Well log prediction while drilling estimates the rock properties ahead of drilling bits. A reliable well log prediction is able to assist reservoir engineers in updating the geological models and adjusting the drilling strategy if necessary. This is of great significance in reducing the drilling risk and saving costs. Conventional interactive integration of geophysical data and geological understanding is the primary approach to realize well log prediction while drilling. In this paper, we propose a new artificial intelligence approach to make the well log prediction while drilling by integrating seismic impedance with three neural networks: LSTM, Bidirectional LSTM (Bi-LSTM), and Double Chain LSTM (DC-LSTM). The DC-LSTM is a new LSTM network proposed in this study while the other two are existing ones. These three networks are thoroughly adapted, compared, and tested to fit the artificial intelligent prediction process. The prediction approach can integrate not only seismic information of the sedimentary formation around the drilling bit but also the rock property changing trend through the upper and lower formations. The Bi-LSTM and the DC-LSTM networks achieve higher prediction accuracy than the LSTM network. Additionally, the DC-LSTM approach significantly promotes prediction efficiency by reducing the number of training parameters and saving computational time without compromising prediction accuracy. The field data application of the three networks, LSTM, Bi-LSTM, and DC-LSTM, demonstrates that prediction accuracy based on the Bi-LSTM and DC-LSTM is higher than that of the LSTM, and DC-LSTM has the highest efficiency overall.

An integrated study on fracture prediction by analyzing seismic anisotropy within a 3D P-wave seismic data set from an oilfield in Northwest China is performed. The research area is dominated by a regional anticline and the target layer is a fractured gas reservoir at a depth of approximately 3 km under the anticline. Two major aspects are considered. First, two P-wave poststack attributes (reflection strength and frequency content in selected lines) are extracted and their features in the target layer are examined, looking for the possible fracture distribution in two prospective zones. Second, the azimuthal attribute analysis approach is applied to estimate fracture density and orientation in the whole area from the prestack data (amplitude and traveltime). The result from the poststack data indicates the high possibility of fracture distributions in these two zones, and the prediction from azimuthal amplitudes in the prestack data is consistent with the fracture orientation from the well interpretation and the observation of outcrops at the top of the anticline but suffers from a poor offset-depth ratio and acquisition footprint. The prediction from azimuthal traveltime is strongly influenced by the regional anticline structure and, therefore, is not reliable. Nevertheless, the prediction from post- and prestack attributes provides more details of the fracture distribution in the target reservoir, which can expand understanding of the reservoir fracture distribution.

Traffic noise is an important type of passive seismic data because it usually includes strong dispersive surface wave components and can be easily accessed. It can be used to extract virtual surface waves via seismic interferometry algorithms for the purpose of imaging subsurface shear wave velocity distribution. In this paper, we propose a scheme to improve the retrieval of surface waves from traffic noise recorded using linear arrays along traffic roads. By deconvolving the decomposed traffic noise wavefield, robust surface wave traces can be computed from a short noise record. First the far‐field component of the traffic noise recording is extracted and separated into unidirectionally propagating components. Then deconvolution interferometry is applied to these separated far‐field wavefield to extract surface wave Green's function. With this scheme, crosstalk noise and near‐field artifacts are excluded from the computation, and surface wave traces with high signal‐to‐noise ratio (SNR) are achieved using short traffic noise traces. In a synthetic test virtual surface waves estimated with the proposed method show significantly higher SNR than those computed with the conventional interferometry workflows, and matches well with simulated active source traces. A field data example with traffic noise recorded in a distributed acoustic sensing experiment also shows that surface waves estimated using the proposed methodology demonstrate higher SNR than those computed with the conventional interferometry schemes and that the virtual surface waves generated using 4 s of traffic noise demonstrate signal quality comparable to the surface waves recorded in this experiment with an active source.

The propagation of seismic waves in fractured rocks is greatly influenced by the fracture system and fluid content. Seismic anisotropy varies with frequency and fluid saturation. Previous theories on frequency-dependent anisotropy (FDA) are mostly limited to the assumption of single-phase fluid, whereas almost all reservoirs are usually partially or fully saturated with one or more fluids. The inelasticity, heterogeneity, and anisotropy exhibited by real earth media seriously challenge the traditional theory of uniform fully elastic media. Therefore, based on the Chapman elastic theory, we consider the viscoelastic model, the relative mobility of saturated fluids, and the coupled effects of commonly independently considered squirt flow and patch effects on the frequency-dependent anisotropic seismic wave propagation in viscoelastic media. By calculating the frequency-dependent anisotropic elastic coefficients of two-phase immiscible fluid-saturated fractured rocks, a new viscoelastic Chapman extension model for FDA in partially saturated rocks, which includes the unified seismic wave propagation effects of squirt flow and patch effects, is constructed. The effect of relative permeability is significant, and compared to the case of complete saturation, the effective fluid mobility in partially saturated rocks may be lower, which may lead to nonmonotonic changes in elastic modulus and water saturation. Numerical experiments are conducted on our viscoelastic Chapman extension model to discuss the coupling effects of squirt flow and patch effects on the FDA of viscoelastic media in cracked rocks with partially saturated reservoirs, under the conditions of no fractures and the existence of fractures. The simulation results of the P-wave modulus, P-wave velocity, and anisotropy parameters confirm the validity of our method and model. We link the anisotropy of viscoelastic media with fluid flow parameters in fractures, which is conducive to improving the understanding of frequency-dependent seismic anisotropy in partially saturated rocks with fractures and the combination of seismology and reservoir engineering.