Time: 03:30~05:00 pm, Nov 15,
2005
Participants: Dir. Kuo, Dr.
Huang, Dr. Guo, Shui-Yong Fan
Dr.
Huang and Shui-Yong have started to prepare the project review to be presented
to BMB at the beginning of December. Shuiyong showed the draft presentation.
Dir. Kuo and Dr. Guo provided their comments and suggestions. Dr. Huang and
Shui-Yong will continue to work on the presentation and should have a modified
version next week.
Shui-Yong
showed the verification diagrams of EXP07
and EXP08 and EXP09.
EXP07:
ncar3dvar with BMB BES. And
namelist.3dvar is the same as that of BMB OPR(operational system) except
len_scaling=1 and var_scaling=1. This means no factors will be added to
background error.
EXP08: an
experiment with Cressman analysis as another reference in addition to EXP0
(OPR) and EXP4 (NoOBS).
EXP09: wrfvar
with new BMB BE generated by gen_be for mm5 with half-month series of BMB OPR
forecast fields. Namelist options are len_scaling=1
/var_scaling=1/chech_rh=2/sfc_assi=1.
The
scores of EXP08 with Cressman analysis are worse than those from most of 3D-Var
experiments. Dr. Huang was pleased with the results as he and Chen Min had the
opposite results in the MM5DFI paper.
It
is very encouraging to see EXP09 has the best scores for all upper air
variables. The surface scores are not the best, but comparable with EXP0 (OPR).
This is the most important progress made so far in the project, as EXP09 uses
the latest version of WRF-Var and the new BE generated by gen_be_stag0 from the
mm5 (as the BMB operational system) forecasts. This will ensure the future BMB
operational system benefits from all future development of WRF-Var. This should
be the starting point for our next round of tuning. Shui-Yong found different
SFC_ASSI options give very different assimilation results. Dr. Huang suggested
that one more experiment with the other SFC_ASSI should be done.
Shui-Yong
has done more tests to the code gen_be for mm5, including T/RH/Psfc difference
field tests. There are no problems found in the current code. Shui-Yong got a
new BE with the code and did an experiment with this new BE (EXP09).
Shui-Yong
also showed some BE structure function plots but more details about the new BE will
be investigated.
Shui-Yong
tried two methods to get the IC of WRF model from CMA T213 global data: one is
changing T213 data to wrfsi intermediate data format and then using wrfsi; the
other is changing MM5 REGRID output to the format wrfsi can deal with by using
the transform package mm5_p_convert. But T213 has no soil temperature and wrf
model cannot run without soil temperature (from Wei Wang). And some cases make
error and damp with the second method.
Next meeting will be at 2:00
pm Tuesday 29 November 2005.