[CIG-SEISMO] Element size specfem3d cartesian

[Moritz Wehler]

Hey there,

I am currently setting up a 3d model with geocubit (using specfem3d 
cartesian). Concerning the element size I was told - as a rule of thumb 
- to use the following formula:

     elem_size = 0.6 times wavelength.

Is the factor 0.6 correct and if so, where does it come from 
(literature?, empirically derived?...)?
If not, what would you suggest to use?

Thanks in advance and best regards,
Moritz