Topic:
How do I replace a bad trace?

Description:
It often happens that a large data volume has one bad sample.  This single
bad sample causes a number of scaling problems, including an apparent
blanking of all time slices.  Following are notes on how to solve this
problem.

Solution:
First you must obtain the data with the spike.

If you still have the original segy tape, that tape includes the spike data.

If you no longer have the original data you can export the bad data to a segy 
file using DataManager > Segy export and select the line with the spike.  
 
In either case you will reload the segy data.
 
The "correct", and simple way to remove the spike is to use the editval
option.  In this simple example the bad value(s) will be replaced by zero.
 
1.) If you are not using an expression for line, cdp, or sp you need to
make one up.  One simple way to do this is to set sp to expression, then
write an expression setting the sp to cdp.  i.e., cdp = sp

2.) Look at the seismic data (using shift+mb1) to get a rough idea what
the amplitudes should look like.  You can then decide what is a maximum
acceptable amplitude.  For example, if the amplitudes of strong reflections are
in the hundreds, you are probably safe to replace all values greater than
100,000 by zero.  Such values are probably noise spikes, not real data.

3.) Below the sp expression line type editval (maximum acceptable value). e.g.,
        editval (100000.) 

Note that this also clips values less than -100,000 to zero.
You are removing both positive and negative spikes.

4.) Reload the bad line.
 






Last Modified on: 19-MAY-99