Seismic wavelets and resolution - Seismic Amplitude - page 28

Table 3.1 Some typical values of Q (after Sheriff and Geldart,
1995 )

180º

Lithology Q

1.0

Sedimentary rocks 20 - 200

Sandstone 70 - 130

Shale 20 - 70

Limestone 50 - 200

Q=30 Q=100

Chalk 135

Dolomite 190

2.3

Rocks with gas in pore space 5-50

Metamorphic rocks 200-400

90º

150º

Igneous rocks 75-300

Figure 3.9 Modelled effect of absorption on wavelet shape over a
1.3 s zone (modified after Angeleri and Loinger, 1984 ) . In a real
example the bandwidth would also change. Note that gas effect (i.e. high absorption in the presence of

gas) may be due to cross-flow of liquids between fully
liquid-saturated pores and surrounding rock with
partial gas saturation. These developments are inter-
esting, but there is as yet no single widely accepted
view of the significance of absorption effects as a
direct hydrocarbon indicator. Sometimes gas sands
show an obvious shift to low-frequency signal at
and for some distance below them but sometimes
they do not.

Q ¼ 30 is

characteristic for example of shales whereas Q ¼ 100 would be

characteristic of limestones and sandstone.

(i.e. near-zero horizontal offset between source and
receiver). This source

receiver geometry is highly
favourable, because raypaths for the top and base of
an interval are essentially coincident. Thus spectral
differences between the arrivals can be reliably
assumed to be due only to the interval properties
between the two depths.
Determining Q from VSPs is of course only pos-
sible where there is a borehole, so it would be desir-
able to extend lateral coverage by measuring Q from
surface seismic. This is not easy to do and it is not
routine practice. Q cannot be derived from stacked
traces owing to mixing of traces with different path
length, the spectral distortion due to moveout and the
incorporation of AVO effects. Dasgupta and Clark
( 1998 ) have devised a method for using pre-stack data
in which amplitude spectra are computed for a par-
ticular reflection for each trace of a CMP gather
individually. They are then corrected for moveout
stretch ( Chapter 6 ) and compared to the source sig-
nature to obtain an estimate of Q.
As can be seen from Table 3.1 , the presence of gas
in a sandstone can give rise to anomalously low Q
(high absorption), particularly at intermediate satur-
ations. This has driven interest in the use of Q in
direct hydrocarbon detection. Chapman et al.( 2006 )
have developed a theoretical framework to explain
abnormally high attenuation in hydrocarbon reser-
voirs. Dvorkin and Mavko ( 2006 ) suggest that the

-

3.5 Idealised wavelets
There are a number of types of idealised wavelets that
are in common use, for example to make well syn-
thetics when the exact wavelet is not known. Some
examples are shown in Fig. 3.10 . In each type the
wavelets can be zero phase, constant phase, or min-
imum phase, and the frequency content is specified by
the user. The Butterworth wavelet is defined by the
lower and upper bandpass frequencies and the slopes
of the response, usually in dB per octave. The Ormsby
wavelet is defined by four frequencies: low-cut, low-
pass, high-pass and high-cut. Both these wavelets
develop oscillatory side lobes if the cut-off slopes are
set too steep. The Ricker wavelet (Ricker, 1940 ) is
defined by a single central frequency and has only
two side lobes. Hosken ( 1988 ) advised strongly
against the use of Ricker wavelets, largely because real
seismic generally has a flat topped as opposed to a
peaked amplitude spectrum ( Fig. 3.10 ). For detailed
work it is probably best to use a custom wavelet
created by the methods to be described in Chapter 4 ,

28