The Tembungo Field area is located about 100 km offshore Sabah, in eastern Malaysia (Figure 1). Geologically, it is a folded anticlinal structure with intense flower-type faulting. The exploration target is a deep turbidite complex in Miocene basin slope fan system. The velocity field in the overburden is severely complicated by the presence of near-surface reef carbonates, gas wipeout and complex faulting. These complications result in strong energy scattering, low signal to noise (S/N) ratio and strong multiple energy that lead to a significant deterioration of the seismic image over the field and across the whole general area. Processing, including the use of prestack depth migration (PreSDM) with a view to extracting adequate signal under these circumstances, has been fraught with difficulties. The commercial importance of the Tembungo Field was a major incentive for seeking ways of resolving the above imaging issues. A new approach that we devised and adopted proved to be immensely successful. The approach consisted of processing the data with Common Reflection Surface (CRS) technology followed by post-stack depth migration (PostSDM) in which the provelocity model was obtained through a pre-stack depth migration (PreSDM) process. The term ‘provelocity’ will be used in the present work specifically to denote the parameter derived from seismic processing as ‘velocity’ because this is a modelling parameter that can be quite different from the true propagation velocity in the ground (Al-Chalabi, 1994). Unlike conventional PreSDM, which aims at enhancing the image of the subsurface, the process in the present case was exclusively employed to produce an optimum PostSDM provelocity model. This is perhaps the most significant ‘innovation’ in the present work. It is driven by the fact that CRS is more amenable to post-stack than to pre-stack processes. Figure 2 shows the overall workflow of the approach. CRS has already established itself as an effective technique for enhancing S/N ratio. The CRS technique was introduced by Müller (1998) as a data driven zero-offset simulation method. Hubral (1999) then refined the technique and followed the strategy of a macro model independent imaging principle. The technique uses larger stacking surfaces rather than relying on a single CMP stack location as in conventional stacking processes. This improves the emergence angle and wavefront curvatures approximation, utilizing seismic wavefield and travel time information. The parameter derivations are based on second order travel time estimations which utilize multi-dimensional global maximum of the coherency function.