A suggestion and a question about subsource codes

[javiquinte]

On behalf of Frederik Tilmann

For subsource codes of seismometer

• For 1,2,Z or 1,2,3 it is not clear whether there is an expectation
  that they form a LHS coordinate system (as NEZ). I think this should be
  imposed, but in any case the spec's should be clear whether both LHS and
  RHS are possible or only LHS

• What's the difference between triaxial components (A,B,C) and raw
  triaxial components (U,V,W)?

[chad-earthscope]

@ftilmann

For 1,2,Z or 1,2,3 it is not clear whether there is an expectation that they form a LHS coordinate system (as NEZ). I think this should be imposed, but in any case the spec's should be clear whether both LHS and RHS are possible or only LHS.

There is this wording:

... if the orientation of the horizontal components is known to deviate more than 5 degrees from true North and East, the respective channels should be named 1, 2 instead of N, E (N->1, E->2).

which identifies a mapping from N,E to 1,2. It could likely be clearer, suggested wording and placement to avoid confusion would be appreciated.

What's the difference between triaxial components (A,B,C) and raw
triaxial components (U,V,W)?

I believe the difference is that ABC is specifically lower edges of a corner-down cube, whereas 123 is a superset case of any non-traditional orientation. I have not seen ABC in use; instead I have seen UVW used often for the "raw" components of a triaxial (e.g. STS-2) sensor. I wondered the same when transcribing them from the SEED manual (which is replete with such things if one looks very closely). I'm happy to drop ABC or clarify otherwise if there are suggestions.

[ftilmann]

There are really two issues, the first one related to 123 being LHS or RHS, the second one related to ABC vs USV. I am sorry for potentially ambiguous phrasing linking the two; the difference between 123 and ABC is clear. So the first issue:

.. if the orientation of the horizontal components is known to deviate more than 5 degrees from true North and
East, the respective channels should be named 1, 2 instead of N, E (N->1, E->2).

which identifies a mapping from N,E to 1,2. It could likely be clearer, suggested wording and placement to avoid
confusion would be appreciated.

I was thinking more of OBS, where there is no deviation from N, E and the horizontal orientation is totally random. Or even for the general case when 3 might not be vertical. My suggested phrasing would be

"1,2,3 must describe a left handed-coordinate system (the same as N, E, Z)."

It's not a big deal really, but defining this clearly would simplify a little bit the requirement for routines for rotation, and philosophically clear recommendations are preferable.

EDIT: I just realise that I always think of 1,2,3 as an orthogonal system, but I guess the wording does not really imply this. If it's meant as a fully flexible system of three axes, not necessarily orthogonal, then probably not reasonable to impose handedness.

[ftilmann]

ABC vs UVW:
I am no expert on seismometry but I think the Galperin configuration of STS2 is orthogonal and all components at same angle to the vertical/surface, which would be equivalent to lower edges of cube sitting on a corner. And most non-derived seismogram traces are 'raw'. Like you, I have never encountered ABC either, and only ever seen UVW.
I guess no great harm done to call the same thing by two different names, but it seems wasteful.

[chad-earthscope]

"1,2,3 must describe a left handed-coordinate system (the same as N, E, Z)."

I like that. We can add "when orthogonal" to clarify. I also think of them as orthogonal and pretty sure most others do as well, but unsure if that's universal enough to add to the spec.

I guess no great harm done to call the same thing by two different names, but it seems wasteful.

Agreed, it is wasteful. Unfortunately it's legacy and may be in use I would be comfortable adding "deprecated, use UVW for this case" in the ABC description if others agree.

[chad-earthscope]

It's been pointed out that 1,2,3 are already clearly documented as orthogonal, so we don't need the "when orthogonal" part.