Adding a preliminary capability of radiance DA in EnKF#496
Conversation
LiaofanLin-NOAA
requested review from
CurtisAlexander-NOAA,
JeffBeck-NOAA,
TrevorAlcott-NOAA,
christinaholtNOAA,
chunhuazhou,
daviddowellNOAA,
guoqing-noaa,
hu5970 and
terraladwig
as code owners
| use_gfs_nemsio=.false.,use_gfs_ncio=.true.,reset_bad_radbc=.true., | ||
| netcdf_diag=${netcdf_diag},binary_diag=${binary_diag}, | ||
| l_obsprvdiag=${l_obsprvdiag}, | ||
| lwrite_peakwt=.true., |
There was a problem hiding this comment.
If this is set to true by default, will it affect other applications? If yes, we need to treat this parameter as configurable (for example, like l_obsprvdiag)
There was a problem hiding this comment.
According to the GSI v3.7 user guide, lwrite_peakwt is an option to write out the approximate pressure of the peak of the weighting function for satellite data to the radiance diagnostic files.
In short, since the diag files are only used by enkf.x in the workflow, adding "lwrite_peakwt=.true.," in gsiparm.anl.sh should not affect other applications in the workflow.
I have further investigated the impact with "lwrite_peakwt=.true." and with "lwrite_peakwt=.false." on the diag files generated via the GSI-observer. Specifically, I checked the following three files from a single cycle of an RRFS-EnKF run:
- diag_conv_t_ges.2022020702.nc4
- diag_iasi_metop-b_ges.2022020702.nc4
- diag_amsua_metop-b_ges.2022020702.nc4
The results show:
- For diag_conv_t_ges.2022020702.nc4, there is no impact. It makes sense, as lwrite_peakwt is for writing out satellite information for radiance DA in EnKF.
- For diag_iasi_metop-b_ges.2022020702.nc4, the file size increases from 4.0 MB (lwrite_peakwt=.false.) to 4.3 MB (lwrite_peakwt=.true.). The larger file has a new variable called Press_Max_Weight_Function, which does not exist in the smaller file. In the global attributes, iextra is changed from 0 to 1 while jextra is changed from 0 to 616. The smaller file (lwrite_peakwt=.false.) has iextra=0 and jextra=0. Those are all the differences.
- For diag_amsua_metop-b_ges.2022020702.nc4, the file size increases from 290 KB (lwrite_peakwt=.false.) to 300 KB (lwrite_peakwt=.true.). Like the IASI one, using lwrite_peakwt=.true. adds variable Press_Max_Weight_Function in the diag files and changes global attributes iextra and jextra from 0 to numbers.
Note that in conventional diag files, there are no such attributes iextra or jextra.
There was a problem hiding this comment.
Thanks, @LiaofanLin-NOAA. This is great documentation of lwrite_peakwt.
| if [ ${GSI_TYPE} == "OBSERVER" ]; then | ||
| if [ ${MEM_TYPE} == "MEAN" ]; then | ||
| if [ ! -d "${SATBIAS_DIR}_ensmean" ]; then | ||
| echo "makding satbias_dir ensmean directory for cycling bias correction files" |
There was a problem hiding this comment.
Thank you! I updated it.
LiaofanLin-NOAA
force-pushed
the
ens_radda
branch
from
949625b to
1bd62d6
Compare
2 participants
DESCRIPTION OF CHANGES:
Adding the capability of radiance data assimilation in RDAS-EnKF. Specifically, a configuration variable “DO_ENS_RADDA” is created. Note that this variable works (i.e., DO_ENS_RADDA=TRUE) only when “DO_RADDA=TRUE”, as the RDAS-EnKF requires running the GSI-observer prior to the ensemble data analysis. The tasks of GSI (EnVar) and GSI-observer (for EnKF) will keep the corresponding satbias files under different folders in NWGES.
Changing the starting satbias files in exregional_run_analysis.sh, by checking RRFS files rather than GDAS files. The relevant PRs are:
Adding “lwrite_peakwt=.true.” in gsiparm.anl.sh. The relevant issue report is at:
TESTS CONDUCTED:
The changes were tested for RRFS_CONUS_13km_ens on Hera, via EnKF cycling runs.
DEPENDENCIES:
None
DOCUMENTATION:
None