Case 7: 4D Hybrid EnVar Case

Case 7: 4D Hybrid EnVar Case cindyhg Tue, 07/16/2019 - 10:59

GSI 4D HYBRID FOR ARW USING GLOBAL ENSEMBLE FORECAST

Introduction

This exercise runs the GSI 4 Dimensional Ensemble-Variational (EnVar) hybrid analysis with the ARW background and conventional data at 18z August 12, 2018. 

Please note the ARW background field is provided at three time levels in netcdf format from case 6, and the NAM BE is employed as the background error covariance in this experiment. The global ensemble files at three time levels are linked to run this GSI 4DEnVar hybrid test.

Please check the Download Practice Data section if need to obtain the background, observation, and ensemble forecast files.

Setting up the Run Script for GSI 4D hybrid analysis

Setting up the Run Script for GSI 4D hybrid analysis cindyhg Tue, 07/16/2019 - 10:59

GSI 4D HYBRID FOR ARW USING GLOBAL ENSEMBLE FORECAST

Setting up the Run Script for GSI 4D hybrid analysis

Copy the sample run script run_gsi_regional.ksh from the practical case 3 (ARW 3DVAR with PrepBUFR) to a working directory and make the following modifications to run GSI 4D hybrid analysis:

  • set the analysis time to 
    ANAL_TIME=2017051318
  • Set the name/path for the analysis run directory to
  • WORK_ROOT=${run directory}
  • Set the location of the ensemble files in the variable ENS_ROOT=...
  • set the BK_ROOT to where you store the background files, WRF forecast at 3 time levels
  • set the path to the background file at the analysis time 
    BK_FILE=${BK_ROOT}/wrfout_d01_2018-08-12_18:00:00
  • set to run 4D hybrid analysis 
    if_hybrid=Yes 
    if_4DEnVar=Yes

As can be seen in the sample run script, for ${if_4DEnVar} = Yes, there are two additional background files at 1 hour before and after the analysis time, as specified in the variable BK_FILE_M1 and BK_FILE_P1; there are also two additional sets of ensemble files before and after the analysis time, as specified in the variable ENSEMBLE_FILE_mem_m1 and ENSEMBLE_FILE_mem_p1. Please make sure those variables setup right.

An example of this run script is available from the link run_gsi_regional.ksh

Running the Script

Running the Script cindyhg Tue, 07/16/2019 - 11:00

GSI 4D HYBRID FOR ARW USING GLOBAL ENSEMBLE FORECAST

Running the Script

If you run on PBS system (Cheyenne), type: 

qsub run_gsi_regional.ksh 

to launch the job. 

The progress of the job can be monitored by examining the tail of the standard out file in the run directory as specified in the variable WORK_ROOT: 

tail stdout

When completed, the contents of this run directory are provided in the following list .

Results

Results cindyhg Tue, 07/16/2019 - 11:00

GSI 4D HYBRID FOR ARW USING GLOBAL ENSEMBLE FORECAST

Results

The standard output file stdout contains the run diagnostics, such as convergence information, and observation distribution from the GSI run. Details of the standard output file are available in section 4.1 of the GSI User's Guide.

Information about the use of observations by the analysis, and the corresponding innovations are available from the fit files (named fort.2*). The fit files located in the run directory should agree with the following fit files for temperature (fit_t1); wind(fit_w1); moisture (fit_q1); surface pressure (fit_p1); and radiance (fit_rad1); and GPS (fort.212); and radar radial velocity(fort.209).

Convergence information is available in the file: fort.220

Visualizing the Analysis

Use the same method as the practical case 3 (ARW 3DVAR) to make plots of the analysis increments. This time, plots will be made for the 2nd level (kmax=1) and level 21 (kmax=20). Once done pdf files GSI_Analysis_increment_1.pdf and GSI_Analysis_increment_20.pdf will be genera ted in the run directory. Compare these images with the reference solution [level 2 ] and [level 21].