One method that takes propagation effects into account is the Delft approach Verschuur et al. (1992). This technique has the advantage of working with the surface data only and for any type of geology. Thus, it is often the method of choice for multiple attenuation in complex geology Miley et al. (2001). To be accurate, the Delft method requires a very dense coverage of sources and receivers. If this condition is relatively easy to meet in 2-D, it becomes much more difficult to fulfill with 3-D surveys. New multiple prediction techniques are therefore developed to circumvent this limitation Levin and Johnston (2001); van Dedem and Verschuur (2001).
A powerful multiple attenuation technique would be one that first takes the wavefield propagation into account, with whatever data we have, and then uses moveout discrepancies to remove multiples. To achieve this goal, we first propose using prestack depth migration as our imaging operator. Assuming that we have the correct velocity and an accurate migration scheme, we can then handle/image any type of complex geology very accurately. In this process, both primaries and multiples are migrated, after which they are transformed to angle gathers using standard techniques. In the angle domain, primaries are flat and multiples are curved, mimicking the situation we have after NMO for simple geology. Finally, we propose mapping the angle gathers into a Radon domain where the signal/noise separation can be achieved. This method has the potential to work with 2-D or 3-D data as long as angle gathers can be estimated. It is also much cheaper than the Delft approach, but it can still handle complicated geologic media.