Phase determination

Along with amplitude and frequency, phase is a fundamental attribute of seismic data.

Seismic data are usually processed to be zero-phase, and we usually assume that the phase is stable in space and time. Indeed, these assumptions are central to most AVO and other quantitative work.

Overview
Based on the advice of Roden & Sepulveda 1999 and Perz et al 2004, there are four simple ways to help determine phase:
 * 1) Inspection
 * 2) Well ties
 * 3) Instantaneous phase
 * 4) Rotation tests

In general, you are unlikely to be able to see a phase difference of 15&deg; or less, and indeed it probably would not matter for horizon picking or even quantitative work. A phase rotation of 30&deg; is probably worth fixing for quantitative work. Anything more than 45&deg; is worth fixing even for interpretation.

Beware: undoing the picking you do on a phase rotated volume is onerous: only make a rotation of your data when you are sure it is more geological that way.

Inspection
Simply examining a strong seismic event that corresponds to an isolated geologic surface of known impedance contrast. It helps if the contrast, which must be spatially consistent in polarity, is fairly strong. Good examples are the seafloor, the Wabamun (in Western Canada), and the Devonian Unconformity (in the Athabasca oil sands). The only thing to really look for is a consistently symmetrical wavelet — this is why the reflector must be isolated, as any tuning or interference effects will spoil the symmetry.

An template of some rotated wavelets here would help with this.

Well ties
A good workflow is to tie wells with a zero-phase wavelet, at least at first. When tying, make a note of the phase disparity at the well — many software tools let you plot correlation coefficient against phase rotation. Once you have a feel for the variance of the well ties, you can start to see if there are spatial trends in this variance. Perhaps most wells tie better with a 90&deg; phase rotation.

Instantaneous phase
This method was described in detail in Perz et al (2004>ref name=perz />). Since we want to pick a phase-independent horizon, we can't just measure instantaneous phase on a horizon. We must do this:
 * 1) Start with the original data, volume D
 * 2) Compute the envelope E (sometimes called instantaneous amplitude or absolute amplitude)
 * 3) Pick a horizon H on a strong peak on E
 * 4) Compute instantaneous phase on H from volume D

The result gives an indication of phase in the data. It should be close to zero.

This method makes it easy to capture the spatial variance, and if you run several horizons, the temporal variance too.

Rotation tests
This method is straightforward but fiddly. Rotate the data by various amounts, in 15&deg; increments (15&deg;, 30&deg;, 45&deg;, etc). Choose a strong reflection and measure the amplitude at the peak or trough. The reflector should have the strongest amplitude when the data are zero phase.

The problem with this method is that it is hard to capture the spatial variance.