Constraining Source Parameters With Sparse Network

Accurate source mechanism,depth and horizontal location of small to middle size earthquakes and reasonable finite fault solutions of meca-earthquakes is not only the basis topic of seismology but also important in earthquake mitigation.Seismogram of earthquakes are composed by source and structure information,theoretically,people can obtain accurate mechanism of earthquakes only when we have a better understanding of structure and enough station coverage.However,there are a few regions in the world have dense station coverage and reliable 3D model.Thus it is necessary to develop methods to obtain accurate source mechanism of earthquakes in regions with sparse network and poor understanding of 3D structure.Base on this background,we test and develop the Cut-And-Paste(CAP) method which inverses source mechanism by waveform modeling and is relative insensitive to 3D structure in following aspects:(1) We test stability and reliability of obtaining source parameters by sparse network and present a crustal tomography model of Southern California which is appropriate for sparse network full parameter inversion;(2) We use different 1D models to approximate 3D structure in different azimuth;(3) We achieve high accuracy source centroid location by combining Ambient Seismic Noise (ASN) technique which can extract Greens' Function of surface wave by cross-correlating noise records and CAPloc method;(4) We develop CAPtele method which can inverse for source parameters basing on teleseismic body waves by improving CAP method,and make sensitivity tests for mantle attenuation factor t*, length of source time function and types of record;(5) We develop CAPjoint method which can use both local and teleseismic records for source parameter inversion;(6) We make sensitivity tests for t*,dip angle,rupture velocity,and maximum rupture depth for finite fault inversion of teleseismic body waves.In chapter 1,we used broadband teleseismic data of ChingFeng station to calculate the Receiver Functions,then obtained a velocity model with H-K stacking method in combination with Crust2.0 and other previous work.With this velocity model and broadband records from 5 CDSN stations,we inverted focal mechanisms of ChiFeng earthquake on August 16th,2003 with the "Cut and Paste"(CAP) method. Then we confirmed the focal depth and source mechanisms by comparing synthetic teleseismic P waves at with broadband records of 9 IRIS stations.Our result shows that the best double couple solution of this Mw5.2 event is 315°,64°and 19°for strike,dip and slip angles respectively,the second nodal plane solution is 214°,74°, and 152°.The focal depth is 25±2km,suggesting that this quake occurred in the lower crust which is much deeper than most continental earthquakes.This lower crust earthquake requires that the rock should be colder than expected.We proposed generation mechanism of this deep earthquake and its implications in rock strength and thermal state.In chapter 2,we first determine source mechanism of YunNan earthquake by local broadband data,and then compare it with teleseismic inversion result.Tests of record type(displacement or velocity),source duration,attenuation factors of P and S body waves are made for sensitivity study of moment magnitude.After weighting regional data and teleseismic data properly according to their amplitudes,they can be used together in a Cut and Past(CAP) inversion process.Fault geometry(strike,dip and rake),moment,depth and duration are determined in a grid search manner.At the end we discuss the distributed slip of this event.In chapter 3,we conduct a detailed test of a recently developed technique, CAPloc,in recovering source parameters from a few stations against results from a large broadband network in Southern California.The method uses a library of 1D Green's functions which are broken into segments and matched to waveform observations with adjustable timing shifts.These shifts can be established by calibration against a distribution of well-located earthquakes and assembled in tomographic images for predicting various phase-delays.Synthetics generated from 2D cross-sections through these models indicates that 1D synthetic waveforms are sufficient in modeling but simply shifted in time for most hard-rock sites.This simplification allows the source inversion for both mechanism and location to easily obtain by grid search.We test one-station mechanisms for 160 events against the array for both PAS and GSC which have data since 1960.While individual solutions work well for mechanism(about 80%),joint solutions using these two stations produce more reliable and defensible results.Inverting for both mechanism and location also works well except for certain complex paths across deep basins and along mountain ridges.Recent waveform modeling methods have been developed to retrieve local Green's functions based on the cross-correlation of ambient seismic noise(ASN) involving station-to-station and convention(source-to-station) inversions.The latter methods provide the most broadband results but require the separation of the source description from the 3D structure.Several new methods overcome this trade-off by iteration involving back-projections,i.e.,the adjoint method.An alternative approach, the cut-and-paste(CAP) technique,allows adjustments in timing between seismic phases(path corrections) making it possible to match waveform data and recover source parameters.These shifts or delays can be estimated from previous earthquakes, phase delay mapping,or they can be generated directly from CAP analysis of 3D synthetics.In contrast,1D Green's functions generated by ASN does not have this inherit source location-origin time issue and can provide excellent independent delay calibrations.Here,we compare the source parameters including location for the recent Chino Hills earthquake(CA) derived from these three methods.These advances make it possible to locate Centroids of local events in near real time.In chapter 6,we inversed source parameters of 2007/06/02 Nin'er Mw6.1 earthquake in Yunnan Province of China.At first,we use four local broadband records of China Digital Seismic Network(CDSN) to inverse for source mechanism of the main event by CAP method,we obtain:Mw=6.0,depth=5.0km,the first nodal plane I is 150°/75°/140°for strike/dip/rake and the second is:252°/52°/40°.And then we use 27 pairs of P/SH teleseismic body waves to make tests for our CAPtele method,which shows that the best t_p^*=1,t_sh^*/t_p^*=5 and length of source time function is 3.5s.Base on these parameters we inverse source mechanism by CAPtele method,in which we use velocity records and corresponding source parameters are:Mw=6.2,depth=2km, nodal planeⅠis:155°/59°/147°,nodal planeⅡis:258°/70°/33°.When we use displacement records during inversion,corresponding source mechanism is: Mw=6.26,depth=2km,nodal planeⅠis:146°/49°/134°,nodal planeⅡis: 259°/66°/46°.CAPtele inversions has another local minimum solution at 5km while local CAP inversion has only one minimum solution in depth.We develop CAPjoint method by combining local and teleseismic record to eliminate this ambiguity,and we have the following joint mechanism:Mw=6.1,depth=6km,nodal planeⅠis: 149°/65°/151°,nodal planeⅡis:254°/60°/29°.Finally,we conduct finite fault inversion on these two fault planes,and our prliminary results show that nodal planeⅠ(149°/65°) might be the real fault plane which accords with intensity distribution. However,we need to combine more data for detailed study of finite fault soluion.In chapter 7,we present a series of rupture models of the Wenchuan earthquake(Mw7.9),based on inverting P and SH teleseismic body waves.A simulated annealing algorithm is used to determine the finite-fault model that minimizes the objective function described in terms of wavelet coefficients.With this approach,we can simultaneously invert for the slip amplitude,slip direction,rise time and rupture velocity.At the first step,experiments conducted on synthetic data are used to assess the ability to recover rupture slip details and teleseismic body waves. We investigate the resolvability of teleseismic body inversion on dip angle,rupture velocity and maximum rupture depth.Then 4 slip models are obtained for the Wenchuan Earthquake by applying this method on 4 single fault planes which are specified by different dip angle.We analysis the resolvability of finite fault analysis by using telesesimic data and how to choose a reasonable dip angle for the Wenchuan Earthquake.Two asperities are observed on different solutions,likelihood of change dip is discussed.We present a series of finite fault models for 2005/02/22 Zarand Earthquake in Iran.At first,we constrain rupture velocity by InSAR image and free surface observation;then we conduct grid searches for the best hypocenter location, which show that the best hypocenter locates about 5km away from eastern side on the fault plane and has a depth of 10km.We also make sensitivity tests for t_p^* and t_sh^*,we find that when t_p^* decrease from 1.0 to 0.7,corresponding Mw change from 6.5 to 6.4. Tests show that it is helpful to use both P and SH waves in finite fault inversion,as SH waves have different sensitivity to kinematic process of earthquakes.All our results prove that Zarand event has 2 or 3 asperities and the biggest asperity locate at eastern side of hypocenter and has a depth of 9km.Incremental improvements in resolving for source complexity will be possible in the near future with more geodetic and near field seismic data combined with space-based observations.