquisitions could happen. Using MT-InSAR analysis, we observe precursory deformation amounting to approximately 50 mm/year in the year 2018-2020, reaching to a maximum of 270 mm/year for the post-failure period from Nov 2020 to June 2021. Before the main landslide failure in June 2020, the average deformation rate was approximately 14% higher in 2018-2020, dominated by above-average precipitation in summer, in comparison to the rate in 2014-2017. Interestingly, MTI analysis also detects a clear signal for the new instability and slow creep in the adjacent slope of the Aniangzhai ancient landslide, previously unrecognised in landslide inventory maps. Besides, the performance of newly designed DCRs is qualified and quantified in the experiments based on intensity time series (in dB), Signal-to-Cluster Ratio (SCR), and results from MTI time series.

We use interferometric synthetic aperture radar (InSAR) observations to investigate the fault model and afterslip evolution within 3 years after the 2017 Sarpol-e Zahab, Iran, Mw 7.3 earthquake. The anti-listric fault which is very similar to flat-and-ramp structure inverted by kinematic afterslip models is proposed to simulate the coseismic slip and afterslip evolution. Compared with listric faults, linear inversions demonstrate that a planar fault can explain coseismic deformation well enough. However, the stress perturbations caused by this basement-involved faulting propagated upward to the sedimentary cover. The transition of sedimentary cover-basement interface inferred by afterslip models is at the depth of ~13 km in the seismogenic zone, which coincides with the regional stratigraphic profile and indicates that the significant afterslip updip of the coseismic rupture is mainly controlled by frictional property. We additionally find the postseismic deformation is dominated by afterslip while the viscoelastic response is negligible with the best-fitting viscosity which is on the order of 1019Pa s. Compared to the best-fitting kinematic afterslip model, the stress-driven afterslip model tends to underestimate early postseismic deformation to the west, which may indicate the spatial heterogeneity of the frictional property of fault plane. Because the coseismic rupture propagated along a basement-involved fault while the postseismic slip was likely to activate the frontal structures and/or shallower detachments in the sedimentary cover, the 2017 Sarpol-e Zahab earthquake may act as a typical event which contributes to both of the thick- and thin-skinned shortening of the Zagros in both seismic and aseismic way.

The coal fires that started over a century ago in Jharia Coal Fields constitute a significant threat to the coal reserves, infrastructure, and residents’ lives. The fires burn underground coal leaving the surface with no support, leading to land subsidence and roof collapse. This will have a multiplier effect as it creates cracks and crevices that pump in more oxygen to aggravate the coal fires further. Despite the various measures taken by authorities, coal fires and land subsidence still have an increasing presence. In this study, we investigated the two hazards and their impact on the coal mines and surrounding settlements. We observed the subsidence and coal fires in the study area with the help of Persistent Scatterer Interferometry analysis of Sentinel-1 images and Temperature anomaly mapping of Thermal Infrared Imagery from Landsat-8, respectively. The subsidence velocity results and the coal fire zones are analysed, and a significant spatial overlap of both hazards is noticed. A few key locations severely affected by both the hazards are identified and examined to understand the mutual effect of coal fires and land subsidence. The subsidence of up to 20 cm/yr is observed in the study area. The results show that nearly 80% of the subsiding area is also affected by coal fires. Kusunda, Bararee and Keshalpur collieries are critically affected by both the hazards and need immediate intervention. Subsidence and coal fires extending towards the residential zones in several collieries is a matter of concern. In conclusion, the study presents an efficient methodology for multi-hazard monitoring, and the findings assist the authorities in enforcing appropriate disaster management strategies.