Interferometric synthetic aperture radar for slow slip applications

Over the last two decades, Slow Slip Events (SSEs) have been observed across many subduction zones, primarily through continuous GNSS networks. SSEs represent shearing of two tectonic plates, at much slower rates than earthquakes but more rapidly than plate motion. They are not dangerous in themselves, but change the stress field and can potentially trigger devastating earthquakes. While highly valuable, GNSS networks at most locations lack the spatial-resolution required to describe the spatial extent of the slow slip at depth. A better constraint of slow slip at depth in combination with other observations from seismology could be essential in addressing key research questions. These include: “Why do slow slip events occurs in some regions and not others?”, “What drives slow slip events?”, “Do slow slip events delay the occurrence of devastating earthquakes?”, and “Can slow slip events trigger devastating earthquakes?”. Interferometric Synthetic Aperture Radar (InSAR) is an established and attractive technique to study surface displacements at high-spatial resolution. Until now, InSAR has not been fully exploited for the study of SSEs. Here, I provide the necessary InSAR methodology, and further demonstrate the use of InSAR for static and time-dependant slow slip modelling. My developments have a direct benefit for various other applications such as earthquake cycle processes. I Specifically address the following two challenges which limit the wide uptake of InSAR: (1) Decorrelation noise introduced by changing backscattering properties of the surface and a change in satellite acquisition geometry, making it difficult to correctly unwrap meaningful signal. I address this problem by applying existing advanced time-series InSAR processing methods. (2) Atmospheric delays masking the smaller slow slip signal. These are mainly due to spatial and temporal variations in pressure, temperature, and relative humidity in the lower part of the troposphere, which result in an apparent signal in the InSAR data. Different tropospheric correction methods exist, all with their own limitations. Auxiliary data methods often lack the spatial and temporal resolution, while the phase-based methods cannot account for a spatially-varying troposphere. In response, I develop a phase-based power-law representation of tropospheric delay that can be applied in the presence of deformation and which accounts for spatial variation of tropospheric properties. I demonstrate its application over Mexico, where it reduces tropospheric signals both locally (on average by ~0.45 cm for each kilometer of elevation) and the long wavelength components. Moreover, I provide to the research community a Toolbox for Reducing Atmospheric InSAR Noise (TRAIN), which includes all the state-of-the-art correction methods, implemented as opensource matlab routines. When comparing these methods, I find spectrometers give the largest reduction in tropospheric noise, but are limited to cloud-free and daylight acquisitions. I also find that all correction methods perform ~10-20% worse when there is cloud cover. As all methods have their own limitations, future efforts should aim at combining the different correction methods in an optimal manner. Additionally, I apply my InSAR methodology and power-law correction method to the study of the 2006 Guerrero SSE, where I jointly invert cumulative GNSS and InSAR SSE surface displacements. In Guerrero, SSEs have been observed in a “seismic gap”, where no earthquakes have occurred since 1911, accumulating a seismic potential of Mw 8.0-8.4. I find slow slip enters the seismogenic zone and the Guerrero Gap, with ~5 cm slip reaching depths as shallow as 12 km, and where the spatial extent of the slow slip collocates on the interface with a highly coupled inter-SSE region as found from an GNSS study. In addition, slow slip decreased the total accumulated moment since the previous SSE (4.7 years earlier) by ~50% Over time and while accounting for SSEs, the moment deficit in the Guerrero Gap increases each year by Mw ~6.8. Therefore I find that the Guerrero Gap still has the potential for a large earthquake, with a seismic potential of Mw ~8.15 accumulated over the last century. Finally, I show the application to use InSAR for time-dependant slow slip modelling. From a simulation of the 2006 SSE, I demonstrate that InSAR is able to provide valuable information to constrain the spatial extent of the slow slip signal. With a future perspective of continued high repeat acquisitions of various SAR platforms, my expansion of the Network Inversion Filter with InSAR will become a powerful tool for investigating the spatio-temporal correlation between slow slip and other phenomena such as non volcanic tremor. Moreover, this approach can apply to earthquake cycle processes. Studying the broader earthquake cycle will further our knowledge of seismic hazard and increase our resilience to such events.