Seismic data acquired by CREWES at an experimental low frequency shoot at Hussar, Alberta, in September 2011 were processed to attenuate unwanted noise and wavetrains and to retain or even enhance the low frequency content. We analyze the stacked data by creating plots of lateral phase-coherency versus frequency. The initial unprocessed data show strong coherency down to 7.5 Hz and weak coherency to 5 Hz but nothing below that.
The data processed by CREWES has radial filters and Gabor deconvolution applied for noise attenuation. The processed data show good coherency to 3 Hz but little in the range of 0-3 Hz. The same data processed by CGGVeritas had much more, and different, noise attenuation, resulting in coherency at the lowest frequencies of 1-5 Hz. Effective noise attenuation appears to be the greatest factor in attaining high coherency at these low frequencies.
We find that the phase-coherency plots are affected by processing procedures such as AGC and the amount of muting before stack. AGC adversely affects the coherency while the mute should not be so harsh as to remove desired frequency content. Geophone instrument response compensation, as far as we can tell at this time, does not enhance coherency at the lowest frequencies.
View full article as PDF (4.99 Mb)