3C VSP tomography inversion for subsurface P‐ and S‐wave velocity distribution

Li, Yingping; Zhao, Xiaomin; Zhou, Ri-Gui; Dushman, David; Janak, Peter

doi:10.1190/1.2148263

Cited by 8 publications

(3 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Another common usage of first breaks in land seismic data is for geometry error corrections (Martin, 2001) and near-surface velocity model building using tomographic inversion of first arrival data (Colombo et al, 2016). In marine and ocean bottom node (OBN) data, first breaks have been used to estimate the depth of sources and hydrophones in deep-towed streamer acquisition (Walia & Hannay, 1999), in vertical seismic profile data processing for full velocity model building from P and S wave direct arrivals (Li et al, 2005) and to diagnose water velocity models and to detect spatial variations in the water velocity in the processing of OBN data (Li et al, 2015), which can be crucial for 4D processing.…”

Section: The Use Of First Breaks In Seismic Processingmentioning

confidence: 99%

First break picking with deep learning – evaluation of network architectures

Zwartjes¹,

Yoo²

2021

Geophysical Prospecting

View full text Add to dashboard Cite

In recent years, various convolutional neural network architectures have been proposed for first break picking. In this paper, we compare the standard auto-encoder and U-net architectures as well as versions enhanced with ResNet style skip connections. The U-net appears to have become the standard network for segmentation, judging from the number of published articles. Still, there is some variety in neural network architectural choices. In this paper, we assess the impact of neural network depth, width and input data size, as well as some small modifications for deep networks offered by the ResNet. In general, results improve as the networks get deeper, but with diminishing returns. The more complex the data, the more benefit the deeper networks bring. We use complete shot gathers, albeit rescaled for efficiency, to train the neural networks. For shot gathers with a simple piecewise linear moveout, this approach yields results with good accuracy when gathers are resampled to 128 × 128 samples. For shot gathers with more complex first break moveout, using our approach it is advised to stay close to the original dimension of each gather for best accuracy, at the expense of increased training times. A good trade-off between network depth, image size and training times is to use a nine-stage U-net with 256 sample images. Despite the advantages in other applications, the basic U-net outperforms a U-net with ResNet features. We show that changing the input data dimensions for trained networks does not work, despite the fact the fully convolutional networks are independent of image size. The U-net based first break picking is not sensitive to picking errors, as in many cases the neural network predictions are better than the training data where the training data have random mispicks. This suggests a practical application; namely, to train or re-train a pre-trained network on a single data set after conventional first break picking with the objective of improving conventionally picked first breaks.

show abstract

Section: The Use Of First Breaks In Seismic Processingmentioning

confidence: 99%

First break picking with deep learning – evaluation of network architectures

Zwartjes¹,

Yoo²

2021

Geophysical Prospecting

View full text Add to dashboard Cite

show abstract

“…He showed that when Poisson's ratio of the upper layer is larger than the lower layer, P-waves from the upper layer propagating downward to the interface at small angles can produce transmitted converted S-waves. Zhao et al (2005) considered that in some P-wave source zero-offset VSP records, the main reasons for the observed pure S-waves produced near the source are: the shear component of the seismic source, near-source heterogeneity, and seismic anisotropy near the source. Li et al (2005) used the observed VSP downgoing P-and S-waves in a P-and S-wave velocity inversion.…”

Section: Introductionmentioning

confidence: 99%

Pure S-waves in land P-wave source VSP data

et al. 2007

View full text Add to dashboard Cite

Conventional land vertical seismic profiling (VSP) exploration usually uses P-wave sources and three-component geophones for receivers, emphasizing P-and converted S-waves. Previous studies show that both dynamite borehole shots and vertical vibrations from controllable seismic sources at the surface will produce relatively strong pure P-waves and weaker pure S-waves. Interfaces with a large Poisson's ratio difference have a positive influence on the formation of strong transmitted converted S-waves. By a comparative analysis of pure S-waves from sources and converted downgoing S-waves, we believe that the main frequency of pure S-waves is usually lower than pure P-waves while the main frequency of downgoing converted S-waves is close to that of P-waves. We have studied zero-offset and offset VSP data from land P-wave sources. Results show that pure S-waves commonly exist in these data with differences in wave intensity. S-wave velocity can be obtained from the P-wave source zero-offset VSP data. Finally, we discuss the bright future of joint application of VSP P-and S-waves and the full use of S-waves in P-wave source VSP data.

show abstract

“…Practically, an available 'best' surface seismic velocity model will be used as an initial velocity model for 3D VSP imaging and then the model is updated by VSP First Arrival Time (FAT) tomography and rarely by VSP reflection tomography. Unfortunately, VSP tomography practices (e.g., Li et al 2003aLi et al , 2005 indicated a typical RMS misfit of over 100 ms for the initial velocity models used for the walk-away VSP and/or 3D VSP with offsets and depths of up to 6 km, implying velocity models are questionable in poorly illuminated regions by surface seismic survey. Although VSP tomography can bring RMS misfits significantly down to about 10-20 ms, sometimes, resulting velocity models are neither unique nor make sense geologically, due to limited receiver coverage and the low fold coverage of VSP surveys.…”

Section: Introductionmentioning

confidence: 99%

Borehole seismic quantitative diagnosis of a seismic velocity model for 3D seismic imaging of subsurface structures

Hewett

2014

Geophysical Prospecting

Self Cite

View full text Add to dashboard Cite

Quantifying uncertainties of surface seismic velocity models is a difficult task. The use of multifarious vertical seismic profiling data provides quantification of velocity model uncertainties. Two borehole seismic diagnosis methods for velocity model uncertainties are developed. The first method is performed with 3D ray‐traced first arrival times through a surface seismic velocity model and compared with measured first arrival times from downhole receivers. It generates a profile of absolute and relative misfits, indicating the velocity uncertainty as a function of depth. The second approach directly compares the differences between the first arrival incident angles measured from 3‐C borehole seismic data and the calculated first arrival incident angles of seismic rays through the velocity model to quantify the model uncertainties. Five seismic velocity models in the Gulf of Mexico are diagnosed using the diagnosis methods with up to six sets of borehole seismic data recorded in three wells. The uncertainty profiles vary with depths and azimuths, revealing the complexity of velocity model uncertainties and providing some clues for further velocity updating. The diagnosis methods are used as velocity updating QC tools and integrated with the conventional seismic velocity updating flow to complement the criteria of flat common image gathers.

show abstract

3C VSP tomography inversion for subsurface P‐ and S‐wave velocity distribution

Cited by 8 publications

References 7 publications

First break picking with deep learning – evaluation of network architectures

First break picking with deep learning – evaluation of network architectures

Pure S-waves in land P-wave source VSP data

Borehole seismic quantitative diagnosis of a seismic velocity model for 3D seismic imaging of subsurface structures

Contact Info

Product

Resources

About