Thread: less smoothing of MUSTANG noise spectra

Started: 2019-09-09 19:50:09
Last activity: 2019-09-09 19:50:09
Topics: MUSTANG Users
Carl Tape
2019-09-09 19:50:09

I wanted to raise a topic that you are probably aware of. (ASL folks
pointed this out to us.) For some applications and interpretations, the
MUSTANG-generated PSDs are overly smoothed. There is a nice test station as
an example: AK.NEA2, which has a 38 m high microwave lattice tower nearby
that had a natural frequency of 2.5 Hz (0.4 s). The sensor is a TA upgrade
(cased borehole into bedrock).

1. The spectral peak is very sharp, as shown in the attached example plot,
which I have flipped for comparison with the other plot (x-axis frequency).
No smoothing is applied to calculate the spectrum. The peak is sharp.

2. From the MUSTANG generated-spectra, you see a mesa-like shape that
arises from the octave smoothing. (You can also see that this "mesa" is
enhanced during windy conditions, which is what you might expect from a
lattice tower.) In this case, the smoothing is detrimental to the

This is part of a recently accepted paper by Kyle Smith and me:

We discuss the issue of smoothing in the Discussion. It is helpful to be
using a data product, which others can directly compare with. However, in
our case (and in others), there is a strong argument against using the
smoothed MUSTANG spectra, and instead calculating the spectra ourselves,
without the smoothing.

Here at the SCEC meeting I noticed other sharp spectral peaks -- from
sedimentary basins recorded by nodal stations. These, too, would get
smoothed out in MUSTANG.

My question is:* Could MUSTANG deliver a less-smoothed version of the PSDs?*
Or perhaps a couple different versions that could be flagged for download?

Thanks very much for this product!


P.S. If you have feedback or questions on Kyle's paper, please email him at
kksmith7<at> (and please CC me too!).

Carl Tape
Associate Professor
Geophysical Institute (office 413D)
University of Alaska Fairbanks
Phone: 907-474-5456
Email: ctape<at>

23:17:12 v.22510d55