search

NCAR / DART

Data Assimilation Research Testbed
https://dart.ucar.edu/
Apache License 2.0
189 stars 142 forks source link

Snow fraction obs converter documentation #531

Open braczka opened 1 year ago

braczka commented 1 year ago

The snow fraction obs converter documentation is incomplete. This applies for both snow_to_obs.f90 (text file) and snow_to_obs_netcdf.f90 (netcdf file). Although it appears both converters are meant to operate on the same data, the original source data formatting (HDF-EOS) presents problems, requiring a 2 step conversion to text or netcdf formatting prior to using the main converters.

At the very least the documentation needs to identify the source data and scripting used for the 1st conversion step.

It seems the direct conversion of HDF-EOS format to obs.seq is a common challenge for DART converters very similar to what is described for the AMSU/AIR converters., in which an incompatibility exists between HDF-EOS and HDF5, necessitating a 2-step conversion. In that case the HDF-EOS tools scripting is used to convert to netcdf formatting.

These HDF-EOS converters could very well be used for the snow fraction obs converter as well, but this needs to be demonstrated and documented. Also, it is unclear if the HDF5 and HDF-EOS incompatibility issue for the AMSU converter applies to the snow converters. The AMSU converter obs_seq files use the RTTOV forward operators (which require HDF5) whereas the snow converters do not.

Given we currently have collaborators working with SNODAS SWE (snow water equivalent) obs converters its a good time to get our snow fraction converters sorted out.

hkershaw-brown commented 1 year ago

hot take: would a version of rttov built without HDF5 be helpful?

braczka commented 1 year ago

I like the hot take idea, but would it work for the radiance converters if RTTOV requires HDF5? To be honest I am not sure why there is a HDF-EOS and HDF5 conflict given the RTTOV forward operator is related, but should be independent from the radiance observation converters and the resulting obs_seq file. I must be missing something.....

With that being said, my real concern for this issue is the snow converters, which are completely unrelated to RTTOV, thus a conflict shouldn't exist, but the documentation still suggests a 2 step conversion is necessary for the snow converters. I am confused.