Reservoir Description And Identification Based On Frequency-dependent AVO Inversion

Reservoir description and identification is the fundamental purpose of exploration. Reservoir description refers to the obtaining of characteristic reservoir parameters from the geophysical data, and reservoir identification refers to the cognition of underground lithology system based on obtained parameter information. Along with objective change of oil and gas exploration from conventional structural reservoirs to unconventional, complex and concealed lithologic reservoir, the geophysical prospecting is required to provide richer and more reliable parameter information in order to reduce the development risk, and that makes the accuracy and information content of seismic inversion a major focus and breakthrough. Through a large number of instances prove that by predecessors, seismic reflection amplitude is not only a function of offset, but also a function of frequency. This frequency-dependent feature used to be associated with reservoir properties, which conventional separate study of Amplitude Versus Offset(AVO) or Amplitude Versus Frequency(AVF) cannot describe or utilize. Defining frequency-dependent response the feature of seismic reflection amplitude as a function of frequency, and frequency-dependent AVO(FDAVO) the multifactor function of frequency and offset(incident angle). Author deem that if only we excavate and utilize the frequency-dependent AVO feature, the accuracy and information content of qualitative and quantitative reservoir identification will be enhanced.In accordance with the order of cognition, development and utilization, starting with forward problem of frequency-dependent AVO, article analyzes physical mechanisms of frequency-dependent response with their characteristics; afterwards, exploring the possibility of developing typical time domain AVO inversions(data-driven and model-driven) to frequency-dependent, and investigating performances in more application fields; then paper discuss application value of inversion results in the field of reservoir lithology identification.To analyze, contrast and filter physical mechanisms which give rise to frequency-dependent responses in seismic frequency band, author test typical rock physical models(e.g. static fluid substitution model, patchy saturation model, squirt flow model, BISQ model and multi-scale fracture model). Source of the effect can be classified into two kinds, meso-scale wave-induced fluid flow(including patchy saturation and meso-scale fracture structure) and thin-layer tuning. Moreover, AVO theories for different mediums are studied(including isotropic medium, porous elastic medium, anisotropic medium and interbed layer to build forward operator. After numerical simulating these three mechanisms, this article finds that: a) for patchy saturation reservoirs, gas saturation determines the velocity high- and low-limit and dispersion degree of reservoir, which affects the frequency-dependent response at interface; b) for fracture medium, level and band of frequency-dependent depend on fracture density and time-scale factor, respectively; c) interbed layer affect reflectivity as a filter, which controls by the combination of thickness of single layer and layer number, in frequency domain. That is, some frequency components of reflection energy are enhanced, while others are attenuated.Seismic attributes, which estimated from conventional data-driven AVO inversion, are usually math combination of several kinds of elastic parameters, with a drawback of its insensitivity to fluid properties. Dispersion attributes, which estimated from FDAVO inversion, take advantage of dissipation mechanism of wave propagation and give expression to extra reservoir information, such as fluid mobility and saturation. This article displays the extraction progress of dispersion attribute and test how the time-frequency analysis affect results. Furthermore, author discusses new application areas of dispersion attributes: a) fluid indicator to verify existence and distribution, where the dispersion responses are corresponding to the dispersive interface, magnitudes are corresponding to dispersion levels and time-frequency analyses determine attributes’ resolutions; b) discrimination and description of Bottom Simulating Reflector(BSR), detection of free gas zone. Dispersion attributes response at BSRs appears to stronger than the seafloors. c) Time-lapse monitoring of CO2 vertical migration. Dispersion attributes will show a pattern of zero to high value and decrease to zero, which corresponding to the three stages of fluid substitution.Conventional data-driven AVO inversion aims to complete model parameterization, but its assumption of elastic and independent brings two restrictions: 1) inaccurate inversion result due to the unsuitable forward model. 2) Only considering elastic properties limit the amount of inversion information. Combing dynamic rock physical model as a forward operator, frequency-dependent AVO for objective function, The L1 norm criterion as constraint condition and optimization algorithm as search method, author suggest inversion framework for model parameterization. According analyzing results of parameter sensitivity, inversion methodology for three frequency-dependent mechanisms are presented: a) simultaneous inversion of fracture density and time scale factor distribution in fracture medium; a) saturation distribution estimating in patchy saturation medium; c) simultaneous inversion of thickness of single layer and layer number in interbed layers. For simulating different reservoir conditions, model selections are: a) 1D depth model; 2D depth-offset model for b) fluid-complex c) background field-complex situation. Moreover, error analysises are employed to test algorithm in jamming environments.Reservoir lithology identification methods are changing from manual method to multivariate statistic means as a trend. Probability statistics methods which represented by Bayesian frameworks, more abundant independent parameter information bring accurate results and make parameters from FDAVO inversion available for application. Hypothetical geological information for prior condition, verification geological information for likelihood condition, and method unifying multivariate information for constraining identification results while agreeing with seismic data. After substitute estimated results from FDAVO inversion, author proof that identification results are modified, even though inversion presents.