Thread: problem with interpolation

I do not see any response to this message, so I will make a couple of
comments.

Looking at your plots, it looks like you have high amplitude information
at high frequencies. The explanation to your problem may simply be that
your data are aliased when you downsample. I suggest you try repeating
your runs using decimate with the antialiasing filter on. Command
interpolate has no antialiasing.

On Tue, 30 Nov 2010, maria manakou wrote:

Hello,

I face some problem when applying the command "interpolate delta XX" in
some noise data.

I have some 1-hour noise data recorded in CMG-5 guralp sensors and I want to
interpolate them for further processing in SAC software. The raw *.cgf files (the
format of the CMG-5 sensors) have already transformed to *.sac files through
Scream V4.4 software provided by Guralp.

The *.sac data (un-interpolated data) have delta 0.01sec and I want to
interpolate them to delta 0.05sec. I use the following commands in my SAC
8/8/2001 V. 00.59.44 (Copyright 1995) software which runs in Fedora 9.0:

r *.sac
qdp off
rmean
rtrend
p1
interpolate delta 0.05
w change .sac .inter

When I plot the un-interpolated (*.sac) and interpolated data (*.inter) to see if
the interpolation works ok (Fig. 1 in the attach file) the result seems ok, but
when I zoom in different time windows the result is not at all good! Specific
spikes in my un-interpolated noise data seem to "move" in time in
the interpolated data as you can see in the attach doc file. This "moving" seems
to happen after some time (e.g. in fig. 2 the coincidence of specific spikes
between un-interpolated and interpolated data is very good, but this does not
happen for the rest time windows,fig. 3-5).

I cannot understand why this happen.
If someone has faced this problem again and have any idea, please let me know.

Best Regards,
maria manakou