2005 CWB project status

Yong-Run Guo

30 July 2005

1, Task 1: Technical support for operational implementation of WRF 3D-Var/NFS at CWB

a, Retrospective experiments in CWB

Ø Summer: 0000 UTC 15 August to 0000 UTC 15 September 2003

§ CWB re-run the GFS (global model) and prepare the 45-km/15-km resolution data from 1) OI+NFS; 2) wrfvar+NFS to NCAR

§ CWB provide the GFS 72-h forecast with 6-h interval from 1200 UTC 28 to 1200 UTC 31 August 2003 and wrfsi code (configuration?) to NCAR

§ CWB do the verification against the analysis for operational run (OI+NFS) and experiment run (wrfvar+NFS).

§ NCAR posts the latest code of the wrfvar system for CWB on web: http://www.mmm.ucar.edu/individual/guo/CWB/CWB_Project.htm

This version of the code has the basic capability of assimilating the CWB conventional data, global bogus data, Typhoon bogus data, and GPS refractivity data.

Ø Winter: 0000 UTC 1 to 31 January 2005

CWB do the retrospective experiment of (wrfvar+NFS) and produce the results of the verifications as did for Summer.

b, Ingest typhoon bogus data into WRF 3D-Var (FM-135 TCBOG)

The latest version of the wrfvar code has already had the capability of assimilating the TC bogus data (40 soundings) from CWB. The problems for further study are

§ The observation error specification needed to be tuned to increase the TC bogus data impact in the final analysis. Currently the default error specifications are 200 Pa for SLP and 2.5 m/s for wind everywhere.

The basic idea is that the smallest OBS errors, for example 50 Pa for SLP and 0.8 m/s for wind, are specified to the center of storm, then the errors are gradually increased to 200 Pa and 2.5 m/s at the outside of the storm. NCAR will develop and test this technique.

§ The temperature in bogus data gave too large bias value of (O-B). Currently only SLP and wind are assimilated, not temperature. How to use the temperature in bogus data is a open question (bias removing technique?). Can the wrf 3D-Var, a multi-variant system, automatically develop the warm core even if the temperature not assimilated? NCAR will do more tests.

§ CWB NFS has the Typhoon re-location technique in the first guess field. To assess the impact of this re-location technique on analysis with the wrf 3D-Var system, CWB should do more study.

c, Ingest of the global model bogus data (FM-135 BOGUS)

The global model bogus data can be assimilated in wrfvar system, but same as TC bogus, the default observation errors are 200 Pa for SLP and 2.5 m/s for wind, and the errors for temperature are same as the sounding data.

CWB (Dr. Terng) suggest to use the same observation errors as soundings. NCAR will modify the OBS preprocessor for this.

Both NCAR and CWB should do the impact study for Global bogus in wrf 3D-Var system.

D, Technical consultancy on background error statistics

§ The code generating the BES for cv_options = 5 was included in wrfvar version 2.1, which has been posted on web. With the 45-km NFS model forecast data (wrf var_NFS) provided by CWB (see Task 1, a), the BES data file has been obtained. According to Dr. Terng (visiting NCAR), CWB will re-run the NFS forecast because some parameters in running the model were not set correctly.

§ NCAR will do the tests with the new derived BES for cv_options = 5. There are two aspects of the work: 1) Single OBS tests in comparing cv_options =5 with the cv_options = 3 in terms of the magnitudes of the increments and the influence range of the OBS; 2) improvements of the analysis and forecast for Typhoon Dujuan case.

§ In addition to 45-km NFS forecast data, CWB will provide 15-km NFS model forecast for the period from middle of August to middle of September 2003. CWB may derive the 15-km BES file by themselves.