Font Size: a A A

Elastic Full Waveform Inversion In Time Domain

Posted on:2017-06-15Degree:MasterType:Thesis
Country:ChinaCandidate:S Y SunFull Text:PDF
GTID:2310330566457055Subject:Geological Resources and Geological Engineering
Abstract/Summary:PDF Full Text Request
Taking the advantage of the amplitude and phase of seismic records,FWI can estimate the accurate and reliable underground medium parameters through the method of interactive updates.It possess the features of high resolution and accuracy,which can provide the accurate velocity for pre-stack migration and the judgement of lithology.However,the acoustic FWI mainly invert the velocity of P-wave and regard the underground medium as the acoustic medium,which cannot show us the real property of the underground medium and provide the velocity of S-wave for seismic interpretation.In order to solve this problem,we need to replace the acoustic FWI by elastic FWI.The inversion parameters will be extended from the velocity of P-wave to P-wave and S-wave and the density.With the development of computer hardware,constrains of the inversion efficiency is gradually resolved.Forward modeling is the basis of inversion,the effects of elastic wave forward modeling is one of the factors which affect the results of EFWI.In this thesis,we solve two-dimensional elastic wave equation by FD schemes in time domain.In the following work,the shot records and snapshots are applied to the process of EFWI.We analyze the stability conditions,source loading and absorbing boundary conditions.In this thesis,EFWI is based on elastic wave forward modeling in time domain.We use the adjoint gradient and weighted-energy method for the parallel multi-parameters FWI process and take the multi-scale method to improve the accuracy of inversion.Model tests prove that we can get accurate velocity results by the two-step method even the initial parameter model is linear.And the density result of the two-step method is much better when compared to the conventional method.The interleaved relationship between elastic parameters and seismic data introduces a new level of complexity in the local optimization methods to solve the problem and it will generate the cross-talk noises in the inversion results.Especially when there is a much big difference among the compressional and shear wave velocity and density of the model structural,the cross-talk noises will greatly influence the iterative convergence of density.In thesis,we compare the strength of cross-talk noises in different model parameterization ways.We revise the two-step method based on the research and get a much better resultThe low-frequency information of field records is usually missing which is much different from the model test.It will greatly affect the outcome of the inversion.The envelope operator can be used to demodulate the low-frequency information of seismic records.We can use the envelope operator to demodulate the low-frequency information and recover the long-wavelength components of the underground medium when the low-frequency information is missing.The envelope EFWI is a nonlinear multi-scale method and much better than the conventional linear multi-scale method.Model tests prove that envelope EFWI can effectively reduce the non-linearity and get a good inversion result when the low-frequency information is missing.
Keywords/Search Tags:elastic wave forward modeling, elastic wave full waveform inversion, multi-parameters, multi-scales, weighted energy, demodulation, envelope operator
PDF Full Text Request
Related items