Playing with parameters

SEPlib programs generally do simple things. They are still very flexible, though, because their default behavior can be modified by appropriate command-line or history-file parameters. Most programs have at least a few options or parameters; some of them have hundreds. Let's look at some relatively simple examples.

We saw already that most of the Txx.H data consists of zero values that are not very interesting for us. Let's trim the data a bit. The SEPlib utility Window is used for this purpose. If you remember, Txx.H consists of 1 plane of 20 traces, with 1024 samples in each trace. In SEPlib notation, n3=1 n2=20 n1=1024. Let's ``zoom in'' on the interesting part of the data between times .4 and .8 seconds and offsets from the smallest (.1) up to 1. (i.e., we will keep samples 200 through 400 of the first 10 traces). To do its job Window needs to know the number of samples and the first sample to accept for each axis. (The defaults are ``all of them that you can'' and ``begin at the beginning'', respectively.) For our example we would have:

Window < Txx.H n2=10 n1=200 f1=200 > Txx_Windowed.H

Fortunately Window is smart enough to understand your command using the values on the two axes too:

Window < Txx.H min1=.4 max1=.8 max2=1. > Txx_Windowed.H

We must warn you that the second method is more risky, since it is possible that you have an error in the sampling parameters such as o1 and d1 in the history file, or you simply forgot to specify them at all (so they defaulted to 0 and 1 respectively). You may prefer to tell Window where to start and end using integer sample numbers. If you are sure that you would never make such mistakes, did you catch the discrepancy in the two examples above? You'll find they don't give quite the same results! Starting at .4 and ending at .8 is 201 samples, not 200. As usual for computer programs, Window does what you tell it to do, not necessarily what you mean for it to do.

Whichever way you did it,

In Txx_Windowed.H

shows us the file Txx_Windowed.H is much reduced in size:

Txx_Windowed.H:
in="/usr/local/sep/scr/joe/_Txx_Windowed.H@"
        expands to in="/usr/local/sep/scr/joe/_Txx_Windowed.H@"
        esize=4
        n1=200  n2=10   n3=1    n4=1            2010 elem       8040 bytes
        d1=.002 d2=.1   d3=1    d4=1
        o1=.4   o2=.1   o3=0    o4=0
        label1=Time, seconds
        label2=Offset, kilometers

Note in particular the new value for o1.

Now let's have a look at what we have done:

Wiggle < Txx_Windowed.H | Tube

So far so good, but let's suppose that the journal you are submitting to insists wiggle plots must have traces that run down instead of across, the order of the traces must go the other way, and the traces must have geophysical-style shading. No problem; from Wiggle's self documentation you find there are three parameters that are likely to do what you need. Try them out:

Wiggle < Txx_Windowed.H transp=yes poly=yes yreverse=yes | Tube

Perhaps it would be better if the trace amplitudes were a little lower? The parameter pclip stands for ``percentile clip''; the default is $98\%$, which is meant to scale the plotting using the effective maximum absolute value while ignoring a few huge abnormal spikes. Our data is mostly zero and has no abnormal spikes, so perhaps clipping on the maximum would be more appropriate:

Wiggle < Txx_Windowed.H transp=y poly=y yreverse=y pclip=100 | Tube

(Figure  shows what this plot should look like.)

 

Wiggle2
Figure 3 Wiggle < Txx\_Windowed.H par=plotpar | Pspen

Note most SEPlib programs don't care whether you type ``yes'' or ``no'', ``y'' or ``n'', or even ``1'' or ``0''. (Although a few old FORTRAN ones only accept the numbers, and a few old C ones only accept the letters.)

Looking at Wiggle's self-doc you may have noticed that Wiggle also supports parameters like ``min1'' and ``max2''. These will usually work just like the ones in Window and probably seem like the preferred way to do windowing of plots. Unfortunately (for now at least) Wiggle does the windowing a lazy way. The whole plot is calculated (and plotted!) just as before, and the graphics driver does all the work of trimming away the excess. You can use these parameters of plot programs like Wiggle to make slight adjustments to the boundaries of a plot, or to make a plot smaller, but don't use them to ``zoom in'' very far! (We'll see an example of a legitimate use of these ``dangerous'' parameters in a few pages.)