[CIG-SEISMO] specfem3d_globe overflow

[Shun Zhang]

Dear all,

I've been trying to use default 1D models(prem, ak135f) to calculate up to ~1 Hz regionally(5 degrees, 240 NEX). 
It failed in the middle and complained that 
forrtl: error (72): floating overflow. 
However, it worked well for NEX = 128(~half Tmin).
The detailed input files and errors are at:
https://github.com/leafriver/specfem3d_globe_test

Thanks,
Shun