Slide 1

Velocity Analysis Introduction to Seismic ImagingERTH 4470/5470 Yilmaz, ch 3.1-3.3.2 Slide 2 Figs 3-1 to 3-3 Velocities of sediment and sedimentary and volcanic rocks increase with depth. For sediment this is due to compaction of pore space with increasing pressure. Increase in rock velocity is due to closure of cracks with increasing pressure. Slide 3 Normal Moveout (NMO) Correction for Flat Layers Slide 4 For single layer (Figs. 3-4, 3-6 and 3-8). NMO depends on {x 2 /v 2 }, so we need to know v(z) in order to flatten CMP gather before stacking. For brute stack we assume a constant velocity (e.g. water) for simplificity, while knowing that this will not give a good image for deeper structures. Slide 5 Slide 6 For multiple layers, t 2 = t 2 (0) + x 2 /v rms 2, so plots of t 2 vs x 2 will give a straight line with slope of 1/v rms. The root-mean square velocity (v rms ) is determined by eq. 3.4 in terms of the interval velocity (v i ) and travel time (t i ) of each layer interval (i). (Figs. 3.9 to 3-11) Slide 7 Slide 8 For real data, we expect moveout of reflectors to decrease with depth (=time) as velocity increases with depth due to compaction Slide 9 Various definitions of velocity (Box6.4) Notice difference between v rms and v av but it is small Note also that v nmo = v rms only for the small offset (spread) approximation (Fig. 3-22). For larger spread offsets, the best fit to flatten the actual moveout is not the same. Slide 10 This results in non-linear expansion of the time axis, which is greater for larger x and smaller v. This changes the frequency of the arrivals. (Fig. 3-13 and Table 3-2). When this effect becomes too large (generally in the upper 1-2 sec TWTT), we need to mute the result (Fig. 3-12). This can be done automatically for stretching greater than a certain amount, or by picking the front mutes by hand as we did to remove the refraction arrivals. (Fig. 3-14). For continuous data (not individual picks) we need to flatten arrivals (ie remove increase in t as function of x and v) by stretching the time axis. Slide 11 Slide 12 Methods for Velocity Analysis Slide 13 Synthetic example with 4 layers showing CMP gather, velocity spectrum and t 2 -x 2 plots. Spectrum is unnormalized, cross-correlation sum with a gated row plot. Slide 14 Real example with 4 primary layers and multiple secondary layers. Spectrum is unnormalized, cross-correlation sum with contour plot. Slide 15 Use of constant-velocity gathers (CVG) for a single CMP gather at various velocities to help detail exact nature of stacking velocities Slide 16 Slide 17 Use of constant velocity stack (CVS) for range of gathers at different stacking velocities. Helpful in sections of low signal-to-noise (e.g. at greater depths in the section) Slide 18 Limitations in accuracy and resolution of velocity estimates Slide 19 Synthetic examples of 4 layers showing various plots of velocity spectra. Effect of spread (offset) length Slide 20 Lack of long offsets reduce resolution of lower (high velocity) layers with smaller moveout Slide 21 Lack of near-offsets reduce resolution of shallow layers Partial stacking (using incomplete fold) can save money (computer time) but can result in reduced resolution Slide 22 Reduced resolution caused by decrease in signal-to-noise Slide 23 Slide 24 Effect of dipping layer Slide 25 Effect of dip is only significant when dip angle is large (i.e. > 20 o )