Next: Rytov traveltime sensitivity
Up: Theory
Previous: Theory
One approach to building a linear finite-frequency traveltime operator
is to apply the first-order Born approximation, to obtain a linear
relationship between slowness perturbation,
, and wavefield
perturbation,
,
| ![\begin{displaymath}
{\bf \delta U} = {\bf B \; \delta S}.\end{displaymath}](img12.gif) |
(5) |
The Born operator,
, is a discrete implementation of
equation (16), which is described in the Appendix.
Traveltime perturbations may then be calculated from the wavefield
perturbation through a (linear) picking operator,
, such that
| ![\begin{displaymath}
{\bf \delta T} = {\bf C \; \delta U} = {\bf C B \; \delta S}\end{displaymath}](img15.gif) |
(6) |
where
is a (linearized) picking operator, and a function
of the background wavefield, U0.
Cross-correlating the total wavefield, U(t), with U0(t), provides
a way of measuring their relative
time-shift,
. Marquering et al. (1999) uses
this to provide the following explicit linear relationship between
and
,
| ![\begin{displaymath}
\delta T =
\frac{\int_{t_1}^{t_2} {\dot U}(t) \; \delta U(t) \; dt}
{\int_{t_1}^{t_2} {\ddot U}(t) \; U(t) \; dt},\end{displaymath}](img18.gif) |
(7) |
where dots denote differentiation with respect to t, and t1 and
t2 define a temporal window around the event of interest.
Equation (7) is only valid for small time-shifts,
.
Next: Rytov traveltime sensitivity
Up: Theory
Previous: Theory
Stanford Exploration Project
4/27/2000