[CIG-MC] CitcomS: layers

[Eh Tan]

Tobias Höink wrote:
> Hi Eh,
>
> Sorry to bother you again today. I have another confusion about the
> layers in CitcomS (visc_layer_file=off).
>
> There are (by default) 4 layers (and properties are assigned with 4
> values of visc0, viscE, etc.), but how are they given by by z_lith,
> z_410, z_lmantle and z_cmd? 
>
> I had assumed that CitcomS started from the top, i.e. 
> layer 1  is between radius_outer and z_lith,  
> layer 2 is between z_lith and z_410,
> layer 3 is between z_410 and z_lmantle 
> layer 4 is between z_lmantle and z_cmb 
>
> Question 1: How is the base, i,e. between z_cmb and r_inner defined? 

Hi Tobias,
(cc to cig-mc)

Layer 4 is between z_lmantle to radius_inner. The z_cmb only controls
the depth of pv-ppv phase change and doesn't affect the viscosity layers.

> Question 2: My assumption seems wrong. 
> Trying visc0=1e1,1e-2,1e0,1e0
> with 
> z_lith=0.045
> z_410=0.1            #also depth of 410-phase change
> z_lmantle=0.15        #also depth of 670-phase change
> z_cmb=0.439           #     depth of another phase change
>
> I expected to prescribe a low visc layer between r=0.98 and 0.955
> (plot 1), but from the profile the step seems to cover layers 2+3.
>
> Can you give a quick idea what I am doing wrong? 

I think you mean that the low visc layer is between r=0.9 and 0.955,
given your z_lith and z_410.

Your viscosity profile looks correct to me. This is a good question that
is not explained anywhere in the manual.

The viscosity is defined on quadrature points and will have the
discontinuous layered structure in the Stokes solver. For
output/visualization purpose, the viscosity is interpolated to nodes. As
a result of interpolation, the nodal viscosity will show a smooth
variation across the (actually) discontinuous viscosity layers.


-- 
Eh Tan
Staff Scientist
Computational Infrastructure for Geodynamics
California Institute of Technology, 158-79
Pasadena, CA 91125
(626) 395-1693
http://www.geodynamics.org