6 CHAPTER 1.

INTRODUCTION

1.3 Dynamic range and seismic data display

Seismic data tends to be a challenge to computer graphics as well as to computer capacity. A single
seismic record can have a tremendous amplitude range. In speaking of this, the term dynamic range
is used which refers to a span of real numbers. The range of numerical voltages that a seismic
recording system can faithfully handle is called its dynamic range. For example, current digital
recording systems use fixed instrument gain and represent amplitudes as a 24 bit integer computer
word3 . The first bit is used to record sign while the last bit tends to fluctuate randomly so effectively
22 bits are available. This means that the amplitude range can be as large as 222 ≈ 106.6 . Using
the definition of a decibel given in equation (1.1), this corresponds to about 132db, an enormous
spread of possible values. This 132db range is usually never fully realized when recording signals for
a variety of reasons, the most important being the ambient noise levels at the recording site and the
instrument gain settings.
In a fixed-gain system, the instrument gain settings are determined to minimize clipping by
the analog-to-digital converter while still preserving the smallest possible signal amplitudes. This
usually means that some clipping will occur on events nearest the seismic source. A very strong
signal should saturate 22-23 bits while a weak signal may only effect the lowest several bits. Thus the
precision, which refers to the number of significant digits used to represent a floating point number,
steadily declines from the largest to smallest number in the dynamic range.

1.3.1 Single trace plotting and dynamic range

Figure 1.2 was produced with Code Snippet 1.3.1 and shows two real seismic traces recorded in 1997
by CREWES 4 . This type of plot is called a wiggle trace display. The upper trace, called tracefar,
was recorded on the vertical component of a three component geophone placed about 1000 m from
the surface location of a dynamite shot. The lower trace, called tracenear, was similar except
that it was recorded only 10 m from the shot. (Both traces come from the shot record shown in
Figures 1.12 and 1.13.) The dynamite explosion was produced with 4 kg of explosives placed at 18
m depth. Such an explosive charge√is about 2 m long so the distance from the top of the charge
to the closest geophone was about 162 + 102 ≈ 19 m while to the farthest geophone was about
√
162 + 10002 ≈ 1000 m. The vertical axes for the two traces indicate the very large amplitude
difference between them. If they were plotted on the same axes, tracefar would appear as a flat
horizontal line next to tracenear.
Figure 1.3 (also produced with Code Snippet 1.3.1) shows a more definitive comparison of the
amplitudes of the two traces. Here the trace envelopes are compared using a decibel scale. A trace
envelope is a mathematical estimate of a bounding curve for the signal (this will be investigated
more fully later in this book) and is computed using hilbert, that computes the complex, analytic
trace ((Taner et al., 1979) and (Cohen, 1995)), and then takes the absolute value. The conversion
to decibels is done with the convenience function todb (for which there is an inverse fromdb ).
Function todb implements equation (1.1) for both real and complex signals. (In the latter case,
todb returns a complex signal whose real part is the amplitude in decibels and whose imaginary part
is the phase.) By default, the maximum amplitude for the decibel scale is the maximum absolute
value of the signal; but, this may also be specified as the second input to todb . Function fromdb
reconstructs the original signal given the output of todb .

3 Previous systems used a 16 bit word and variable gain. A four-bit gain word, an 11 bit mantissa, and a sign bit

determined the recorded value.

4 CREWES is an acronym for Consortium for Research in Elastic Wave Exploration Seismology at the University

of Calgary.