Analysis and identification of geohazards using multi-source remote sensing data

Geohazards can result in substantial losses, yet human-induced geohazards are predictable, monitorable and manageable, particularly for land subsidence triggered by ground extraction and mining hazards. Over the past decade, earth observation has incorporated a diverse array of sensors and platforms, empowering us to monitor and identify these human-induced geohazards. The integration of multi-source remote sensing data can detect more different types of geohazards, can effectively improve the accuracy and efficiency for identifying geohazards, can improve the reliability of geohazard mapping, can support the classification of geohazards, and can be used for geohazard risk analysis and decision-making suggestions for the implementation of mitigation strategies. The temporal analysis of multi-source remote sensing data over time allows us to identify dynamic alterations in high-risk regions, yielding valuable insights into the evolution and behavior of geohazards. Consequently, the objective of this thesis is to identify, monitor and classify human-induced geohazards, using multi-source remote sensing data. Previous research primarily emphasizes the assessment of land subsidence using InSAR and GNSS remote sensing methods. The utilization of InSAR, including factors such as temporal decorrelation, geometrical decorrelation and atmospheric artifacts, imposes several constraints when conducting comprehensive assessment of land subsidence. It is important to highlight that in prior research, aerial LiDAR datasets have been employed as supplementary data for InSAR processing or to detect changes in small areas. Therefore, the primary innovation of this research lies in the use of a multi-temporal open-access and non-customized (i.e., non-specifically acquired for this study) point cloud datasets to assess vertical deformations resulting from land subsidence at wide-spread scale. This approach successfully addresses the challenges associated with utilizing extensive LiDAR data archives. Firstly, we developed a novel methodology for point cloud differencing to identify areas affected by land subsidence on a basin scale, which is founded on the multiscale model-to-model cloud comparison (M3C2) algorithm. To this aim, the iterative closest point (ICP) algorithm is employed for registration of both point clouds, demonstrating a highly stable and robust performance. Furthermore, a method that integrates gradient filtering and cloth simulation filtering (CSF) algorithm was utilized to eliminate non-ground points. The discrepancies in the internal edge connection across various flight lines were rectified using a method that smoothens the point cloud. The accuracy of the maps of changes derived from LiDAR was evaluated by comparing them with the rate recorded by continuous GNSS stations and an InSAR dataset. Additionally, a comparison was made between LiDAR outcomes and the distribution of compressible soil thickness, which revealed a distinct correlation. Lastly, we analyzed the uncertainty associated with directly processing point cloud to create a cumulative map of land subsidence, and we explored the methodological benefits of working directly with point clouds, particularly when compared with InSAR. Active deformation areas (ADAs) present a substantial risk in mining regions due to their significant potential to trigger slope failures, which typically impact open pits and waste and tailing disposal facilities. Previous studies have explored various aspects related to active deformation areas, including stability assessment of disused open-pit mines and waste dumps, mapping ground movements and analyzing mining subsidence in specific regions of the Sierra de Cartagena-La Union. This dissertation not only enhances the existing ADA inventories for the entire Sierra de Cartagena-La Union but also introduces a comprehensive systematic approach for the automatically mapping and preliminary supervised categorization of the ADAs that impact the region. The central concepts of this paper introduce a methodological approach that leverages the combined strengths of LiDAR and InSAR remote sensing techniques for identifying active deformation regions within mining areas. The suggested approach not only facilitates the detection of landslides but also allows for the mapping and initial categorization of additional phenomena commonly observed in mining regions, such as consolidation of waste dumps, earthworks, subsidence and erosion. Consequently, the utilization of the LiDAR technique serves as a complementary method for updating active areas in this study. Additionally, we demonstrate the potential of utilizing InSAR technique from satellite and LiDAR techniques from the airborne, in conjunction with data derived from a basic slope stability geotechnical model, to create and update inventory maps that track active deformations in mining regions. To achieve this, we processed imagery from Sentinel-1 that was captured in ascending and descending orbits. The displacements were then broken down into vertical and east-west components. In addition, LiDAR point clouds, which are open-access and non-customized, were employed to examine changes and movements on the ground surface. Moreover, the infinite slope stability modelling approaches offer a practical and easy way to create a map of the slope stability safety factor (SF) in the study area, considering various depths of the slip surface (D) and soil-phreatic level ratio (Hw) values. The results provided by this limit equilibrium method serve as a valuable tool for the identification and categorization of ADAs placed in areas that are susceptible to landslides. Finally, the maps derived from InSAR, LiDAR, and the limit equilibrium method were combined to update an existing inventory map of landslides and to carry out a preliminary classification of various active deformation areas, aided by optical images and a geological map. Additionally, an activity level index was established to verify the reliability of the identified ADAs. Finally, the key characteristics of various techniques and the inevitable uncertainties encountered while identifying and categorizing ADAs were discussed. Synthetic Aperture Radar Interferometry (InSAR) monitoring of mining subsidence dynamics allows for remote exploration and analysis of ground surface deformations resulting from underground resource extraction. These changes result from a combination of natural geological environmental factors and human activities. Properly managing the direction of mining operations and identifying deviations from original production plan are crucial for maintaining the quality of the mining exploitation process and for detecting cases of unauthorized mining. Likewise, the face advance rate plays a pivotal role in planning and optimization of mining operations. These parameters directly influence productivity, safety, and project schedules. Furthermore, surface deformation can also have implications for essential infrastructure, posing risks and disruptions to everyday life. Monitoring underground mining without prior knowledge of mining conditions and working operation parameters presents a significant challenging. In this context, we introduce an approach to extract parameters related to mining activities' dynamics, which can enhance the exploitation process and aid authorities in regulating underground operations. Notably, Chinese LuTan-1 (LT-1) mission represents L-band bistatic spaceborne SAR mission for civil applications. LT-1 significantly improves global SAR data availability and provides real-time dynamic monitoring of subsidence caused by underground mining activities. Therefore, in this dissertation various InSAR datasets and technology with additional data have been combined to examine geohazard and the associated damage. LT-1 offers continuous imagery for the analysis of ground surface deformations via differential SAR interferometry, and it also provides valuable velocity results. The findings of subsidence bowl obtained from LT-1 were found to be in good concordance with the outcomes derived from Sentinel-1 in the same period. Additionally, the DInSAR outcomes obtained from LT-1 and Sentinel-1 were confirmed with data from continuous GNSS stations from the same time frame. In conclusion, we monitored the dynamics of surface deformation related to underground mining activities by utilizing DInSAR outcomes from four distinct time periods. These DInSAR findings highlights considerable alterations in the form and geographical position of the subsidence bowls over time. Finally, the velocity and the gradient maps, which were obtained from DInSAR outcomes, were superimposed on the layout of vital infrastructures in the study areas. Finally, the exposure of critical infrastructures in different classes to assess their susceptibility to mining subsidence was analyzed, which was validated using GF-7 satellite high-resolution optical images. In conclusion, this dissertation emphasizes the capability of open-access and non-customized LiDAR in tracking spread and magnitude of vertical deformations in regions susceptible to land subsidence induced by groundwater withdrawal. Additionally, the findings underscore the efficiency of these two remote sensing methods, namely InSAR and LiDAR, when used together with basic geotechnical models, supported by orthophotos and geological data, to update inventory maps of ADAs in mining regions. Furthermore, this dissertation also opens a new perspective to demonstrate the ability of Chinese satellite in monitoring surface deformation dynamics and analyzing the potential damage caused by mining subsidence.

Author

• Hu, Liuru

• Multi-source remote sensing data
• InSAR
• LiDAR
• Human-induced geohazards
• LT-1

Identifier

hdl:10045/153937

OAI identifier oai:identifier

oai:rua.ua.es:10045/153937