Statfjord 3D/4C data - Azimuth Processing.

"Alice through the looking glass"

Mark Thompson, Statoil

             

Conventional marine 3D seismic acquired using surface streamers has long proved a valuable tool in the understanding of the subsurface the world over. However, most conventional marine 3D seismic data sets are still not 'true 3D', because they illuminate a restricted range of azimuths in the subsurface, typically within twenty degrees of the nominal sail line azimuth. It is only today with the options of true 3D geometry possible in OBC acquisiton that we can begin to truly image a 3D world "through the looking glass".

In 1997 Statoil acquired a dense 3D/4C dataset set over the Statfjord field providing data with a full spectrum of azimuths. The survey was recorded by Geco-Prakla in four swaths, using two 5km long cables seperated by 300m in each swath, with a receiver spacing of 25m. The source lines were 11km long, shot in parallel to the receiver cables, with a shot interval of 25m (flip-flop) and sail line seperation of 100m, resulting in a source grid of 50m by 50m. This P/Z data has subsequently undergone conventional 3D processing and azimuth processing, the results of which are presented here.

The aim of the azimuth processing was to determine whether restricted azimuth stacks would produce a better image of the structurally complex East Flank area. The data were sampled into four azimuth subsets, two subsets of opposing inline azimuths and two subsets of opposing crossline azimuths. These were processed seperately using the same parameters, and the results compared to the conventional full azimuth processing.

It is seen that when all azimuths were included in the stack, a variable image was produced. The data quality is excellent in the northern portion of the survey and deteriorates towards the south. The azimuth subsets show a generally improved image, but still vary in quality compared to the full azimuth stack. We can relate the improved images to the fact that the subsurface is not equally illuminated by all azimuths, with some events aligned in certain azimuth domains. Faults for example could be preferentially highlighted by azimuth susbset processing, and we see indications of this. When these events are stacked using all azimuths, destructive interference caused by wavefields from very different raypaths may result in a degraded stack.

Our findings suggest that in a full azimuth 3D/4C dataset such as Statfjord, P/Z processing is non-trivial and requires 'True 3D' processing techniques which accurately handle data with greatly different ray paths.. Preliminary conclusions suggest that an improved subsurface image can be achieved by splitting the data into azimuth cubes and that critical processing steps such as velocity analysis and multiple attenuation should be tuned to optimise each azimuth cube.