[CIG-SEISMO] Homogeneous earth model

[Dimitri Komatitsch]

Hi Stefan,

Thanks for your question. There are two problems related to using a 
constant velocity model:

1/ the resolution of the mesh in SPECFEM3D_GLOBE is adapted to PREM,
i.e. the mesh is doubled (mesh density is divided by 2, i.e., mesh cells
become twice bigger) several time (3 or 4 times) with depth; therefore
if you use a constant velocity model this will result in poor sampling 
in some regions of the earth (unless you use low frequencies only for 
the source, say 40 seconds or above, but no energy below) (you can do 
this easily in postprocessing: convolve the Green seismograms from 
SPECFEM3D with a low frequency source); you won't be able to run high 
frequency comparisons

2/ indeed, we use a fluid velocity potential in the outer core, which 
implies Vs = 0 (by construction); there is no way we can set Vs to 
something different than zero in the outer core without major 
modifications in the source code.

Therefore what I suggest you do is keep three regions: the mantle, the 
outer core and the inner core, keep the outer core fluid, and make each 
of the three regions homogeneous (with different values in each, e.g. 
not too far from the average of the PREM values in each region).

Of course this would imply using a fluid-solid formulation in DG,
e.g. something along the lines of the Wilcox-Hesthaven-Ghattas JCP 
preprint we discussed in Munich two weeks ago.

Hope this helps,
Dimitri.

On 07/09/2010 09:27 AM, Stefan Wenk wrote:
> Dear CIG-Seismo Team,
>
>
> this is Stefan Wenk from Munich University. Recently we've tried to compare SpecFEM3D with
> our DG Code on global scales. Therefore, the comparison we would like to start
> as simple as possible, using a homogeneous earth model.
>
>
> For that reason I've tried to change the isotropic PREM routine model_prem.f90
> in the source code and implemented constant material values for rho, vp, and vs
> instead of the polynomials. The compilation worked fine but the code crashed due
> to instability problems. I think this is caused by the constant S-wave velocity
> unequal to zero I've chosen in the outer core.
>
>
> Does the calculation of the potential you do in the outer core assume vs=0? Is
> there any possibility to circumvent this issue without huge changes in the code?
>
>
> Best regards,
>
> Stefan
>

-- 
Dimitri Komatitsch - dimitri.komatitsch aT univ-pau.fr
Professor, University of Pau, Institut universitaire de France,
CNRS and INRIA Magique3D, France   http://www.univ-pau.fr/~dkomati1