Pre-stack Inversion: Evolving the Science ofInversion

Over the past 2 decades seismic inversion, ostensibly the process of deriving rock properties from seismic measurements, has evolved significantly. The early methods of recursive inversion converted seismic traces to well log traces providing a measurement of the "pseudo acoustic impedance". The acoustic impedance could also be expressed as "pseudo-acoustic velocity" by assuming a simple relationship between velocity, density and acoustic impedance. In any event though, the inverted property was still acoustic impedance.

While the property of acoustic impedance is more of a geophysical measurement than a geologic rock property, it did yield some indication of actual rock types. Most importantly, it demonstrated that valuable physical information was present in seismic data which was being overlooked by conventional wiggle traces.

The resolution of recursive inversion was limited to the bandwidth of the seismic data (hence the name bandlimited inversion). By using spike detection algorithms to convert the seismic trace to a high frequency sparse reflectivity series prior to inversion, sparse spike inversion algorithms could achieve high resolution. The "blocky" lithologic boundaries created by sparse-spike methods more accurately modeled actual geologic conditions although the output physical quantity was still "pseudo-acoustic impedance".

Recently, model-based inversion schemes have evolved which essentially rely on the fact that the forward model of a "good" inversion should very closely match the actual seismic data. Using iterative forward modeling schemes, these methods perturb an initial acoustic impedance model until it's forward model matches the seismic traces. These methods have the advantage of allowing some degree of control over the starting point and hence the resulting inversion. In any event, model based inversions still derive acoustic impedance.

AVO techniques have demonstrated that measurement of the conversion of compressional energy to shear energy at interfaces can yield information about the fluids and lithology present. More recently, advances in pre-stack imaging and analysis have resulted in significantly improved pre-stack signal quality with better preservation of lithologic information.

We present LithSeis, a pre- tack inversion technique which combines inversion and A VO technology with anisotropic petrophysics. LithSeis uses pre-stack seismic data as well as sonic, density and gamma ray logs to directly derive clastic rock properties including sand/shale content, gas saturation, water volume, and effective porosity. The new technique demonstrates that significantly more information is contained in the seismic wavefield than simply acoustic impedance and that we can reliably quantify rock properties from seismic data.

We demonstrate the technique on North American and international seismic data. Ongoing research will soon yield oil content and carbonate mineralogy venturing closer to the ultimate goal of deriving complete rock properties from seismic data.