Hanli:

Looks like the tgcm21 run w/ mfilsech=12 completed ok,
and the secondary histories look good. I made a run w/
the processor and plotted from all 25 histories. You
should test these histories thoroughly w/ the processor,
but then you can go ahead and continue the run. You may
want to skip history 0,0,0 when plotting diagnostic
fields (tendencies), as they are zeroed out as we discussed
yesterday.  

BTW, I noticed that there was apparently no SDS overflow
this time, so increasing -lQ to 12.0Mw apparently fixed
that problem. I really don't think this is related to the
sec hist bug, but will keep it in mind.

When you continue the run don't forget to comment out
SOURCE, and start SECOUT w/ SECGW04 (even tho there's only
1 history on SECGW03). You should be able to run it out
for 2 more days in the next run, as Ray suggested. If so,
make sure to give enough volumes for SECOUT.
Good luck -- let me know how it goes. I will check in again
later this afternoon.

--Ben