New Velocity Model Building Techniques to Reduce Sub-Salt Exploration Risk

Stephan Petmecky¹, Martin L. Albertin², and Nick L. Burke^2
1BP Exploration, Sunbury, United Kingdom
²BP Exploration, Houston, TX

Seismic data in most sub-salt provinces are characterized by lack of angular illumination and low signal to noise ratios creating a challenging exploration environment. Three novel techniques of building more accurate sub-salt velocity models will be presented. All three techniques take advantage of geological information to constrain estimates of effective stress from which interval velocities can be derived.

1) De-salting simply corrects extrapolated sub-salt velocities by accounting for the density differences between sediments and salt. Keeping the initially estimated sub-salt pressures constant the reduction in overburden pressure, due to presence of salt, leads to an effective stress reduction which in turn lowers the originally inferred interval velocities.

2) Structural modeling takes advantage of the observation that crests of anticlines commonly coincide with velocity lows, whereas synclines tend to be characterized by faster interval velocities. This is caused by lateral pressure transfer in permeable units. Overpressure estimates for potentially connected sand bodies can therefore be used to build structural control into the sub-salt stress model.

3) The recent discovery of a very large oil accumulation in the deepwater GoM demonstrates that integrating basin modeling derived effective stresses into the sub-salt velocity model building process can significantly reduce exploration risk. A fully calibrated 3D basin model provided an effective stress/velocity cube which replaced the original sub-salt velocity field. Remigrating the seimic data resulted in major geometrical changes in the deeper parts of the targeted prospect, reducing the initially identified charge access risk. Analyzing well data from this successful exploration well confirmed that the basin modeling derived sub-salt velocities were more accurate than the initial model.

AAPG Search and Discover Article #90078©2008 AAPG Annual Convention, San Antonio, Texas