[RegCNET] SST interpolation crossing date line

[Raymond Arritt]

Dear RegCNET friends,

I'm attempting to run RegCM-4.4rc8 for a large North America domain 
using HadGEM2 initial and boundary conditions.  When running the 'sst' 
preprocessing step the code fails with the error message
NEST DOMAIN TOO NEAR TO PARENT.

I've traced this back to the part of mod_interp.f90 that searches for 
the global data grid points that surround the RegCM4 grid points.  
Through reading the code and some trial-and-error I'm almost certain 
that this arises because the northwest corner of the domain crosses the 
date line, and the code does not account for the wraparound of 
longitudes at +/-180.  That is, the code does not recognize that (for 
example) RegCM4 grid longitude 179.888 is near the global data longitude 
-180.

I have already done a large suite of current climate runs for the same 
domain that were driven by ERA-Interim.  These all worked fine, because 
'sst' performs the interpolation for ERA-Interim without having to 
search for surrounding points.

Before rewriting mod_interp.f90 to account for such cases I would like 
to ask if anyone else already has done so, or if there is another fix 
you could recommend.

Thanks for whatever help you can suggest.

Ray Arritt