The targets of seismic exploration for oil and gas are becoming increasingly complex. Clastic rock reservoirs, volcanic rock reservoirs, carbonate rock reservoirs, and unconventional reservoirs have all become core ta...The targets of seismic exploration for oil and gas are becoming increasingly complex. Clastic rock reservoirs, volcanic rock reservoirs, carbonate rock reservoirs, and unconventional reservoirs have all become core targets for oil and gas exploration. In response to these challenges, seismic wave imaging has evolved from qualitative, band-limited reflection coefficients to quantitative, broadband acoustic impedance. This transition is essential for accurately characterizing complex reservoirs. While band-limited reflection coefficients qualitatively or semi-quantitatively describe the geometric structure of subsurface media in three-dimensional depth space, broadband acoustic impedance provides a more quantitative depiction of lithologic variations, making it more suitable for complex lithologic reservoir characterization. Both imaging approaches represent parameter estimation problems under conditions of insufficient information. Full waveform inversion(FWI), grounded in Bayesian parameter estimation theory, has emerged as a key technology for seismic imaging. However, FWI is a highly nonlinear inversion problem, and the conventional gradient-based iterative approach often fails to achieve convergence to the full wavenumber spectrum(across from low to high), especially when applied to low-quality onshore seismic data. To address these limitations, we propose an alternative FWI-based seismic imaging method aimed at obtaining stable broadband acoustic impedance models for lithologic reservoir interpretation. The proposed workflow consists of three key steps:(1) Characteristic wave inversion(CWI) is used to establish the background velocity, and a spatial data fusion algorithm is employed to obtain the background density;these two parameters are then fused to generate the background impedance.(2) Based on common-image gathers(CIGs), quantitative weak sidelobe reflection coefficients are estimated through true-amplitude imaging, seismic-well calibration, and imaging-domain deconvolution.(3) Broadband acoustic impedance is reconstructed through information fusion by solving a constrained optimization problem. This method avoids the instability and nonconvergence issues commonly associated with direct inversion-based approaches. Moreover, the framework allows the integration of additional prior information, further enhancing modeling accuracy. Unlike conventional impedance inversion methods, the proposed approach clearly defines the source of each wavenumber component: low-frequency information is derived from the background impedance model, high-frequency content from high-resolution reflection coefficient imaging, and mid-frequency content is supplemented by improved background velocity and structural imaging, as well as enhanced low-to-mid-frequency information in the bandlimited reflection coefficients. This broadband acoustic impedance modeling method is particularly well-suited for low-quality onshore seismic data. Field data application results validate the feasibility, robustness, and practical value of the proposed workflow and its key technologies.展开更多
基金the sponsors of the WPI group for their financial supportfinancially supported by the National Natural Science Foundation of China(Grant Number:42474142,42304124,42174135,42074143)+4 种基金the China Postdoctoral Science Foundation(Grant No.2023M732633)the National Key R&D Program of China(Grant Number:2018YFA0702503)the RIPED and BGP of CNPCGeophysical Research Institute of Shengli Oilfield,Geophysical Research Institute(Nanjing)and its Northwest Branch of SINOPECResearch Institute and Zhanjiang Branch of CNOOC for their funding
文摘The targets of seismic exploration for oil and gas are becoming increasingly complex. Clastic rock reservoirs, volcanic rock reservoirs, carbonate rock reservoirs, and unconventional reservoirs have all become core targets for oil and gas exploration. In response to these challenges, seismic wave imaging has evolved from qualitative, band-limited reflection coefficients to quantitative, broadband acoustic impedance. This transition is essential for accurately characterizing complex reservoirs. While band-limited reflection coefficients qualitatively or semi-quantitatively describe the geometric structure of subsurface media in three-dimensional depth space, broadband acoustic impedance provides a more quantitative depiction of lithologic variations, making it more suitable for complex lithologic reservoir characterization. Both imaging approaches represent parameter estimation problems under conditions of insufficient information. Full waveform inversion(FWI), grounded in Bayesian parameter estimation theory, has emerged as a key technology for seismic imaging. However, FWI is a highly nonlinear inversion problem, and the conventional gradient-based iterative approach often fails to achieve convergence to the full wavenumber spectrum(across from low to high), especially when applied to low-quality onshore seismic data. To address these limitations, we propose an alternative FWI-based seismic imaging method aimed at obtaining stable broadband acoustic impedance models for lithologic reservoir interpretation. The proposed workflow consists of three key steps:(1) Characteristic wave inversion(CWI) is used to establish the background velocity, and a spatial data fusion algorithm is employed to obtain the background density;these two parameters are then fused to generate the background impedance.(2) Based on common-image gathers(CIGs), quantitative weak sidelobe reflection coefficients are estimated through true-amplitude imaging, seismic-well calibration, and imaging-domain deconvolution.(3) Broadband acoustic impedance is reconstructed through information fusion by solving a constrained optimization problem. This method avoids the instability and nonconvergence issues commonly associated with direct inversion-based approaches. Moreover, the framework allows the integration of additional prior information, further enhancing modeling accuracy. Unlike conventional impedance inversion methods, the proposed approach clearly defines the source of each wavenumber component: low-frequency information is derived from the background impedance model, high-frequency content from high-resolution reflection coefficient imaging, and mid-frequency content is supplemented by improved background velocity and structural imaging, as well as enhanced low-to-mid-frequency information in the bandlimited reflection coefficients. This broadband acoustic impedance modeling method is particularly well-suited for low-quality onshore seismic data. Field data application results validate the feasibility, robustness, and practical value of the proposed workflow and its key technologies.
