DISCUSSION

We present a methodology for using daylight data to image source locations or reflector boundaries for v(x, y, z) media. Traces are crosscorrelated in time to form the correlation kernel in equation (3), this kernel is weighted by the appropriate migration operator, and summation over the g and g' indices is carried out to give the migrated image. Our theoretical formulae partly validate the conjecture: a crosscorrelogram is a trace that can be migrated in the same fashion as a trace generated by a source at g and a receiver at g'.

Other possibilities can be explored using the mathematical formalism in this paper. It is straightforward to develop a non-linear inverse methodology by using crosscorrelation misfit functions and obtaining its gradient.

It is interesting to note the analogy of inverting correlograms with that of constructing holograms. Light intensity images are used to reconstruct holograms, which are images of the object polluted by noisy interference terms. Seismic correlograms can also be used to reconstruct images of the object, but the other terms in equation (3) suggest that the data are polluted by noisy interference terms. However, we can design migration operators to reconstruct the object function (e.g., reflectivity) or information from the interference terms (e.g., source locations). An alternative is to simultaneously reconstruct both coherent noise and signal models with least squares migration filtering.