Processing for Meaningful 4D Differences
Standard 3D acquisition and processing approaches almost never are adequate for 4D monitoring. Two “legacy” surveys optimally imaged for 3D use may be interpreted for large changes, such as gas-cap formation, but almost certainly they will not be suitable for measuring small differences or determining reliable details of waterfloods. “Legacy” is the general euphemism used to describe surveys that have not been acquired and processed carefully for 4D analysis.
If surveys have not been acquired specifically to have repeatable geometry for 4D analysis, the results usually are both late and poor, even after lengthy processing. Trying to perform 4D analysis on poorly repeated acquisition is nearly always a waste of time and money. With properly repeated surveys, the processing can be quick and simple and can yield timely results.
Our inability to equalize surveys with different shooting geometries should tell us that something is incomplete in our industry's current approach. It is not our inability to remove “random” noise that is the problem, nor is it our inability to stack smooth coherent signal in phase. The problem is our inability to remove systematic semicoherent shot-generated effects, such as overburden distortion and multiples. In 4D monitoring, we should aim to repeat those problems and difference them away.
We will not go through all the details of seismic processing. Instead, we will outline a general philosophical approach and then look at some of the unique opportunities available for 4D analysis.