Present practice for simultaneous source surveying is to separate shots using the randomness of the shot initiation times. Random shot times make the interference between shots appear as random noise in receiver or common-midpoint gathers. This interference between shots may then be removed either by random noise attenuation methods or by inversion methods that use the coherency of the signal as a criteria for separation.
This shot time randomness is important to the quality of shot separation. If the random time differences are too small, separating the interference from the signal becomes difficult. The lowest frequency of interest determines the minimum range of the random time differences required. This minimum time range then determines the maximum number of sources that can be used simultaneously.
Care must be taken to avoid undesired regularities in the scheduled shot times. While simple random number generators typically work well for scheduling shot times when a small number of shots are involved, we are generally dealing with very large numbers of shots. Spurious regularities in the random numbers generated may generate misleading events in the seismic images if care is not taken to avoid these undesired regularities.
We introduce an approach for analysis of microseismic waveform data. More precisely, we propose a method to assemble recordings into common-receiver gathers sorted in a way that neighboring traces in a gather correspond to events located close to each other. We assume that microseismic sources located at a small distance are expected to produce similar traces recorded at the same receiver. Therefore, apparent inconsistency between neighboring traces can be considered as an indication of some error in the data. We present an algorithm for sorting microseismic recordings in the case when events are located and for the situation when only arrival time picks are defined. The algorithm is applied to a cloud of more than a thousand induced microseismic events recorded at two borehole arrays during one stage of hydraulic fracturing. We construct both arrival time based and location based gathers and show examples of wrong arrival time picks in a dataset which can hardly be revealed from the analysis of individual recordings or common-shot gathers. We demonstrate that this approach can be used for detecting secondary arrivals such as reflections. Presented algorithm is relatively simple and can be easily included into the microseismic data processing workflow as a quality control tool.
Yang, Xiaohui (China University of Petroleum) | Cao, Siyuan (China University of Petroleum) | Liu, Qiong (China University of Petroleum) | Wang, Yuzhou (China University of Petroleum) | Yuan, Dian (China University of Petroleum)
Conventional NMO correction applied to CMP gathers with long-offset usually causes wavelet stretching that debases the frequency content of the NMO corrected data. Muting is the primary method of overcoming the wavelet stretching, which usually affects subsequent processing. In this paper, based on curve fitting method, we develop a new scheme to avoid event discontinuity produced by nonstretching NMO correction using matching-pursuit algorithm. Finally, we apply our method to synthetic and real data sets. Results indicate that this method provides nonstretched NMO data with relatively higher precision.
Wide-azimuth, long-offset seismic exploration plays an important role in the unconventional exploration. Wavelet stretch generated in NMO correction process is a challenge for us to utilize the wide-azimuth long-offset seismic data. Muting is the most common method to avoid wavelet stretch. However, some negative factors are caused by muting. For instance, the stacking fold of muted CMP gathers is certainly lower than the unmuted one and the long-offset data which is benefit to AVO interpretation and anisotropy analysis is lost.
Buchholtz (1972) firstly points out that conventional NMO correction applied to CMP gathers often generates amplitude distortion with stretch at long-offset. The earliest method of nonstretched NMO correction, called blockmove- sum (BMS), is introduced by Rupert and Chun (1975). This method may cause wavelet repetition and discontinuity due to the overlap of adjacent blocks. Subsequently, various nonstretched NMO correction methods which are similar to BMS are developed. Shatilo and Aminzadeh (2000) propose a constant normal movement (CNMO) correction, which applies a constant movement for a finite time window of a seismic data. Perroud and Tygel (2004) and Masoomzadeh et al. (2010) put forward a similar non-stretched NMO correction method which is achieved by adjusting velocity function. The main drawbacks of those methods mentioned above are the inability to deal with CMP gathers whose reflection events are overlapping each other in long-offset. Besides that, it is difficult to estimate appropriate block lengths. Thus, Zhang et al.(2013) proposes a matching-pursuit-based NMO (MPNMO) correction method which can provide relatively non-stretched results. However, amplitude anomaly and discontinuity can be brought in this method where events are crossing.
Wei, Jianxin (China University of Petroleum) | Di, Bangrang (China University of Petroleum) | Ding, Pinbo (China University of Petroleum) | Wang, Lingling (China University of Petroleum) | Di, Xiao (China University of Petroleum)
We constructed set of crack samples using solid matrix material filled with thin penny-shaped inclusions, with crack density from 0% to 12%. The velocity of the shear wave parallel and perpendicular to the cracks were measured with the three sets ultrasonic transducers with different main frequencies, the relation between fast and slow shear wave velocity and crack density were obtained at different frequencies. The results show that the velocity of the fast shear wave parallel to the cracks decreases slightly with an increase of crack density in all the three frequency measurements, whereas the slow shear wave velocity decreases more rapidly with increasing crack density, and that the shear-wave anisotropy increases with increasing crack density. The shear wave velocities and anisotropy change with frequency which is related to the scattering caused by the cracks. This phenomenon is more apparently at high crack density.
A high-fold full-azimuth, full-offset 3D P-P reflection survey was acquired and processed for the purpose of characterizing a naturally fractured carbonate reservoir. The reservoir is a thick carbonate, overlain by shale, which will flow oil with a sufficient fracture network. The on-going drilling and completion program is horizontal wells plus hydraulic stimulation. The azimuthal seismic attributes were compared to ISIP (instantaneous shut-in pressure) gradients, taken as indicating minimum horizontal stress, and other calibration data as discussed in the companion papers presenting the azimuthal amplitude results. Azimuthal interval velocities derived from azimuthal prestack depth migrated data provide insight into the in-situ horizontal stress field. The success of the hydraulic stimulations depended upon having a low minimum horizontal stress and the presence of natural fractures that flow oil.
The results of this study are: 1) The magnitude of the slow interval velocity (VINTslow) in the reservoir showed a local proportional sensitivity to the local minimum horizontal stress, as estimated from the ISIP gradients. 2) the interval velocity anisotropy (VINTfast-slow) was seen to be locally inversely proportional to ISIP gradients throughout much of the survey. 3) Based on the observations in the first two results, pre-drill qualitative predictions for ISIP gradients were corroborated in a horizontal well drilled after we had finished processing and interpretation
Surface microseismic monitoring faces low signal-to-noise ratio environments that challenge our abilities to detect and locate microseismic events. In these scenarios, the amplitudes of the signals of interest are strongly distorted by the noise and propagation effects. The polarity information, on the other hand, can be better preserved. In this work, we introduce the concept of sign-bit compressive sensing imaging to constrain the search for event locations in surface microseismic monitoring. We exemplify the concept with a location algorithm that works in the compressed domain, where the dimensionality of the forward propagation operator is reduced via multiplication by a random matrix. Numerical modeling in a synthetic scenario demonstrates gains in image quality over the unconstrained solutions obtained with and without compression. Limitations of the location method are also discussed.
Traditionally, survey merging is achieved by applying the optimal bulk time shift, phase rotation and a time-varying amplitude scalar obtained to match two or more different surveys. With an increase in demand for seismic reservoir characterization, the amplitude-related seismic attributes are of greater importance. Therefore amplitude-spectrum matching is critical to survey merge processing. In contrast, the use of Fourier Transform in amplitude spectrum matching is often taken for granted, by ignoring the requirement for stationarity of the seismic signal under investigation. Our study presented here shows that ignoring the non-stationarity requirement imposed by the theory of Fourier Transform could be problematic in practice. Using the Gabor transform, a solution to this problem is illustrated by application to real data examples from offshore northwest shelf of Western Australia.
The amplitude-variation-with-offset (AVO) inversion is an important technology in estimating elastic parameters which are sensitive to reservoirs and fluid. The fluid factor based on the Biot-Gassmann equation is very popular in the fluid discrimination study. However, in the deep reservoir, seismic data with low signal-to-noise ratio often lack the information of large angle of incidence. Therefore, the three-term AVO inversion will not be stable. The main purpose of this study is to help improve the stability of fluid factor inversion for deep reservoir. Under the guidance of the theory of multi-porous media theory, we derive a new two term AVO approximate equation in terms of Gassmann fluid term and shear modulus. The new equation has the nearly same accuracy as the three term AVO equation. The analysis of information contained in reflections demonstrated that the information content didn’t lose, so we can utilize the two term AVO approximation equation to invert the fluid factor to discriminate the fluid saturated in deep reservoir. The L1 norm regularization and smooth model constraints were performed in the inversion method. Tests on synthetic data and field data application show that both fluid factor and shear modulus can be estimated reasonably. Therefore, it can be used for fluid discrimination in deep reservoir.
The absence of low-frequency information in real field seismic data prevents Full Waveform Inversion (FWI) applications from obtaining smoothed velocity model for accurate seismic imaging. In addition, for land datasets, FWI with acoustic wave equation is not suitable in describing wave propagations to accommodate surface and converted waves. The seismic data used in this paper was acquired in Saudi Arabia by using an acquisition configuration based on dispersed source arrays having three different frequency bands, namely, 1.5 to 8 Hz, 6.5 to 54 Hz and 50 to 87 Hz. Alternatively, elastic FWI is applied for the estimation of three parameters, namely, the P-wave, S-wave velocities and density to one of this dataset containing 1.5 to 8 Hz. In an algorithm of elastic FWI, wave modeling is performed in the time domain by the first-order wave equation with the staggered grid scheme while other procedures such as calculating the partial derivative wavefields are conducted in the Laplace- or Laplace-Fourier domains based on the second-order wave equation with the finite element method. To demonstrate the validity of the elastic FWI, acoustic reverse time migration was implemented on the initial and inverted P-wave velocities with the data containing 6.5 to 54 Hz.
This paper introduces the implementation and application of high productivity vibroseis acquisition in a case study. It is a high-density and wide-azimuth land seismic acquisition project with 60,000 channels and 30 vibrators. BGP has developed and applied several efficient methods to significantly increase the productivity and decrease cost. These include: 1) the dynamic slip-sweep (Sercel, 428XL Manual), which greatly reduces the source circle time through shortening slip time; 2) a geometry with appropriate source receiver ratio that can help the crew efficiently use all receiver channels and vibrators applied in the project; 3) a proper vibrator deployment that keeps the vibrator fleets as far apart as possible and increases the possibility of shooting with dynamic slip-sweep and distance-separated simultaneous sweep; 4) a sophisticated DSS guidance system (BGP, DSS Mannual) for the stakeless vibroseis operation, enabling vibrator fleets to safely and efficiently move to the preplanned vibrating points; 5) the SeisPro system (BGP, SeisPro Manual), which backs-up and automatically QCs huge volumes of data (e.g., 3 TB) acquired daily. So far, the average daily productivity has reached more than 10,000 VPs with high quality and low cost.