Open ADCollard opened 1 year ago
GMAO have shared one of their YAMLs for aircraft. This does not include the bias correction, so we can work on adding that.
Looking at the aircraft YAML from GMAO.
There are some additional filters required that are not available in UFO develop.
ObsErrorFactorDuplicateCheck
ObsErrorFactorPressureCheck
There is also a weird error
Test "ufo/ObsFilters/testFilters" failed with unhandled eckit::Exception: Bad Conversion:
Cannot convert (1.4,1.4,1.4,1.4,1.4,1.4) (List) to double @
(/scratch1/NCEPDEV/global/spack-stack/spack-stack-v1/cache/build_stage/emc.nemspara/spack-stage-eckit-1.20.2-cxvre73ysjwofwoqpbwssnsqeucjlg2h/spack-src/src/eckit/value/Content.cc +75 badConversion)
Without the new filters testing will be limited, but we can look at aircraft bias correction and humidity QC (which is not included by GMAO).
Finally got the system working. Air temperature shows significant differences as would be expected from not having bias correction in place:
Final ob errors are fairly close but differ too:
Humidity QC is not included in the GMAO YAMLs.
H(x) differences are at the 1.e-8 level:
Something is wrong with the obs error differences:
Wind H(x) differences are very small (as expected):
The obs error differences are very large, however. We certainly need to dig.
Re-starting this work with updated GMAO YAML files
Test YAMLs are in https://github.com/NOAA-EMC/GDASApp/tree/feature/gmao_eval/parm/atm/obs/testing/swell
Differences for sondes
@ADCollard thanks for the plots. After comment out a few filter lines in sondes yaml, I was able to run the test and my results are very similar to your results except one for the stationPressure obs error plot:
@PraveenKumar-NOAA I am glad you are looking at it. Take a look at my YAML here. It runs without needing to delete any filters.
@ADCollard thanks. I will look into this.
@ADCollard I noticed an additional line 'pressure: MetaData/pressure' at few places in your yaml, it worked now without any comment and I produced exactly your results.
@ADCollard after making the following minor changes in your yaml:
I got the following results for the stationPressure:
If we use the GSI _da_psfcscheme instead of UKMO, the results are very similar:
Splitting this YAML into one for sondes and one for aircraft.
Evaluate the GMAO conventional data YAMLs in the context of the GDASApp.