search

GEOS-ESM / GEOSana_GridComp

Repository containing code for data analysis for the GEOS Earth System Model data assimilation
Apache License 2.0
1 stars 3 forks source link

GSI reproducibility with a new observational data set #151

Closed gmao-jjin3 closed 1 month ago

gmao-jjin3 commented 1 year ago

GSI results may be not reproducible when a new monitoring observational data set is added. I found that GMI departures somehow are changed when a new monitoring ATMS data are added in the middle of the name list "OBS_INPUT" in the file gsi.rc.tmpl. However the issue disappears after the new ATMS data set is added at the end of the name list. The reason is unknown.

Here are part of the name lists with a new monitoring "ncep_atms_n21_bufr" which is added right after the other ATMS data set ncep_atms_bufr and before other observations in gsi.rc.tmple which causes different GMI departures.

OBS_INPUT::
!  dfile          dtype       dplat       dsis                  dval    dthin  dsfcalc  obsclass
   prepbufr       ps          null        ps                    0.0     0      0        ncep_prep_bufr 
   ......
   atmsbufr       atms        npp         atms_npp              0.0     1      0        ncep_atms_bufr
   atmsbufr       atms        n20         atms_n20              0.0     1      0        ncep_atms_bufr
   atmsn21bufr    atms        n21         atms_n21              0.0     1      0        ncep_atms_n21_bufr
   crisfsrbufr    cris-fsr    npp         cris-fsr_npp          0.0     3      0        ncep_crisfsr_bufr
   crisfsrbufr    cris-fsr    n20         cris-fsr_n20          0.0     3      0        ncep_crisfsr_bufr
   ......
   gmibufr        gmi         gpm         gmi_gpm               0.0     1      0        gmao_gmi_bufr
   amsr2bufr      amsr2       gcom-w1     amsr2_gcom-w1         0.0     1      0        gmao_amsr2_bufr

This results differences in GMI O-F values: 

      it      satellite instrument     # read         # keep      # assim   penalty      qcpnlty       cpen      qccpen 
< o-g 01 rad  gpm       gmi            26230409        54561        18704    6955.6       6955.6      0.37188      0.37188    
> o-g 01 rad  gpm       gmi            26230409        54561        18703    6983.0       6983.0      0.37336      0.37336

The issue is resolved with zero difference by adding "ncep_atms_n21_bufr" at the very end of the name list:

OBS_INPUT::
!  dfile          dtype       dplat       dsis                  dval    dthin  dsfcalc  obsclass
   prepbufr       ps          null        ps                    0.0     0      0        ncep_prep_bufr 
   ......
   atmsbufr       atms        npp         atms_npp              0.0     1      0        ncep_atms_bufr
   atmsbufr       atms        n20         atms_n20              0.0     1      0        ncep_atms_bufr
   crisfsrbufr    cris-fsr    npp         cris-fsr_npp          0.0     3      0        ncep_crisfsr_bufr
   crisfsrbufr    cris-fsr    n20         cris-fsr_n20          0.0     3      0        ncep_crisfsr_bufr
   ......
   gmibufr        gmi         gpm         gmi_gpm               0.0     1      0        gmao_gmi_bufr
   amsr2bufr      amsr2       gcom-w1     amsr2_gcom-w1         0.0     1      0        gmao_amsr2_bufr
   atmsn21bufr    atms        n21         atms_n21              0.0     1      0        ncep_atms_n21_bufr
gmao-jjin3 commented 1 month ago

Edit the above comment: "The issue is resolved with zero difference by adding "ncep_atms_n21_bufr" at the very end of the name list".