Case 2: ARW Case with conventional data (PrepBUFR) plus other data

Case 2: ARW Case with conventional data (PrepBUFR) plus other data cindyhg Tue, 08/20/2019 - 12:55

Case 2: ARW Case with conventional data (PrepBUFR) plus other data

Introduction

This case of 00z analysis on February 13th, 2014 was used here as the example case to introduce how to setup run scripts for the GSI observer and EnKF and how to check the EnKF analysis results to determine if a particular EnKF application was successful. In this case, in addition to conventional observations from PrepBUFR, satellite radiance data are also assimilated.

Case Data

The information on downloading the case data can be found from this link.

Setting Up the Case Data and Run Directory

  • Create the data directory ./enkfdata and download, untar the data files. The case data is now populated in:

               (PATH)/enkfdata/arw 
          
  • Create a working directory or use an existing directory as the working directory.
  • In this case, we use run as working directory. 

Setting Up the GSI Observer Run

Setting Up the GSI Observer Run cindyhg Tue, 08/20/2019 - 12:57

Case 2: ARW Case With Conventional Data (PrepBUFR) Plus Other Data

Setting Up the GSI Observer Run

Running GSI as observer is the same as running GSI as 3DVar analysis, except for couple of run scripts setups. Users should create a a basic GSI run script following the GSI tutorial: Prepare Base Run Script.

Based on run_gsi_regional.ksh_basic, make the following changes to run GSI as observer:

  • Define run GSI observer only and ensemble members
  • When set if_observer=Yes, GSI only runs observation operator part and skips the minimaztion:
      if_observer=Yes
      no_member=20
      BK_FILE_mem=${BK_ROOT}/wrfarw.mem
    
      
  • For this exercise, in addition to conventional observations, satellite radiance data are also assimilated. To do so, un-comment the links to the other observations
      ln -s ${srcobsfile[$ii]}  ${gsiobsfile[$ii]}
  

The example script can be found : here 

Run GSI observer

Run GSI observer cindyhg Tue, 08/20/2019 - 12:58

Run GSI observer

How to run GSI job depends on computer system. On Cheyenne (LINUX with PBS), run it by: 

   qsub  run_gsi_regional.ksh

 

Check the GSI observer run results

Check the GSI observer run results cindyhg Tue, 08/20/2019 - 12:58

Case 2: ARW Case With Conventional Data (PrepBUFR) Plus Other Data

Check the GSI observer run results

cd gsidiag_arw/. 
 

  • Check the stdout file from member 001 and compare it to the reference stdout here 
  • Check the files in the run directory for member 001 before the directory clean and compare them to reference file list here
  • Check the files in the run directory for member 020 before the directory clean and compare them to reference file list here
  • No need to check fit files and minimization
  • should have diag files for the previous 19 members and mean

Setting Up the EnKF Run

Setting Up the EnKF Run cindyhg Tue, 08/20/2019 - 12:59

Case 2: ARW Case With Conventional Data (PrepBUFR) Plus Other Data

Setting Up the EnKF Run

Here, we will setup the the EnKF run script based on the basic run script from Prepare basic EnKF run script. After copy: 

cp run_enkf_wrf.ksh_basic run_enkf_wrf.ksh 

Make the following additional modifications to the script run_enkf_wrf.ksh:

  • Define ensemble parameters
  NMEM_ENKF=20
  BK_FILE_mem=${BK_ROOT}/wrf_en
  NLONS=129
  NLATS=70
  NLEVS=50
  
  • define ensemble background and observations used in analysis
  IF_ARW=.true.
  IF_NMM=.false.
  list="conv amsua_n15 amsua_n18"
  

 

The example script can be found : here

Run EnKF

Run EnKF cindyhg Tue, 08/20/2019 - 12:59

Case 2: ARW Case With Conventional Data (PrepBUFR) Plus Other Data

Run EnKF

It can be run with the same way as GSI runs, for example on Cheyenne (with PBS): 

   qsub   run_enkf_wrf.ksh

 

Check the EnKF run results

Check the EnKF run results cindyhg Tue, 08/20/2019 - 13:00

Case 2: ARW Case With Conventional Data (PrepBUFR) Plus Other Data

Check the EnKF run results

cd ${WORK_ROOT} 
 

  • Check the stdout file and compare it to the reference stdout here 
  • Check namelist file in run directory and compare with reference namelist file here