CWB Project

Welcome to the CWB Project page!

(Updated 12 November 2010)


A,  WRFVar Shell  and  code

      End-to-end testing run

          Obseevation decoder

              Observation distribution(pdf) from 2009060818Z NCEP prepBUFR data file.
                   It is interesting that most of GPSPW data are from Taiwan. The wind profiler data are most fro Japan, but one from Hong Kong.
                   These two types of data are not included in CWB WRFDA/WRF syste.

              Observation distribution (pdf) from 2010022400Z NCEP prepBUFR data file.

              CDAAC GPSRO BUFR data decoder: gps_bufr_decoder.tar.gz

                   This decoder program can generate two types of the LITTLE_R files: "profile" type and "one_lev_per_record" type. If user wants
                   to consider the ray-smearin in GPSRO data assimilation, he/she should use the "one_lev_per_record" type LITTLE_R file.

              Data assimilation
               
                   You can downloaded the Test version of WRFDAV3.2 (1 July 2010) :
                     
This version is from the trunk code Revision 4335 plus the two modified subroutines:
                            var/da/da_physics/da_transform_xtogpsref_lin.inc
                            var/da/da_physics/da_transform_xtogpsref_adj.inc
                      The modifications are for the GPSRO refractivity N observation operator ignoring the pressure perturbation. From the high level (about 15km, eta_level=37) single N tests,
                      we found that the significant moist increment at the low levels was created. Most likely this low level moist increment is from the pressure increment, so we ignored
                      the P-perturbation in the N operator. This is not a final decision. Hope to have more testing results to confirm if this mods can improve the GPSRO assimilation.


                  Updated version of WRFDA V3.2 (17 October 2010):
                      1) The main update is to implement the conversion of the GPSRO geometric height to the geopotential height, which will reduce the positive bias of (O-B) above 12km;

                          Ted Iwabuchi tested this versio of WRFDA with the conversion code for period of 2010022400Z to 2010030918Z. The results are IWABUCHI_height (ppt file).
                          Note that the logarithm vertical interpolation has NOT been implemented in this version.

                           If you want to re-consider the P-perturbation in the GPSREF operator, you can replace

                                    /WRFDA/var/da/da_physics/da_transform_gpsref_lin.inc  with /WRFDA/var/da/da_physics/da_transform_gpsref_lin.inc.org
                                    /WRFDA/var/da/da_physics/da_transform_gpsref_adj.inc with /WRFDA/var/da/da_physics/da_transform_gpsref_adj.inc.org

                      2) The capability for outer-loop with the variable tuning factors for CV3 BE has been implemented, but not fully tested.  If you wat to use this  capability,  you should use

                                   /WRFDA/var/da/da_setup_structures/da_setup_be_ncep_gfs.inc.new  to replace  /WRFDA/var/da/da_setup_structures/da_setup_be_ncep_gfs.inc
                                   /WRFDA/var/da/da_main/da_solve.inc.new to replace /WRFDA/var/da/da_main/da_solve.inc

                            The namelist.input setting for CV3 BE tuning factors: as1, as2, as3, as4, and as5 is explained in CV3_BE_Tuning (ppt file).

                Training courses in CWB

                    Yong-Run Guo gave 3 talks during his visit to CWB in March 2010:

                    1, Adjoint development techiniques in WRFVar: WRFVar_Code_development_I.ppt on the basic techniques of tangent linear and adjoint code development;
                    2, Observation operators in WRFVar: WRFVar_code_development_II.ppt on operators of GPSRO, GPSZTD, and surface 2-m T, q, 10-m wind, and Psfc;
                    3, Formulation to diagnose the pressure and geopotential height in WRFVar: GeoHt_calc_March2010.ppt

                      Sample of program for RH tangent linear code correctness check.
                      Sample of program for adjoint code correctness check.

                Results:

                    Surface data assimilation code update by Eric Chiang, CWB, 11 November 2010: Chiang_Surface_Work.ppt

                    GPSRO data assimilation work by Pai-Yu Chen, CWB, 11 November 2010: (1) H-correction.ppt; (2) H-correction+P-perturbation-off.ppt.

                    We conducted the tests for 2009060818Z case for assimilating the different types of the GPSRO data:
                         (1) wetPrf in "profile" type used in CWB OP211;
                         (2) bfrPrf in "profile" type used in NCAR tests;
                         (3) bfrPrf in "one_lev_per_record" type used in GSI GPSRO BUFR input.
                     The results can be found from GPSRO_Data.ppt

          WRF v3.2:

B, WRF-DART ----- EnKF


C. Ground-based GPS measurements

D, COSMIC II (OSSE)

E. Land Surface Model (HRLDAS)

    
H, Annual report

        TASK#1: TASK-1_review_June (ppt file)


For comments, send email to guo@ucar.edu

Last Modified: 12 November 2010