Sparse Sampling Data Reconstruction Method And Comparison Of Full Waveform Inversion Results

With the constant improvement of the seismic exploration degree and exploration in the field of extending unceasingly,the exploration targets become increasingly complex,in the process of field data acquisition,due to restrictions in wild terrain conditions and equipments,leads to the irregular seismic data,the incomplete situation becomes increasingly serious,affect the later data processing and interpretation.The traditional reconstruction method has a higher sampling rate,and it cannot be used in the field to collect the samples,and the cost of collecting is higher.Based on compressive sensing theory,the sparse data reconstruction,acquisition of seismic data in reducing the amount of data collection,under the premise of cost savings,to obtain the good reconstruction data,Based on compressed sensing theory,less than traditional Nyquist sampling rate of the reconstruction of the incomplete data will also be able to get good effect,based on the theoretical framework,this paper mainly studies the different ways of sparse sampling data reconstruction and the effect of different sparse representation of base and the seismic data reconstruction effect of different reconstruction methods.In the sparse collection mode,the effect of random sampling and the sampling on the reconstruction effect was compared.In terms of the Fourier transform and the Curvelet transform,has studied the SPGL1 method based on Fourier transform data reconstruction and SPGL1 method based on curvelet transform data reconstruction,compared the effect of reconstruction data,more outstanding the reconstruction of the curvelet transform data from seismic data quality is higher.On the reconstruction method,compared the SPGL1 reconstruction method based on curvelet transform and the reconstruction of the POCS reconstruction method based on curvelet transform effect,SPGL1 reconstruction method and the reconstruction of the mature method compared with the reconstruction effect is good is not good,the method to improve accuracy,has the larger development space.In the reconstruction of the above data,the selection is based on the reconstruction of the POCS method data,used to make the full waveform inversion,and the original data,in comparison with the results of full waveform of data reconstruction of full wave inversion result is more ideal,only in the deep speed is somewhat fitting error.