next up previous print clean
Next: Synthetic tests Up: Artman and Fomel: Fourier Previous: Introduction

Theory

The space-domain shot-profile imaging condition including subsurface offset for shot-profile migration Rickett and Sava (2002) is  
 \begin{displaymath}
I(x,h)\vert _{\omega,z}=U(x+h)\,D^*(x-h)\;,\end{displaymath} (1)
Where I is the migrated image produced by cross-correlating the up-coming, U, and down-going, D, wavefields at every depth and frequency. Both x and h can be areal vectors. * represents complex conjugation. To derive the Fourier-domain equivalent, we will perform a piece-wise proof and begin with the Fourier transform D* to $\hat{D}^*$ (neglecting Fourier scaling)
\begin{displaymath}
\hat{I}(x,h)=U(x+h)\,\int \hat{D}^*(k_s)\,e^{i\,k_s\,(x-h)}\,dk_s \;.\end{displaymath} (2)
Continue by Fourier transforming the variable x to find
\begin{displaymath}
\hat{I}(k_x,h)=\int U(x+h)\,\int \hat{D}^*(k_s) e^{i\,k_s\,(x-h)}\,dk_s
 e^{-i\,x\,k_x}\,dx \;.\end{displaymath} (3)
By reordering variables, the equivalent form
   \begin{eqnarray}
\nonumber
\hat{I}(k_x,h) & = & \int\hat{D}^*(k_s)\, e^{-i\,k_s\...
 ...\,h\,(k_x-2\,k_s)} 
 \int U(x')\,e^{-i\,x'\,(k_x-k_s)}\,d x'\,dk_s\end{eqnarray}
(4)
is achieved. From here, we can recognize the inner integral is the Fourier transform of the wavefield U which can be replaced directly to yield
\begin{displaymath}
\hat{I}(k_x,h)= \int\hat{U}(k_x-k_s)\,\hat{D}^*(k_s)\,
e^{i\,h\,(k_x-2\,k_s)}\,dk_s \;.\end{displaymath} (5)
With the use of the definition of offset, $k_h=k_x-2\,k_s$, we can replace several of the above arguments with equivalent expressions to find
\begin{displaymath}
\hat{I}(k_x,h)= \frac{1}{2}\,\int\hat{U}\left(\frac{k_x+k_h}...
 ...hat{D}^*\left(\frac{k_x-k_h}{2}\right)\,e^{i\,h\,k_h}\,dk_h \;.\end{displaymath} (6)
The last integral is recognized as an inverse Fourier transform, this time over the kh variable. Using this fact, we arrive at the multi-dimensional (over x and h, which can be two-dimensional themselves) Fourier transform of the general shot-profile imaging condition  
 \begin{displaymath}
\widehat{I}(k_x,k_h)=\frac{1}{2}\,\hat{U}\left(\frac{k_x+k_h}{2}\right)
\,\hat{D}^*\left(\frac{k_x-k_h}{2}\right)\;.\end{displaymath} (7)

From this equation, the result that the Fourier-domain equivalent to the conventional space-domain imaging condition for shot-profile migration is again a lagged multiplication of the up-coming and down-going wavefields at each frequency and depth level. Evaluating the arguments inside the wavefields to produce a component of the image shows that the wavefields, in the wavenumber domain, will need to be interpolated by a factor of two to calculate the image space output. The Table 1 showing example calculations of the components of the image space looks like

  
Table 1: Layout of wavenumber components in Fourier-domain imaging conditions
\begin{table}

\begin{displaymath}
k_h\end{displaymath}\begin{displaymath}
k_x 
...
 ...{3}{2})\hat{D}^*(\frac{1}{2}) \\  \hline \end{array}\end{displaymath}\end{table}


next up previous print clean
Next: Synthetic tests Up: Artman and Fomel: Fourier Previous: Introduction
Stanford Exploration Project
5/3/2005