FIR normalized warning

Users Mailing List Archive
1

Hi all,

Does anyone know where the warning:

"
WARNING: FIR normalized: sum[coef]=9.781101E-01;
"

comes from when removing the instrument response from some IRIS data?
The interesting part is that sum[coef] seems to be always =9.781101E-01
(even in similar threads:

http://lists.swapbytes.de/archives/obspy-users/2014-January/001157.html ).
Is it some kind of systematic error?

Even though this seems to be a common case, I couldn't find an answer to
this in the internet. Also I think this error doesn't come from obspy
but from the underlying libraries, correct?

Does anyone know the answer or where to find it?

Thank you,

Matthias

2

Hi Matthias

I can confirm that the warning may arise from evalresp library which is used by obspy (source code in .../obspy-0.10.2/obspy/signal/src/evalresp/). "may" since the warning should be followed by some information about the network station, location, channel.

A quick inspection of the code (lines 471-479 in resp_fcnts.c) gives that the FIR filter is expected to be normalized to 1 within a tolerance of 0.02. If not, a warning will be issued and the filter will be re-normalized to full fill this criteria.

regP