[MITgcm-devel] scaling of output fields

[Ed Hill]

Hi folks,

Gael, Dimitris, Chris, and I have talked about how to better handle
precision issues in the output diagnostics.  The basic problems are:

 - writing Real*8 fields results in unmanageably huge output files 
   for Dimitris' high-res integrations, and

 - writing Real*4 fields results in catastrophic precision loss 
   for some diags (eg. Gael is unable to compute salinity variance 
   using SALT and SALTSQ data due to precision loss).

Heres two relatively cheap and easy ways we can improve things:

 1) Add a per-diagnostic option that allows for the selection of 
    either Real*4 or Real*8 output.

 2) Add a mechanism to scale: (scaled_diag = alpha * diag + beta)
    each diagnostic so that we can effectively fit more precision 
    in a Real*4 variable.  This is routinely done in netCDF files
    but it can just as easily be done for MDSIO.

So, to get started, are there any comments on how people would like to
see this implemented?

Ed

-- 
Edward H. Hill III, PhD
office:  MIT Dept. of EAPS;  Rm 54-1424;  77 Massachusetts Ave.
             Cambridge, MA 02139-4307
emails:  eh3 at mit.edu                ed at eh3.com
URLs:    http://web.mit.edu/eh3/    http://eh3.com/
phone:   617-253-0098
fax:     617-253-4464