Towards operational use of combined Split-Band Interferometry and Multidimensional Small Baseline Subset: application to geohazard monitoring in the Kivu region

The vulnerable region of Kivu is strongly affected by geohazards such as volcanism and landslides. An essential ingredient for modelling the geophysical processes from which these geohazards originate is the measure of the Earth’s surface displacements. In this complex region, monitoring these ground deformations can be efficiently performed by using Synthetic Aperture Radar Interferometry (InSAR). Nevertheless, geophysical modelling requires accurate measurements of the displacements over time, which are not always manageable with conventional InSAR due to area-related limitations such as geometric distortions. Consequently, we consider advanced processing techniques such as Split-Band Interferometry (SBInSAR), which potentially allows to calculate the absolute interferometric phase, and Multidimensional Small Baseline Subset (MSBAS), which reconstructs the history of deformations along the horizontal and vertical directions. The aims of this thesis are the investigation of the SBInSAR and MSBAS techniques for improving the ground deformation measurements, the development of these techniques in order to achieve an operational tool of monitoring, and eventually to combine the information provided by both methods. Regarding the Split-Band Interferometry, we consider its theoretical applicability, determine optimum processing parameters and propose a method that corrects phase ambiguities by using SBInSAR as a complement to classical phase unwrapping, which we call SBInSAR-assisted phase unwrapping. We establish that the best criterion for the detection of spectrally stable pixels is the slope standard deviation. We also study the temporal behaviour of spectrally stable pixels and show that few of them persist over time. We analyze the behaviour of artificial reflectors with respect to SBInSAR and provide strong evidences that there exists a correlation between the signal-to-clutter ratio of a target and its detectability as a spectrally stable pixel. Regarding the Multidimensional Small Baseline Subset, we present the InSAR mass processing chain developed in collaboration with the European Center for Geodynamics and Seismicity for MSBAS preprocessing, and we propose to optimize the spatial coverage of MSBAS measurements by using interpolation or global masking. An MSBAS analysis is carried out over the city of Bukavu for the period from January 2016 to September 2017. We also study temporal decorrelation over the Virunga Volcanic Province and we model it using a decreasing exponential function. As a complement, we implement the Range Split-Spectrum technique in the CIS software for the ionospheric phase compensation in low frequency interferograms. We finally discuss the options for merging SBInSAR and MSBAS together.