A Multivariate Time Series Analysis of Ground Deformation Using Persistent Scatterer Interferometry

Abstract

Ground deformations in urban areas can be the result of a combination of multiple factors and pose several hazards to infrastructures and human lives. In order to monitor these phenomena, Interferometric Synthetic Aperture Radar (InSAR) techniques are applied. The obtained signals record the overlapping of the phenomena, and their separation is a relevant issue. In this framework, we explored a new multi-method approach based on the combination of Principal Component Analysis (PCA), Independent Component Analysis (ICA) and Hierarchal Clustering (HC) on the standardized results to distinguish the main trends and seasonal signals embedded in the time series of ground displacements, to understand spatial-temporal patterns, to correlate ground deformation phenomena with geological and anthropogenic factors, and to recognize the specific footprints of different ground deformation phenomena. This method allows us to classify the ground deformations at the site scale in the metropolitan area of Naples, which is affected by uplift cycles, subsidence, cavity instabilities and sinkholes. At the local scale, the results allow a kinematic classification using the extracted components and considering the effect of the radius of influence generated by each cavity, as it is performed from a theoretical point of view when the draw angle is considered. According to the results, among the classified cavities, 2% were assigned to subsidence and 11% to uplift kinematics, while the remaining were found to be stable. Furthermore, our results show that the centering of the Spatial-PCA (S-PCA) is representative of the region’s main trend, whereas Temporal-PCA (T-PCA) gives information about the displacement rates identified by each component.

Keywords

• InSAR
• Ground deformation
• Subsidence
• PCA
• ICA
• Hierarchical Clustering