YAML Configuration | GMI

[danholdaway]

Agreement with GSI using GeoVaLs in UFO (Hofx no bias correction)

• GPM

Agreement with GSI using GeoVaLs in UFO (Bias correction)

• GPM

Agreement with GSI using GeoVaLs in UFO (Quality control)

• GPM

Agreement with GSI using GeoVaLs in UFO (Hofx with bias correction)

• GPM

Agreement with GSI using Swell

• GPM

JEDI PRs

• https://github.com/JCSDA-internal/ufo/pull/2179
• https://github.com/JCSDA-internal/ufo/pull/2230
• IODA converters (sat bc predictors in the YAML) #111

Current issues:

• Cloud index needed in the bias correction. Currently not available in the UFO.
• YZ: suggestion to add cloud calculations to bias correction part.

[gmao-jjin3]

There is a pending issue (JCSDA-internal/ufo-data#164) that adding bias correction to hofx values before they are used for cloud retrievals. As a result, validation of GMI quality control on-hold at the moment.

[gmao-jjin3]

Clouds can be derived correctly in JEDI. X-axis, clouds from GSI; Y-axis, clouds from JEDI.

[gmao-jjin3]

GIS obervational errors (x-axis) are reproduced in JEDI as well (y-axis).

[gmao-jjin3]

Departures of GMI data in GEOS are fully reproduced in UFO after its bias correction is finalized in UFO. There is git-hub pull request for the latest update: https://github.com/JCSDA-internal/ufo/pull/2179. Here are departure difference ranges and observational error difference ranges:
*********** ch01 differences 'UFO_ObsValue-hofx_BC_QC' - 'ObsValue-GsiHofXBc_BC_QC': -4.57764e-05 4.57764e-05
*********** ch02 differences 'UFO_ObsValue-hofx_BC_QC' - 'ObsValue-GsiHofXBc_BC_QC': -1.52588e-05 2.28882e-05
*********** ch03 differences 'UFO_ObsValue-hofx_BC_QC' - 'ObsValue-GsiHofXBc_BC_QC': -1.52588e-05 3.05176e-05
*********** ch04 differences 'UFO_ObsValue-hofx_BC_QC' - 'ObsValue-GsiHofXBc_BC_QC': -1.52588e-05 3.05176e-05
*********** ch05 differences 'UFO_ObsValue-hofx_BC_QC' - 'ObsValue-GsiHofXBc_BC_QC': -3.05176e-05 3.05176e-05
*********** ch06 differences 'UFO_ObsValue-hofx_BC_QC' - 'ObsValue-GsiHofXBc_BC_QC': -3.05176e-05 6.10352e-05
*********** ch07 differences 'UFO_ObsValue-hofx_BC_QC' - 'ObsValue-GsiHofXBc_BC_QC': -1.52588e-05 1.52588e-05
*********** ch08 differences 'UFO_ObsValue-hofx_BC_QC' - 'ObsValue-GsiHofXBc_BC_QC': -3.05176e-05 3.05176e-05
*********** ch09 differences 'UFO_ObsValue-hofx_BC_QC' - 'ObsValue-GsiHofXBc_BC_QC': -3.05176e-05 3.05176e-05
*********** ch10 differences 'UFO_ObsValue-hofx_BC_QC' - 'ObsValue-GsiHofXBc_BC_QC': -3.05176e-05 3.05176e-05
*********** ch11 differences 'UFO_ObsValue-hofx_BC_QC' - 'ObsValue-GsiHofXBc_BC_QC': -3.05176e-05 3.05176e-05
*********** ch12 differences 'UFO_ObsValue-hofx_BC_QC' - 'ObsValue-GsiHofXBc_BC_QC': -3.05176e-05 3.05176e-05
*********** ch13 differences 'UFO_ObsValue-hofx_BC_QC' - 'ObsValue-GsiHofXBc_BC_QC': -3.05176e-05 3.05176e-05

*********** ch01 differences 'UFO_EffectiveError_QC' - 'GsiFinalObsError_QC': -1.90735e-06 1.90735e-06
*********** ch02 differences 'UFO_EffectiveError_QC' - 'GsiFinalObsError_QC': -3.81470e-06 3.81470e-06
*********** ch03 differences 'UFO_EffectiveError_QC' - 'GsiFinalObsError_QC': -3.81470e-06 3.81470e-06
*********** ch04 differences 'UFO_EffectiveError_QC' - 'GsiFinalObsError_QC': -3.81470e-06 3.81470e-06
*********** ch05 differences 'UFO_EffectiveError_QC' - 'GsiFinalObsError_QC': -3.43323e-05 4.19617e-05
*********** ch06 differences 'UFO_EffectiveError_QC' - 'GsiFinalObsError_QC': -2.86102e-05 2.95639e-05
*********** ch07 differences 'UFO_EffectiveError_QC' - 'GsiFinalObsError_QC': -0.000869751 0.000839233
*********** ch08 differences 'UFO_EffectiveError_QC' - 'GsiFinalObsError_QC': -3.81470e-06 3.81470e-06
*********** ch09 differences 'UFO_EffectiveError_QC' - 'GsiFinalObsError_QC': -3.81470e-06 3.81470e-06
*********** ch10 differences 'UFO_EffectiveError_QC' - 'GsiFinalObsError_QC': -7.24792e-05 9.53674e-05
*********** ch11 differences 'UFO_EffectiveError_QC' - 'GsiFinalObsError_QC': -3.81470e-06 3.81470e-06
*********** ch12 differences 'UFO_EffectiveError_QC' - 'GsiFinalObsError_QC': -4.57764e-05 5.72205e-05
*********** ch13 differences 'UFO_EffectiveError_QC' - 'GsiFinalObsError_QC': -5.53131e-05 7.24792e-05

[gmao-jjin3]

[gmao-jjin3]

[gmao-jjin3]

There are a few more data screened out by UFO filters than by GSI. That is because a minor 'bug' in GSI. The retrieval error flag is not checked when clouds are retrieved using simulated Tb and simulated Tb assuming clear-sky conditions.

           if(gmi) then
             call gmi_37pol_diff(tsim_bc(6),tsim_bc(7),tsim_clr_bc(6),tsim_clr_bc(7),clw_guess_retrieval,ierrret)
             call gmi_37pol_diff(tb_obs(6),tb_obs(7),tsim_clr_bc(6),tsim_clr_bc(7),clw_obs,ierrret)
           end if
           ......
           if (ierrret /= 0) then
                varinv(1:nchanl)=zero
                id_qc(1:nchanl) = ifail_cloud_qc
           endif

Only the 2nd ierrret is checked in the GSI program setuprad.f90.

[gmao-jjin3]

Waiting for the merge of the UFO PR: https://github.com/JCSDA-internal/ufo/pull/2230 .

[danholdaway]

Great stuff @gmao-jjin3! Thanks for working through all the challenges.

[gmao-jjin3]

Here is the PR to correctly convert GMI bias correction coefficients:
JCSDA-internal/ioda-converters#1042

[gmao-jjin3]

@gmao-yzhu @danholdaway The HofX of GMI in x0048 indeed can be reproduce by UFO with CRTM v2.4.0.

[danholdaway]

Thanks for checking that out @gmao-jjin3

[gmao-jjin3]

New validation using results of the x0048 experiment. Get the same conclusion as for x0044 in August 2022. GSI results of O-F and QC can be reproduced by UFO, but there are a few more observations tossed by UFO for channels 10-13. The reason was posted in #66 (comment).
A GSI PR GEOS-ESM/GEOSana_GridComp#137 is made for future GEOS update.

[gmao-jjin3]

Here are comparisons of observational errors in JEDI and GSI.