The Nighth BMB-NCAR Project Meeting

Time: 02:00~04:00 pm, Oct 12, 2005

Participants: Dr. Huang, Dr. Guo, Shui-Yong Fan

 

1.   Discuss the project progress

á      Parallel runs

Shui-Yong showed the diagrams of EXP03 on the meeting. The scores of EXP02 were best in all of experiments. The scores of EXP03 were the same as EXP00 that the analysis scores were better but forecast scores after 12h became worse.

  

   EXP03: replace ncar3dvar by wrfvar_v2.1. Shui-Yong Found when the option SFC_ASSI is 2, wrfvar_v2.1 has some problems, and after the discussion with Dr. Guo, he changed the SFC_ASSI option to 1. Other configurations of EXP03 are the same as those of ncar3dvar.

 

á      Gen_be work

Shui-Yong completed the code of transform test from Psi/Chi to U/V and U/V back to Psi/Chi for MM5. The difference of Psi to transformed Psi is the same as that of WRF.

á      Verification

Shui-Yong also showed some statistics of EXP03, including initial and final cost functions and the ratio J_min/N_obs. Dr. Huang and Guo thought that the analysis of EXP03 are too close to observation and It made the analysis scores better but forecast scores worse. They suggested that the background error-δb should be decreased to make the analysis a little closer to background field.

 

Shui-Yong also showed the results of EXP03 single_obs test. Dr. Huang and Guo found the T response to single T obs had an unreasonably big negative value. After discussing with Dale, they thought that the bad balance and different vertical levels could give wrong thing by using the interpolated BES (Shui-Yong has proved this by single_obs tests with KMA domain and KMA BES and with BMB domain and BMB BES). So they suggested that the BES derived from the same model forecasts and the same synoptic situation should be used.

 

Shui-Yong also showed the results of precipitation verification of two cases (05060612 and 05060700). The most of TS scores of EXP03 were better than EXP00 and EXP02.

2.   Make a task list

á      Do observation quality control for a week to get a set of good observations. This set of observations can be used for the verifications of all experiments.

á      Run more experiments, including with no3dvar/only sounding observations/only surface observations/BMB BES, et al., and just only domain1.

á      Clean up the diagrams of precipitation verification to make them easier to compare.

á      Continue to write some texts for the results obtained.

á      Speed up the MM5 Stage0 work. 

 

3.   Next meeting

 

Next meeting will be at 2:00 pm, Oct 26, before Dr. Guo leave.