CWB Project

Welcome to the CWB Project page!

(Updated 30 December 2008)

   
    The status of the tasks for 2008 CWB-NCAR project is listed  in the Table(20080429) <doc file 20080429>
   
    Teleconference between CWB and NCAR on July 24, 2008 at 8:30pm (Boulder time) on Task#1:

         The summary of the teleconference is written by Maxine Chen and reviewed by Y.-R. Guo (Summary, doc file).

        The summary of the teleconference is written by Eric Chiang (Summary, doc file)


    Mid_Term review in NCAR, September 2008

        CWB_status_report (ppt file) by Dr. Terng (20080909)

        CWB WRF_report (ppt file) by Jing-Shang Hong

        CWB HRLDAS report (ppt file) by Jing-Shang Hong

        CWB WRFV3.0 report (ppt file) by Jing-Shang Hong 
 
        CWB DFI 20080821report (ppt file) from Jing-Shang Hong
 
        CWB TC relocation (ppt file) from Jing-Shang Hong

        CWB TC relocate (ppt file) by Der-song Chen

        Review_Task#1 (ppt file) by Yong-Run Guo

        Review Task#2 (ppt file) by Hui Liu

        Review Task#5 (ppt file) by Mike Barlage.


      Basic formulation for WRF and WRF-Var (Mass coordinate) (doc file) documented by Yong-Run Guo
       
A,  Shell  and  code

       Observation processing

             The obsproc was updated same as V3.0 obsproc: 3DVAR_OBSPROC (20080506):

                   Some of the minor modifications: (1) Bug fixed in module_recoverh.F90; (2) Discarded the multiple levels SHIP and BOUY data;
                   (3) Improve the printing stuffs; and (4) improve the MAP_plot. Please read the src/README file.

          ZTD decoder
                   The updated ZTD and PW decoder can be downloaded here  ( GPSPW_decoder). After gunzip and un-tarred GPSPW_decoder.tar.gz,
                  
you can a file: GPS_Note_on_code, from which, you may know something about the code.

                   Also the updated 3DVAR_OBSPROC can be downloaded here  3DVAR_OBSPROC (20080725). From the README file, you can
                   learn what were changed for GPS ZTD. 

                  **  In the current implementation, the default ZTD observation error is set to be 0.5 cm based Lidia Cucurull work and discussed with
                  Chris Rocken (COSMIC) before. Recently, Finnish Meteorologist used the ZTD error variance of about 1.0 cm. This should be tested
                  for CWB Ground-based GPS observations.  

                   ** An example for ZTD data processing can be downloaded from  GPS_proc_exmpl, which includes the shell scripts and the related
                   namelist files, station list file, and GPS data file for date of 20080520. You can use this script with minor modifications to generate
                   either PWV or ZTD data. The decoded little_r file: DPTD_081410.decoded, 3dvar obs file: obs_gts.3dvar_ZTD (or obs_gts.3dvar_PWV),
                  and gmeta file will be produced.

       Data assimilation

             CWB staff reported that SATEM data assimilation may have some problems. After carefully check the WRFVar_2.1 code, and discussed
             with Qinongnong Xiao, the original developer, we found that there are few bugs in the code (see SATEM_WRFvar.pdf). The bug fixed
             code wrfvar_2.1.tar.gz is updated (20080222, recently updated on 11/26/2008).

             In order to improve the typhoon BOGUS data assimilation, the multiple outer loops with the different tuning factors technique was
             introduced to wrfvar_2.1 code. The details for this technique can be found from "Impact of WRF-Var (3DVar) Background Error Statistics 
             on Typhhon Analysis and Forecast" (http://www.mmm.ucar.edu/wrf/users/workshops/WS2006/WorkshopPapers.htm by click P4.2). The code
            wrfvar_2.1_outerloop (updated [20080923], modified da_check_rh.inc on 11/26/2008) and an example of the namelist.3dvar file can be
            downloaded here. Note: currently this code can ONLY be run in serial  mode with single processor. Need more software engineer work to
            make it working in mpi mode future.

        Background error statistics

             BES files were derived based on the forecasts initiated from June 2007 CWB GFS (NODA) by using WRFVar_2.1 gen_be (bintype=5).
             The files can be downloaded from CWB ftp site:

                   datwncr@1632.9.179.171:/guo/be_45km_200706.cv5
                   datwncr@1632.9.179.171:/guo/be_15km_200706.cv5                  
                   datwncr@1632.9.179.171:/guo/be_05km_200706.cv5

            
            
ZTD assimilation
               The wrfvar version 2.1 code is upgraded to include the ZTD assimilation. The code is in the testing stage, and can be downloaded from here
               wrfvar_2.1_ZTD (20081230). In the current implementation, the ZTD observation errors are from the obs file, i.e. the errors vary site by site. With
               minor modification, a constant observation error (.5 cm or 1.0 cm ...) could be set. This may need to try in future.

               The version 2.1 of wrfvar code with ZTD assimilation and outer-loop+BES_tuning is developed here (updated [20081230]version: wrfvar_2.1_ZTD).

                ** Bug fix for da_check_rh.inc to prevent the adjusted relative humidity out of the range {minimum_rh, maximum_rh} (11/26/2008)
                ** Bug fix for da_transfer_xatowrf.inc: qvf2_b = 1./(1.+qvf1) ==> qvf2_b = 1./(1.+qvf1_b) (12/30/2008)

          ** NOTE
                For WRFVar v2.1, now there are two versions:
                        (1) without capability of outer-loop+BES_tuning, and
                        (2) with the capability of 
outer-loop+BES_tuning
                Both of these version give the identical results, and with ZTD assimilation capability and bug fixed.

             Radiance Data Assimilation

             The radiance data assimilation work is supported by CAA project, and Hui Shao, NCAR/DATC, conducted a series of experiments for 2007
             and 2008 CAA project (see https://wiki.ucar.edu/display/mmm/CAA07). She completed the one-month (August 2008)  impact study  and  the
             Typhoon SEPAT forecast experiments  ( CAA_RadianceAssi_HuiShao.pdf).

             The code for WRFV3 and WRFVAR 3.0 used by Hui Shao can be downloaded here. Note that the code structure of this WRFVAR 3.0 is
             somewhat different from different from the  released WRFVar 3.0.  After  setup  all of the environment variables for libraries link, you should
            "cd build", then "./configure" and "./compile all_wrfvar". All of the excutables will be generated under the directory "build".

            The namelist files for WRFVar  can be downloaded here: bias correction preparation: namelist.input, radiance assimilation: namelist.input.
            For details of running radiance assimilation, you may read the user guide for radiance assimilation

            New CRTM library(ibm)  or CRTM_library (pgi) (updated on 20081031) could be downloaded here.
            New RTTOV library (20080828 could be downloaded here.

            The k-shell scripts for running Hui Shao's amsua radiance data assimilation for 2007081918Z case can be downloaded from SEPAT_amsua.ksh
           and da_run_suite_wrapper_caa1-v3.0.ksh.

       Update BC

            Jim Bresch informed me that CWB reported  when low_bdy_only = .true., the code is not correct, and the printed messages are also not make sense.
     
After discussed with Wei Huang, the original developer of this program, the bugs are fixed in WRF_BC_v2.1/src/update_wrf_bc.F, and more comments
      were added in  WRF_BC_v2.1/README. The updated code can be downloaded from WRF_BC.

       WRFVar_v3.0

       WRFVar_v3.0 was released om 4 April 2008. The updated version WRFVar_v3.0.1.1 will be released on 29 August 2008.

       WRFVar_v3.0.1.1 includes some accessory functions, such as gen_be, etc. To test the gen_be function, the scripts for running gen_be and  plotting
        gen_be results are gen_be_wrapper.ksh and gen_be_plot_wrapper.ksh.

        The  released version of WRFDA v3.0.1.1  has been installed in  CWB  outside-firewall  IBM ,  hpcog1: /nwpr/fdda/xb07/CODE/WRFDA.

       The neccessary libraries for compiling the WRFDA have been also installed in pcog1: /nwpr/fdda/xb07/external. To install these libraries, users
        can find all the "setup" files for different platforms/compilers from WRFDA/var/scripts/compile_external. For example, for IBM machine,
        all setup files are located under the directory: ibm_powerpc. With these setup files, it is straightforward to install all of the libraries. There are
        only three things need to pay attention:

            1) To install the rttov library, you need to delete the line "#lapack.f \" in the file: rttov/current/src/Makefile_lib.  
            2) Edit the setup.ksh file for CWB IBM machine in order correctly to set the environment variables linking the libraries.
           3)  To compile the Hui Shao's version of WRFVAR code (an old version), the PATH in .profile file need to be edited to include
                ~/external/makedep90/current.

       An example of the c-shell script, da_run_wrfvar_3.0.csh, and the namelist file, namelist.input, are posted here for running the WRFVar_v3.0 (WRFDA)
       assimilation  with CWB 45-km domain, 2007081300Z case.

             Verification with WRFVar 3.0:

             In WRFVar v3.0, there is a package for verification against the observation and analysis. A variety of plots can be created with NCL
            software. NCAR has tested the verification against the observation data (SOUND and SYNOP) with the CWB sample data files. The
            related shell scripts and a README file can be downloaded from V3.0.1.1_VERIFY.tar.gz, and the testing data files can obtained
            from CWB ftp site: 163.29.179.171:/guo/Verif_V3_data.tar.gz After un-tarred "V3.0.1.1_VERIFY.tar.gz", read the README file first,
            then try to use this package step by step.

      B,  Project progresses

           See Table of status..

       * Since M.-H. Chen came to NCAR, we confirmed that all of the components in the WRFVar_2.1/WRFV2 system
         used in both CWB and NCAR are identical. With the same version of xlf compiler:

         IBM XL Fortran Enterprise Edition V10.1 for AIX
         Version: 10.01.0000.0004

         and the same compiler option: -O2 -qarch=auto, the results from the cold-start runs in CWB(hpc) and NCAR
         IBM(blueice) for 2007080100Z case are identical. The configure files  for WRFV2  is  configure.wrf
         and for WRFVar is configure.wrf.

     C, CWB visitors to NCAR

          Mei-Hsin (Maxine) Chen came to NCAR, Boulder, CO., on 4 April 2008 for 6 months visit.

          Maxine's tentative working plan in NCAR (7 April 2008):

         1, Setup the computer account in all NCAR (SCD and MMM) computer system    two days.
         2, Download the 2007 August OBS data to blueice (NCAR IBM)    one days
         3, Run one day, for example 1 August 2007, WRFVar/WRF in blueice.   
                First guess:  a, NCEP GFS download from NCAR/MSS                      
                                    b, NCEP GFS from CWB    1 to 2 days    
                BES:
                                    a, CWB NODA CV5 BES             
                                    b, CV5 BES  based on 200706  computed in NCAR.             
                                    c, CV3 BES    
                To check the results from NCAR machine and CWB machine are identical,    at least very similar, or not.      total 2 weeks. 
         4, According to the results  above, adjust the namelist settings     one week.
         5, Try the objective BES tuning based on the fort.50 data. Mei-Hsin, do you bring    this data here?     one week
         6, Typhoon forecast experiments: SEPAT?    If CV5 BES results are not satisfied, we may try to multiple-outer loop approach 
            with different tuning factor.     two weeks
         7, GPSRF exps.     two weeks 
         8, WRFVar_3.0       I hope that we take 1 to 1.5 months ti make WRFVar_2.1/WRF_2.2.1 working well,
             the gradually to move on the WRFVar_3.0.
      

    D, Refrences

                 

    E, WRF-DART ----- EnKF


    F. Ground-based GPS measurements

        The progresses of Task 3: Near Real-Time Data Collection and Display of GPS Precipitable Water Vapor Estimates in Support of TIMREX can be seen

             http://taccop3g.cwb.gov.tw/timrex/index.html

    G. Land Surface Model

        Presentation of Land Surface Processes and its Modeling in Mesoscale Numerical Prediction Model (ppt file) by Fei Chen, NCAR/RAL/TIMES
        at CWB, Taipei, on 26 March 2008.

       The HRLDAS presentations, codes, and scripts by Mike Barlage, NCAR/RAL, can be found from

             http://www.rap.ucar.edu/~barlage/CWB_wiki/

    H, Annual report

      



For comments, send email to mesouser@ucar.edu

Last Modified: 30 December 2008