BMB-UCAR WRF 3D-Var project

Summary for 23 September13 October 2006

Reporting time (for this report): 15 October 2006

 

 

Task 5. Meteorological performance

5.1. BE

 

Shuiyong re-calculated the BE for 27 and 9km domains using SI+ wrfvar (BE interpolation) + WRFv2.1 with the operational model forecasts in August 2006. The interpolated 9km BE was also tuned using the factors 1.5/1.5/2/2/2. He analyzed the structure of 9km BE and performed single-point observation and AWS+GPS data assimilation experiments. According to the results of experiments, except for specific humidity, the BE structures are reasonable for most of the variables except humidity, which needs more time to investigate.

 

5.2. WRFDFI

 

Min has performed the cycling tests for the three options of moisture filtering for DDFI. But no report is prepared this time.

 

5.3. Comparison

Min and Zuofang has submitted a report about the verification for the operational precipitation forecasts of MM5 for the summer of 2006. The results showed that the scores of the forecasts initiating from 1200UTC are better than those from 0000UTC for whatever the verification thresholds, that is, using the interpolated AVN analysis as initial condition may lead to better precipitation forecasts skill than objective analysis on the background of AVN forecasts. In addition, the forecast skills for precipitation under different types of synoptic systems are verified. In general, the precipitation forecast skills are greatly improved comparing with last year, which can be regarded as the results of the improved model configurations and the shift of background from T213 to AVN.

 

The performance comparison between operational MM5 and pre-operational WRF runs during the summer of 2006 is still ongoing. The results will be reported next time.

Task 6. Monitoring of pre-operational runs and case studies

 

6.1. Verification

 

The codes relating with vertical level verification has been prepared. But further tests will be performed.

 

6.2. OSE

 

Jiqin has confirmed the types of observation data we can acquire and decode in real-time. The real-time data collecting system will be ready in short time. She also made a plan for the OSE task: 1) to get some idea about the number and distribution of observations which can be absorbed by WRFVAR; 2) to get some idea about the observation quality with statistics of O-B; 3) simple correction for observation bias; 4) numerical experiments.

But thereÕre still some problems need to resolve: 1) is there any difference between the GTS data collected from BMB and from NCAR? 2) what is the framework the OSE numerical experiments will be performed under (WRFSI+WRF or WPS+WRF)?

Dr HuangÕs comments: Jiqin should start her GTS experiments using the latest system: WPS+WRF according to Shuiyong's definition.

 

6.3. Cycling

 

Min started to work on the operational cycling system script. But no report is prepared this time.

 

6.4. Tuning

 

Yongrun found the problem in the interface of WPS and WRFVAR. WPS gave the overshooting in terrain interpolation along the coastal line, so there are many negative values in terrain field and PHB. This caused the floating point in WRFVAR. Yongrun provided a GEOGRID.TBL file as a temporary solution to this problem. He promised to fix the WRFVar code to account for the negative terrain values in near future.

 

 No report is required from Zuofang.

 

6.5. MM5DFI

 

The work was planned to start next year. No report is required.

 

Next report time: 27 October, 2006.