Noise-reduction algorithms in seismic processing are often based on the statistical assumption that noise is Gaussian, white, and spatially not correlated. Unfortunately there is often no physical reason why this should be the case. The problem of Blind Separation of Signals is a rapidly emerging topic in the signal processing community. In mature areas, detection of by-passed hydrocarbons and new satellite fields have a high business priority in order to extend the life of the field.

BLISS proposed an analysis of the algorithms recently developed to solve the problem of Blind Separation of Signals. These techniques are based on the assumption that noise and signals display mutual statistical independence which is arguably a physically more justified condition.

The project will be divided in three work packages between teams from Leeds and Grenoble. Leeds will start with a critical evaluation of existing signal-separation techniques. Grenoble will lead theoretical development while keeping in mind the statistical characteristics of seismic data. Leeds University will then use the acquired expertise to incorporate the most suitable techniques into the noise-reduction toolbox and apply them on select datasets.

These newly developed techniques will be implemented for advanced data processing and noise reduction to increase the resolution of conventional seismic data and thereby exploration prospects.

Until now only a few isolated attempts have been made to employ these techniques in seismic data processing. However, initial results look promising and significantly improved signal-to-noise ratios were obtained. A second phase of the consortium commenced at the end of 2010 with the support of two oil company sponsors.

One of the Principal Investigators and project manager, Dr Mirko van der Baan, has re-located from the University of Leeds to the University of Alberta in Edmonton Canada.