Application Of High-rate GPS For Earthquake Early Warning And Rapid Modeling

Earthquakes of different scales occur frequently every year worldwide,posing a huge threat to life and property safety.Earthquake prediction and forecasting are difficult to solve in the short term.In this case,earthquake early warning and emergency response are particularly important.This depends on fast and reliable analysis and modeling of the earthquake source immediately after the rupture initiation.Current work mainly includes real-time determination of moment magnitude and rapid inversion for seismic moment tensor and rupture process.For example,the Global Centroid Moment Tensor Catalogue(GCMT)Project and the U.S.Geological Survey(USGS)provide centroid moment tensors and fault plane solutions for earthquakes of different scales worldwide.However,they rely on broadband teleseismic data.Since most teleseismic stations are located at epicentral distances of 300 to 90°,the average P-wave arrival time is?10 min,which can be considered as the efficiency limit of inversion of teleseismic data,lagging the emergency response and output after the earthquake.Near-field broadband seismic data are usually affected by clip and station tilt during strong earthquakes.Although strong motion data are not saturated,they suffer from baseline drift.High-rate Global Positioning System(GPS)directly records the surface displacement,which effectively avoids problems such as clip,station tilt and rotation,and errors caused by data integration.With its unique advantages,highrate GPS has great potential for application in seismological research,which has led to the emergence of an interdisciplinary subject,namely,GPS seismology.This study focuses on the applicability,timeliness,and reliability of high-rate GPS in earthquake early warning and rapid response.Taking the 2019 Mw 7.1 Ridgecrest,California earthquake as an example,the main work is as follows:(1)In the Introduction Section(Chapter 1),the progress of high-rate GPS seismology are reviewed,including the application of high-rate GPS for co-seismic surface displacement monitoring,earthquake early warning,and inversion for seismic moment tensor and source process,etc.At present,most research on earthquake early warning and rapid response rely on the seismic network,but the mentioned inherent disadvantages of broadband seismograms and strong motion data make it restrictively applicable for earthquake hazard mitigation.Therefore,the objectives of this study are put forward.I quantify the timeliness of earthquake magnitude estimate,centroid moment tensor solution,and kinematic rupture model based on high-rate GPS observations.(2)In Chapter 2,I introduce the basic theory and method of high-rate GPS positioning,and obtain displacement waveforms.The peak ground displacements(PGD)observed by high-rate GPS are used to estimate the warning magnitude of the 2019 Ridgecrest earthquake.The influence of different regression coefficients of PGD scaling law and the number and distribution of stations used on the warning magnitude calculation are analyzed.Stations located on both sides of the fault are beneficial to report a robust estimate.Taking the timeliness and accuracy of earthquake early warning into account,a four-station PGD warning will be issued at 24 s after the earthquake onset,with warning magnitude of Mw 7.03,which is consistent with the Wphase solution reported by the USGS.(3)In Chapter 3,I describe the theory of seismic moment tensor inversion,including characterizing earthquake source using the moment tensor,and its decomposition and inversion.The results of real-time inversion for moment tensor of the 2019 Mw 7.1 Ridgecrest earthquake using high-rate GPS data are given,indicating that nodal plane solutions have been initially recovered at 30 s after the origin time,with a strike-slip mechanism of the event already resolved;by 50 s,the moment magnitude reaches Mw 7.06,which is consistent with the W-phase solution published by the USGS,and the centroid location coincides with the GCMT result.In addition,I supplement the analysis with the available unclipped broadband seismograms.The results of post-event inversion for focal mechanism solutions are in good agreement with those of real-time inversion,ensuring the reliability of rapid inversion.(4)In Chapter 4,I introduce the basic principles and methods of inversion for finite fault slip distribution and rupture process.Based on the fault plane solution derived from real-time moment tensor obtained in Chapter 3,I further image the time-dependent slip history of the 2019 Mw 7.1 Ridgecrest earthquake.The variance reduction results corresponding to two nodal plane solutions suggest that nodal plane 1(with strike/dip/rake of 322°/60°/-169°)is the seismogenic fault plane.The results of rapid modeling of the source process are in good agreement with published results.This earthquake is a predominant dextral strike-slip event,with both extension and thrust components resolved as well,indicating faulting complexity.The source duration is?28 s.The total scalar seismic moment is 5.18×1019 N·m(Mw 7.11),most of which are released within 10 s.The maximum slip is?4 m.The main asperity extends?50 km in a region with clear surface rupture,with most slip near the hypocenter and concentrated between 0?10 km depths.The slow average rupture velocity(?2.1 km/s)indicates a young,immature fault system,which is likely governed by fault geometrical complexity,strength and its roughness.Jackknife sensitivity test shows that the kinematic source model obtained in this study is robust and reliable,and is not sensitive to data distribution.I also discuss the quasi-real-time inversion for the source process.Based on the fast moment tensor solution recovered within 30 s after the rupture initiation,I cut 30-second waveform recordings from five nearest stations and run the inversion.A similar rupture model which matches key characteristics with that derived from full waveforms is obtained.To supplement the analysis,I also conduct a retrospective simulated real-time inversion for the rupture process of the 2016 Mw 7.8 Kaikoura earthquake and the 2010 Mw 7.2 El Mayor-Cucapah earthquake.A reliable rupture model is available at 64 s and 100 s,respectively,after the earthquake occurrence.Thus,I quantify the timeliness of a reasonable source model,that is,a reliable rupture model of the 2019 Ridgecrest earthquake,the 2016 Kaikoura earthquake,and the 2010 El May or-Cucapah earthquake can be obtained within 30 s,64 s,and 100 s,respectively,after the origin time.This study highlights the scientific and practical significance of high-rate GPS for earthquake emergency response and hazard mitigation.