Southeastern (SE) Tibet is one of the most seismically active regions in mainland China, but the spatial distribution of interseismic coupling that quantifies seismic hazard is unknown along most major faults. In this study, we constructed an elastic block model to invert Global Positioning System data for slip rates and locking coefficients along 20 major faults in SE Tibet. Our results identify 27 strongly coupled fault segments with locking coefficients >0.5, defined as potential seismogenic asperities, extending laterally for 36–330 km. Quantitative calculations of seismic moment budgets on these seismogenic asperities indicate that they are capable of generating Mw 6.4–7.7 earthquakes in the next few decades, of which the Anninghe, Daliangshan and Red River faults have the potential for Mw ≥ 7.5 earthquakes. The interseismic coupling model provides a component for probabilistic analysis of future seismic hazards in densely populated Southwest China.

On 21 May 2021, an Mw 7.4 earthquake occurred in Maduo County, Qinghai Province, China. A dense network of high-rate (1 Hz) Global Navigation Satellite System (GNSS) stations around the epicenter provided one of the most complete recordings of GNSS kinematic displacements in Tibet to date. We used these data retrospectively to test a prototype earthquake early warning (EEW) system. Here, we present the results of that test, in which the EEW system archived high-rate GNSS displacement streams, and then used them to track evolving magnitude and slip. The geodetic module in the EEW system issued the first alert 34 s after the earthquake onset. The first alert had a moment magnitude (Mw) of 6.65, which then increased until reaching a stable value (Mw 7.4) 61 s after the earthquake onset. Testing results show good agreement with rapid GNSS displacements and postprocessed GNSS displacements; moreover, the rapid finite-fault inversion is consistent with postinversion magnitude and rupture dimensions. Our findings confirm the viability and usefulness of this EEW system in quasi-real-time magnitude estimation and finite-fault slip inversion.