Motivation

Most of the large structural onshore and shelf hydrocarbon plays have already been delineated, and seismic imaging efforts have begun to concentrate on subtler phenomena, such as those related to the presence of hydrocarbons in rocks. A resonant note is struck in crustal seismology by the need to delineate the extent of lithospheric melts. AVO is one of the most common methods of characterizing the fluids in rocks in either exploration Yilmaz (2001) or crustal Makovsky and Klemperer (1999) surveys.

The FEAVO anomalies are much stronger than regular AVO effects, rendering AVO analysis impossible. Their removal will thus allow AVO analysis. A byproduct of the FEAVO removal process is a very accurate velocity model [White et al. (1988) shows that velocity contrasts as small as 2% can generate FEAVO], and this will also highly benefit AVO analysis, which is highly sensitive to the velocity used for prestack migration [Clapp (2002), Mora and Biondi (2000)].

FEAVO removal is also desirable for reasons beyond the obvious practical ones described above: in principle, the reflectivity that seismology seeks to recover is the high spatial frequency component of the impedance field. A FEAVO-contaminated image is simply inaccurate. Imaging the correct reflectivities is in line with the modern efforts towards true-amplitude imaging [Biondi (2001b); Sava and Biondi (2001a)]. And the by-product - an accurate velocity model describing the low spatial frequencies of the velocity field - is every bit as important as the reflectivity image itself Claerbout (1999).

 

prodef Figure 1 a: Part of a CMP gather exhibiting FEAVO anomalies. The strong event at 2.3s shows a slight departure from hyperbolic moveout, too subtle to allow successful classical traveltime tomography. b: FEAVO anomalies in the midpoint-offset space (Kjartansson ``V''s). The preprocessing consisted in: muting, spherical divergence correction, bandpass filter, interpolation of missing or noisy traces, hydrophone balancing, f-k filtering, and offset continuation to fill in the small offsets (with a forward and inverse DMO cascade using the log-stretch DMO in the Fourier domain described in Zhou et al. (1996) and implemented by Vlad and Biondi (2001)). The figure has been produced exactely as in Kjartansson (1979): square and vertically stack the data between 1.5 and 3.5 seconds, then take the logarithm to increase the dynamic range. Offset continuation does not predict the FEAVO anomalies (the tips of the ``V''s are not extended into the extrapolated small offsets)

 

whitebig Figure 2 The physical explanation for the expression of FEAVO anomalies in CMP gathers (Figure 1a). If the frequency of the waves is high enough or the anomaly large enough, we will see a small triplication. Otherwise, only offset-dependant amplitude focusing (FEAVO) is visible. The traveltime delays are negligible, as the velocity anomaly changes only very little the length of the rays. Figure taken from White et al. (1988).

 

vilus
Figure 3 The physical explanation for the expression of FEAVO anomalies in midpoint-offset space (Kjartansson ``V''s, Figure 1b). In the upper picture, the blobs are transmission anomalies and the arrows are raypaths for the zero offset and for the maximum offset recordings. For case A (anomaly on the reflector), only a single midpoint is affected, for all offsets. Case C (anomaly at the surface), is actually a static: its ``footprint'' is a pair of streaks slanting 45^o from the offset axis. Case B (in between) gives a pair of streaks with angles smaller than 45^o.