Multi-scale spatiotemporal inversion filter of a dense GNSS observations network: Application to the deformation anomalies detection related to the 2021 Yangbi earthquake of Ms 6.4

Summary

The Global Navigation Satellite System (GNSS) now makes it possible to monitor crustal deformation with unprecedented accuracy and spatial coverage. However, transient tectonic signals are obscured in raw GNSS time series due to complex triggering mechanisms and a low signal-to-noise ratio. We here propose a multi-scale spatiotemporal inversion filter (MSIF) method based on dense GNSS network. First, multi-scale spatial decomposition of the regional GNSS deformation field is performed to identify anomalous strain accumulation zones. Then, continuous GNSS observations from targeted regional stations are incorporated to extract transient fault slip signals and constrain their spatiotemporal evolution by the variational Bayesian Independent Component Analysis-based inversion approach. Numerical simulations are conducted to verify the minimum GNSS network density required for detecting Mw 5.5-level slip events and evaluate the fault slip detection capability of the current GNSS observation network. We further apply this method to continuous GNSS time series associated with the 2021 Ms 6.4 Yangbi earthquake.

Distinct pre-seismic deformation anomalies are identified during the preparatory stage of the earthquake, with strain accumulation concentrated at a small spatial scale of approximately 11 km. The inversion results reveal that the rupture zones of foreshocks and the mainshock are dominated by extensional motion and right-lateral strike-slip. The corresponding released moment magnitude reaches Mw 5.5, and the induced Coulomb stress changes amount to 0.021 MPa and 0.036 MPa in the hypocentral regions of the foreshock and mainshock, respectively. These results demonstrate that the detected pre-seismic slip anomaly may have facilitated the occurrence of subsequent foreshocks and mainshock, further verifying that aseismic creep and stress perturbation-induced foreshocks can coexist and interact during earthquake nucleation.