[RegCNET] Problem in Latent Boundary Condition with latest version of RegCM3

[Thang Luong]

Dear RegCNET users,

I ran the latest version of RegCM3 to test the simulation differences
between a long run a short one.
My expectation was the results must being the same for to cases.
Configurations are set for 2 experiments. First, 4 months from
1996050100 to 1996090100.
Second, 9 months from 1996010100 to 1996090100.
I took monthly mean of skin temperature for comparison. It's turned to
be very different between to experiment as comparing last month
results. Somewhere is the ocean, the differences are larger than 7
degree. This doesn't happen with older 2006 version.

I'm running all experiments on a RH EL machine with parallel option.
RegCM3 was compiled with PGI_7.1.4 and mpich2_1.0.7. All cases using
default regcm.in except changing igcc from 2 to 1 and iocnflx from 2
to 1. Input data are exactly the same.
Any idea?
-- 
Thang
--------------------------------------------
Thang Luong, B.S.
Vietnam National University at Hanoi - College of Science
Department of Meteorology

Tel: +84912693936
Add: No 185/26/18A Minh Khai st., Hanoi, Vietnam
Email: thang.luong at gmail.com
--------------------------------------------