Youngseo Kim

Frequency-domain reverse-time migration with source estimation

Although artificially generated seismic sources such as dynamite, vibroseis, and air guns are used in seismic exploration, it is not easy to exactly recover the source wavelet in field recording or in data processing. For this reason, seismic data processing often assumes that an explosive-source wavelet can be described by a well-known function (e.g., a Ricker wavelet), a near-offset trace, or a deconvolved wavelet. In frequency-domain waveform inversion, it has been proven that a source wavelet can be estimated by an optimization method, and incorporating the source wavelet estimation into an inversion algorithm yields better inversion results. We have developed source wavelet estimation into 2D two-way frequency-domain reverse-time migration. The source wavelet is first estimated independently of reverse-time migration by an optimization method such as the full Newton method. It is then used in reverse-time migration. This source-wavelet-incorporated reverse-time migration algorithm is applied to three...

Acceleration of stable TTI P-wave reverse-time migration with GPUs

When a pseudo-acoustic TTI (tilted transversely isotropic) coupled wave equation is used to implement reverse-time migration (RTM), shear wave energy is significantly included in the migration image. Because anisotropy has intrinsic elastic characteristics, coupling P-wave and S-wave modes in the pseudo-acoustic wave equation is inevitable. In RTM with only primary energy or the P-wave mode in seismic data, the S-wave energy is regarded as noise for the migration image. To solve this problem, we derive a pure P-wave equation for TTI media that excludes the S-wave energy. Additionally, we apply the rapid expansion method (REM) based on a Chebyshev expansion and a pseudo-spectral method (PSM) to calculate spatial derivatives in the wave equation. When REM is incorporated with the PSM for the spatial derivatives, wavefields with high numerical accuracy can be obtained without grid dispersion when performing numerical wave modeling. Another problem in the implementation of TTI RTM is that wavefields in an area with high gradients of dip or azimuth angles can be blown up in the progression of the forward and backward algorithms of the RTM. We stabilize the wavefields by applying a spatial-frequency domain high-cut filter when calculating the spatial derivatives using the PSM. In addition, to increase performance speed, the graphic processing unit (GPU) architecture is used instead of traditional CPU architecture. To confirm the degree of acceleration compared to the CPU version on our RTM, we then analyze the performance measurements according to the number of GPUs employed.

Improved logarithmic waveform inversion considering the power‐spectrum of the wavefield

Local minima of an objective function often prevent solutions of logarithmic waveform inversions from converging to the global minimum in cases where an initial velocity model for the inversion is not close to the true velocity structure. In particular, forward-modeled wave fields with small power-spectrum values influence the numerical stability of the gradient direction. Accordingly, it is important to remove these small values to allow a solution to the misfit function to converge to the global minimum. Therefore, we developed a waveform-inversion technique using forward-modeled wavefields with relatively large values of the power-spectrum.

Frequency-domain Homotopy Inversion Using the Perturbation Theory

Full waveform inversion is one of the indirect inversion methods for finding the solution to nonlinear problems. Although this technique gives high-resolution subsurface images, it requires high-performance computers and has the potential to lead to useless and misleading results. To solve these problems, this study proposes a new inversion technique based on perturbation theory, which is referred to as homotopy inversion. The proposed technique also depends in large measure on an initial model for successful processing. Unlike full waveform inversion, however, this technique can obtain a final result from the first iteration. In addition, since a step length required for descent methods is not used in homotopy inversion, we do not need to worry about failure as a result of the difficulty of determining a proper step length. To demonstrate our technique, we test the numerical examples on a simple anticline model. Furthermore, we compare true seismograms with inverted seismograms to confirm the feasibility of our algorithm.

A time-domain waveform inversion using filtering techniques

If local minima are present in a calculation of an objective function, it is difficult to identify subsurface information. Therefore, it is essential to exploit new technique to reduce the number of local minima in seismic inversions. To make the steepest descent direction head toward the global minimum, we propose using a full waveform inversion with filtering techniques. During the inversion process, we allocated a time window that includes the time corresponding to the highest amplitude of each observed trace. In the calculation of the objective function using an 2 L norm  , data contained by the