RTM using effective boundary saving: A staggered grid GPU implementation

Yang, Pengliang; Gao, Jinghuai; Wang, Baoli

doi:10.1016/j.cageo.2014.04.004

Cited by 70 publications

(20 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…4). 따라서 GPU를 사용한 거꿀 참 반사 보 정이나 완전 파형 역산의 경우 알고리즘을 수정하여 계산량을 늘리는 대신 메모리 복사를 없애는 연구들도 진행되고 있다 (Yang et al, 2014(Yang et al, , 2015. …”

Section: 실험 구성unclassified

Time-domain 3D Wave Propagation Modeling and Memory Management Using Graphics Processing Units

Kim¹,

Ryu²,

Ha³

2016

Geophysics and Geophysical Exploration

View full text Add to dashboard Cite

Abstracts: We used graphics processing units for an efficient time-domain 3D wave propagation modeling. Since graphics processing units are designed for massively parallel processes, we need to optimize the calculation and memory management to fully exploit graphics processing units. We focused on the memory management and examined the performance of programs with respect to the memory management methods. We also tested the effects of memory transfer on the performance of the program by varying the order of finite difference equation and the size of velocity models. The results show that the memory transfer takes a larger portion of the running time than that of the finite difference calculation in programs transferring whole 3D wavefield.

show abstract

Section: 실험 구성unclassified

Time-domain 3D Wave Propagation Modeling and Memory Management Using Graphics Processing Units

Kim¹,

Ryu²,

Ha³

2016

Geophysics and Geophysical Exploration

View full text Add to dashboard Cite

show abstract

“…One key problem of GPU-based implementations of FWI is that the computation is always much faster than the data transfer between the host and device. Many researchers choose to reconstruct the source wavefield instead of storing the modeling time history on the disk, just saving the boundaries (Dussaud et al, 2008;Yang et al, 2014). For 2N -th order finite difference, regular grid scheme needs to save N points on each side (Dussaud et al, 2008), while staggered-grid scheme required at least 2N − 1 points on each side (Yang et al, 2014).…”

Section: Wavefield Reconstruction Via Boundary Savingmentioning

confidence: 99%

“…Recent advances in computing capability and hardware makes FWI a popular research subject to improve velocity models. As a booming technology, graphics processing unit (GPU) has been widely used to mitigate the computational drawbacks in seismic imaging (Micikevicius, 2009;Yang et al, 2014) and inversion (Boonyasiriwat et al, 2010;Shin et al, 2014), due to its potential gain in performance. One key problem for GPU implementation is that the parallel computation is much faster while the data communication between host and device always takes longer time.…”

Section: Introductionmentioning

confidence: 99%

A graphics processing unit implementation of time-domain full-waveform inversion

2015

Self Cite

View full text Add to dashboard Cite

The graphics processing unit (GPU) has become a popular device for seismic imaging and inversion due to its superior speedup performance. In this paper we implement GPU-based full waveform inversion (FWI) using the wavefield reconstruction strategy. Because the computation on GPU is much faster than CPU-GPU data communication, in our implementation the boundaries of the forward modeling are saved on the device to avert the issue of data transfer between host and device. The Clayton-Enquist absorbing boundary is adopted to maintain the efficiency of GPU computation. A hybrid nonlinear conjugate gradient algorithm combined with the parallel reduction scheme is utilized to do computation in GPU blocks. The numerical results confirm the validity of our implementation.

show abstract

“…However, for a huge 3D problem, we need a large number of computational resources to save forward and backward wavefields on every grid point and at every time step. To reduce computational cost, the excitation approach (Kalita and Alkhalifah ; Oh, Kalita and Alkhalifah ) and the boundary‐saving method were suggested, in which we only store the forward wavefields at the model boundary and then we reconstruct the full forward wavefields by back propagating the boundary values (Mitter ; Yang, Gao and Wang ). This approach provided reasonable FWI results for a large‐scale 3D elastic problem (Raknes and Arntsen ; Raknes and Weibull ).…”

Section: Introductionmentioning

confidence: 99%

Study on the full‐waveform inversion strategy for 3D elastic orthorhombic anisotropic media: application to ocean bottom cable data

Alkhalifah

2019

Geophysical Prospecting

View full text Add to dashboard Cite

A B S T R A C TThe multi-parameter full waveform inversion is an essential tool to estimate subsurface anisotropic properties in a reservoir that may have complex geological behaviour requiring an elastic orthorhombic medium description. For such elastic orthorhombic media, finding a proper inversion strategy to mitigate parameter trade-off and reduce the Null space is crucial considering the large number of parameters describing the medium. We apply our recently developed strategy for orthorhombic medium inversion on synthetic and real ocean bottom cable data, and find the most efficient inversion strategy. At first, we analyse the trade-off patterns in three elastic orthorhombic parameterizations for a hockey-puck-shaped model. We, then, compare the performance of these orthorhombic parameterizations on a 3D synthetic ocean bottom cable data, which are obtained from a channel-shaped model. By interpreting the inverted models based on analytic radiation patterns of the nine elastic orthorhombic parameters in each parameterization, we observe that parameterizations, which have one P-wave velocity, one S-wave velocity and seven dimensionless parameters, can be optimal to recover subsurface isotropic properties in the early stages of inversion. Then, we show that the choice of three anisotropic parameters and their deviations along the horizontal plane helps us mitigate the complexity of the multi-parameter inversion by decoupling anisotropic features, which have vertical and horizontal symmetric axes, respectively. This decoupling of isotropic and anisotropic properties enables us to perform the multi-parameter anisotropic inversion in a multi-stage manner. In addition, for the marine acquisition, we show that the number of parameters can be reduced from 9 to 4, which makes the multi-parameter inversion more practical. Finally, we apply the elastic orthorhombic inversion to real ocean bottom cable data.

show abstract

RTM using effective boundary saving: A staggered grid GPU implementation

Cited by 70 publications

References 37 publications

Time-domain 3D Wave Propagation Modeling and Memory Management Using Graphics Processing Units

Time-domain 3D Wave Propagation Modeling and Memory Management Using Graphics Processing Units

A graphics processing unit implementation of time-domain full-waveform inversion

Study on the full‐waveform inversion strategy for 3D elastic orthorhombic anisotropic media: application to ocean bottom cable data

Contact Info

Product

Resources

About