Deriving centimeter-level coseismic deformations and source parameters of small-to-moderate earthquakes from time-series Sentinel-1 SAR images

Small-to-moderate size earthquakes occur much more frequently than large ones but are general less studied by InSAR, despite that they also provide critical information about the physics of faulting and earthquake mechanisms. The weak coseismic deformations contaminated by severe atmosphere turbulences make them difficult to be studied by single interferogram. Since the launchings on April, 2014 and April, 2016, Sentinel-1 A/B satellites began to provide large-coverage SAR images in short revisited period (6 or 12 days) with 250 km frame width. The high-temporal sampling rate of Sentinel-1 data produce sufficient images for the stacking process to greatly reduce the local atmospheric turbulence that is difficult to be handled by numerical weather models. This procedure allows the extraction of very weak coseismic deformation (i.e. sub-centimeter) for small-to-moderate size earthquakes and systematical static slip inversions of the earthquakes in a tectonically active region by InSAR.

 

Here we report this stacking method and a new downsampling strategy based on quadtree mesh obtained from preliminary slip model to efficiently reduce the number of unwanted data points. Applying the proposed methods, we successfully retrieve coseismic deformations for 33 earthquakes (Mw4.1-Mw6.6) occurred in west China from Nov, 2014 to Jul 2020. Among these earthquakes, the smallest peak Line-of-Sight coseismic deformation is only ~6 mm. These InSAR-based earthquake catalogs show robust and precise absolute location (latitude, longitude and depth), therefore can be used as benchmark events to calibrate seismic based catalogues. However, strong trade-offs between earthquake source parameters (e.g. fault size vs slip) exist when the earthquake magnitude is small (in general smaller than Mw5.5). Such trade-offs are rooted due to the smaller deformation gradient in comparison with larger earthquakes. For the moderate size earthquakes (Mw6.0-6.6), the comparison between equivalent moment tensor from InSAR slip models and GCMT/W-phase solutions show that large CLVD components, as shown in the seismic-based moment tensor solutions, are mostly not necessary to explain the InSAR data. We suggest to combine geodetic and seismic datasets for more comprehensive and accurate earthquake source parameter inversions.