Research On Least-squares Reverse Time Migration Of Multiples

Seismic exploration data,especially ocean data,contains a lot of multiples.Multiples are multiple reflection of the subsurface,contain abundant information.Multiples not only can be regarded as noises.Compares to primaries,multiples have a longer path,a wider range of illumination.Multiples can cover the shadow areas where primaries can not arrive.Multiples have smaller angle of reflection which results a higher vertical resolution.Therefore we should change our mind that multiples are regarded as noises to be suppressed or removed.Instead research on migration of multiples should be improved which will have very important significance for the full use of seismic data information and quality improvement of the imaging of subsurface structures.The paper first reviews the basic theory of the multiples.Then we carry out forward modeling of the surface-related multiples through the finite difference numerical simulation method.It provides database for the following multiples imaging methods.About research method of surface-related multiples,this paper first discuss the reverse time migration proposed by Liu Yike in 2011.Then we analyze feasibility,range of illumination and imaging results of simple model and complex model,and reverse time migration of multiples shows the effectiveness of migration,especially for the deep structures.But it still needs multiple prediction,which is time consuming and prone to error.Hence the paper next proposes a new reverse time migration approach that can simultaneously migrate the primaries and the free-surface related multiples,in which no multiple prediction is needed.But we must pay attention to the selection of source wavelet.There are some difficulties for reverse-time migration technology in the imaging process of deep reservoirs.In addition,there is crosstalk in the migration image.Another difficulty is that the imaging condition is the correlation between the downgoing and upgoing fields,which can give rise to ringiness in the multiples migration image because the recorded data can be considered as a very ringy wavelet.To overcome these problems,we propose least-squares reverse time migration of multiples(LSRTMM),which can improve the image quality by removing artifacts,balancing amplitudes,and suppressing crosstalk.