Layer Parallel Smoothing (LPS)

Horizon-driven smoothing with fault preservation

Seismic noise is a common problem to almost all datasets. Linear and non-linear coherent noise are frequently attenuated using FK or Tau-p dip filtering processes, while random noise is attacked using methods such as FX or FXY deconvolution. However, some types of noise fall between these two classifications, being coherent over very short distances or perhaps more coherent in one direction than another. These noise types can be difficult to attenuate satisfactorily. Trace mixing can be effective, but conventional smoothing operates on constant time samples and tends to attenuate dipping primary energy.

Layer parallel smoothing (LPS) is a technique that acts as a powerful noise attenuator by smoothing along the local reflector dip (therefore, protecting dipping primary events), while preserving local discontinuities such as fault intersections. We have now upgraded this technology to allow the use of the same dip model for smoothing of multiple datasets. This improves stability when processing, for example, time-lapse seismic cubes.

LPS has particular value in time-lapse (4D) processing. In a recent application from the North Sea, 4D difference data were observed to have significant contamination from dipping noise. The noise was non-repeatable from the baseline to repeat survey, increasing the level of 4D noise in the difference dataset. Conventional dip filtering proved ineffectual because the apparent dips were similar to the dips of nearby primary events, while K filtering reduced the noise but also degraded the dipping reflectors.

We applied LPS to smooth along the dominant dip directions, reducing the 4D noise and resulting in a difference volume where true 4D signals stood out more clearly than before.


Request More Information

LPS improves 4D difference by enhancing 4D signal

LPS attenuates noise while preserving local discontinuities.LPS improves 4D signal in time-lapse surveys.
PrevNextZoom1 of 2