8
The processing approach described above has succeeded in reducing the striping errors and noise spikes
substantially. Subtracting the final data set from original shows the spikes and striping errors removed – see
plot below.
You can see that the striping error changes in magnitude over the site and increases in magnitude over higher
resistance areas - 1 ohm over most of the site but up to 10 ohms in the high resistance region. Even so, the
processing approach has succeeded in removing the majority of this error. You can see from the difference plot
that the restoration of the large scale background changes using addition may have over increased the
magnitude of the linear features running from top to bottom. Changing the magnitude of the low pass filtered
data set to between 70% and 100% before addition would help reduce this effect. Also, in some situations a
wider low pass filter window may produce better results, especially where the background features are
extensive.
Note that any archaeological features that are parallel with the traverse direction may be removed by this
processing technique, so, depending on their importance, you may need to use the Cut and Combine function to
temporarily store and preserve such features in a temporary composite and then repaste them back in their
original positions after the above processing.
