[CIG-SEISMO] why rotating tensor in model_aniso_mantle.f90 in specfem3D_globe

[Zhitu Ma]

Hi, I am trying to benchmark some calculations using an anisotropy model. In model_aniso_mantle.f90, subroutine model_aniso_mantle, the code first calls build_cij to obtain d_ij and then calls rotate_aniso_tensor to obtain c_ij. 

I am wondering what the differences are between d_ij and c_ij.  And the rotation depends on lat/long. I am not sure why this is true because d_ij seems to be constructed from local Gc, Gs etc, which have been rotated to east, north, up already?

Thanks,
Zhitu Ma