Clarify frame of reference for SF and SS in MS_phasevels

[andreww]

This was queried by a user, my answer was:

I think the vector is in the coordinate system you are after and you don't need to transform it. The
reference frame for the unit vectors describing the polarisation direction (displacement vector) of the
fast quasi-shear wave, SF (and the slow one, SS) is the same as that used to describe the elastic
constants tensor. So if SF = [ 1 0 0 ] and SS = [ 0 1 0 ] the sense of particle motion is to strain the
material in the sense resisted by C(1,1) and C(2,2), respectively. The wave propagation direction will
be along (or, if the symmetry is low enough, close to) the X3 direction (inc = 90.0 and the result should > be independent of azimuth).

but some clarification in the documentation would be useful.

Also, why don't we output the qP polarisation direction?

[jwookey]

That looks right to me. No particular reason that P-polarisation is output, just haven’t really needed it for anything up to now.

Cheers,

James

On 4 Sep 2014, at 10:03, Andrew Walker [email protected] wrote:

This was queried by a user, my answer was:

I think the vector is in the coordinate system you are after and you don't need to transform it. The
reference frame for the unit vectors describing the polarisation direction (displacement vector) of the
fast quasi-shear wave, SF (and the slow one, SS) is the same as that used to describe the elastic
constants tensor. So if SF = [ 1 0 0 ] and SS = [ 0 1 0 ] the sense of particle motion is to strain the
material in the sense resisted by C(1,1) and C(2,2), respectively. The wave propagation direction will
be along (or, if the symmetry is low enough, close to) the X3 direction (inc = 90.0 and the result should > be independent of azimuth).

but some clarification in the documentation would be useful.

Also, why don't we output the qP polarisation direction?

—
Reply to this email directly or view it on GitHub.

[andreww]

Except what we actually return in SF and SS is the eigenvectors projected onto the plane normal to the propagation direction (we return S1P and S2P calculated from:

  %  ** calculate projection onto propagation plane      
  S1N = V_cross(XI,S1) ;
  S1P(ipair,:) = V_cross(XI,S1N);
  S2N = V_cross(XI,S2) ;
  S2P(ipair,:) = V_cross(XI,S2N);

I think this is because this is the vector we need to plot in e.g. MS_sphere (so the vector is tangent to the sphere) but isn't what you would get from, for example, doing finite difference of wave propagation. Maybe what we should return is S1 and S2 (the sorted eigenvectors, which are, I think, what I've described above) and push the projection into MS_sphere and MS_plot?

But we do need to do the projection calculation inside MS_phasevels in order to calculate pol (which is in the ray frame). It's just a question of what we need to return to the caller - maybe both S1 and S1P should be returned (this can be done in a backwards compatible way).

[alanfbaird]

For computing group velocities it would be useful for MS_phasevels to output the non-projected eigenvectors, and possibly also the slowness vectors (i.e. XI/V(i)).