Multiple elimination is broadly classified into data-driven and model-driven methods. This classification is based on the domain in which the multiples and primaries as separated. The model-driven (e.g. Radon Filter), transforms the data into a domain, where the separation is done mostly based on velocity. This approach becomes problematic when both multiples and primaries have similar velocities. Conversely, the Data-driven approach predicts exclusively all the surface related multiples for a particular data. Filters are then used to match the predicted multiples to those in the input data, which are then subtracted from the input data. In addition, the data-driven approach can handle data which can’t be easily fit into parabolic or hyperbolic curvature which is a requirement for the Radon based methods.
SRME is a data driven approach. It has been applied successfully in marine data but its application to land data has proved daunting because of the effect of the irregularity of the land surface. The prediction is done such that for every source receiver pair, all the possible ray paths are represented and accounted for.
It has been demonstrated that Land SRME can be achieved using a practical implementation of the SRME algorithm. This flow generally contains four steps; preprocessing (*only difference is that the data is corrected to surface as opposed to a flat datum), then data is then preconditioned to stabilize the multiple prediction. Finally, least-squares subtraction is performed in higher dimensions to remove the multiples and preserve the primaries.
However, SRME in general can only be done applied conveniently to 2D data owing to computational limitations of performing such predictive operations for 3D volumes (*many receivers and sources not necessarily aligned in a straight line) and also because data with high signal to noise ratio is required it to be implemented successfully.
Juefu Wang* and Shaowu Wang, 2013, Practical implementation of SRME for land multiple attenuation GeoConvention 2013: Integration.