Fahdi Maula

In 3D and 2D surface seismic interpretation, seismic waveform shapes and attributes can identify facies and reservoir parameters laterally with more details than traditional amplitude mapping. Herein, a method on 1D zero offset VSP (ZVSP) data was adapted, giving a unique perspective of lateral heterogeneity analysis using VSP seismic waveform shapes and attributes. The downgoing wavefield of VSP measures seismic wave variation in the vertical direction. When enough VSPs are covering an area, we can combine them to get an insight into both vertical and lateral variations.

An unsupervised machine learning clustering algorithm based on K-means and self-organizing maps (SOM) was used to group the waveforms based on their shape similarity and attributes (frequency spectrum). The algorithm produced a cluster map, a probability map, and a typical wavelet for each cluster. These were then used to analyze the vertical and lateral heterogeneity from well to well based on VSP waveform attributes. The used example data were an open dataset, the Poseidon 3D data from the NW Shelf, Australia (Browse Basin), provided by GEO Science Australia. Six wells were available with VSP datasets.

This technique can be of use to incorporate additional attributes from VSP into extensive 3D subsurface interpretations. For precautions, the VSP measurement or data preconditioning must be done reliably prior to clustering. Such A method may function well for vertical well ZVSP where variation was noticed because of the vertical seismic ray path.

In this study, the application of VSP data has been extended from the conventional single well-to-well basis. The value of integrating VSP characterization has been investigated from various wells and numerous measurements to discern both vertical and lateral heterogeneity in a studied area.