Geoff:

That is indeed a strange one. When I tried my rdtgcm utility on 
/ROBLE/RGR95/TSALP05 it also died, but with operand range error.
Then I put the new processor on it and it ran fine. Then, I reran
rdtgcm on it and it also ran fine. Maybe it was a memory problem,
although those usually dump in hpalloc.

This is a tgcm20 run, meaning our new version of the dynamo time-gcm.
There is no reason I know of that timesproc would not accept it, but 
I am going to give you the new processor, as it appears to work ok
on this volume.

The new processor is called tgcmproc, and is the most "generic" of
them all, so you should never have to use another post-model processor 
for the rest of your life, and then some :) (yeah, right)
It is easy to set up:

1. Make a directory ~/tgcmproc (probably also a /d/crowley/tgcmproc)
2. Copy ~foster/tgcmvis/tgcmproc/newuser/* to ~/tgcmproc.
   This will be 3 files:

   a. README (small documentation file for new users -- please read)
   b. tgcmproc_sut (script to run the processor in "single ut" mode,
      e.g., maps, slices, etc.
   c. tgcmproc_mut (script to run the processor in "multi ut" mode,
      e.g., plots with ut on x-axis.

3. Set up tgcmproc_sut and/or tgcmproc_mut for the run you want
   (read comments after semicolons), and submit either one via MIGS 
   in same way as previous jobs. (be sure to set sendcgm and senddat
   appropriately to get output, and that the specified directories 
   exist).

Note there are several "emissions fields" not in previous processors
that you may be interested in.

I am working on fuller documentation of each keyword=value input
parameter, but its not done yet, so call/write me if you have questions.

I am also working on a concurrent model processor for tgcm20 which
you may want to use later. tgcm20 will be the base model for subsequent
development.

--Ben