The discrete wavelet transform (DWT) has potential as a tool for suppl
ying discriminatory attributes with which to characterize or cluster g
roups of seismic traces in reservoir studies. The wavelet transform ha
s the great advantage over the Fourier transform in being able to bett
er localize changes. The multiscale nature and structure of the DWT le
ads to a method of display which highlights this and allows comparison
of changes in the transform with changing data. Many different sorts
of wavelet exist and it is found that the quality of reconstruction of
a seismic trace segment, using some of the coefficients, is dependent
on the choice of wavelet, which leads us to consider choosing a wavel
et under a 'best reconstruction' criterion. Location shifts, time zero
uncertainties, are also shown to affect the transform, as do truncati
ons, resampling, etc. Using real data, examples of utilizing the DWT c
oefficients as attributes for whole trace segments or fractional trace
segments are given. Provided the DWT is applied consistently, for exa
mple with a fixed wavelet, and non-truncated data, the transform produ
ces useful results. Care must be exercised if it is applied to data of
different lengths. However, as the algorithm is refined and improved
in the future, the DWT should prove increasingly useful.