The output of an impulse-response program always looks awful.
The main reason is the large area of -space
which is near the Nyquist frequency or above the evanescent cutoff.
Since we rarely sample data in time as coarsely
as the Nyquist criterion permits,
the program below defaults to final filtering with the filter (1+*Z*)/(1-.8*Z*).
This filter still passes a lot of energy outside the usual bandwidth
of seismic data.
Since all land data and most marine data
do not have the zero frequency component,
the program contains an option to filter further
with (1-*Z*)/(1-.8*Z*).
I haven't displayed anything with this extra filter
because I wanted this book to show
all the artifacts you might encounter.
Furthermore, I deliberately enhanced the visibility of artifacts
on wiggle-trace, variable-area plots
by plotting with the nonlinear gain.
(Perspective hidden-line drawings always have linear gain).
Since my plots are necessarily about 10 cm square in this book
and in practice you will look at plots of about a hundred times the area,
I plotted only one second of travel time.

10/31/1997