I would like to save the GeoVals values in a file needed for a UFO test including the model variables at observation locations and the main metadata such as latitude, longitude and datetime.
I have successfully run the obs filter GOMsaver, however, it only saves the requested model variables at observation locations under variables group. I have yet not found a way to save the necessary metadata. If this is possible could you please advise, which routine I need to call to request MetaData to also be saved in these files. I feel like I have overlooked something very simple.
The MetaData is stored in either of the āobs spaceā input or output files. It is stored in the same order of locations as the GeoVaLs feedback files, as long as the per-processor files are read in order. Currently there is not a facility to write the MetaData to the GeoVaLs feedback files.
Yes I can see the MetaData from the obs space but I was hoping to produce the geovals.nc4 files from the code not manually (which I understand from what you are saying is the case for all of the geovals files in the UFO testdata folder).
Thank you, it is useful to know that the observations are kept in the same order in hofx and geoval outputfiles for each file per processor. I could thus modify e.g. combine_obsspace.py file in the āgsi_ncdiagā folder to process hofx and geoval files in order as a pair per processor to read in geovals to produce a single geovals file with metadata. I think that should work, but it would be great if MetaData was automatically added within the geovals output files when running GOMsaver filter
On a similar note, is there a way to include geovals within the hofx output files?
I find an easy way to do the matching of the MetaData" and the geoval outputs. Just add the following lines to output the obs metadata information into multiple files which is consistent for the both:
obsdataout:
engine:
type: H5File
obsfile: sfc_obs_2022052619corrected.nc4
io pool:
max pool size: 36 (for example only, to be consistent with the ntask number)
write multiple files: true
Then use python or ncl to read the files in pairs. This should do the tricks.