Feature/directory tree

[frolovsa]

suggested initial directory structure closes issue #1

[jswhit]

how will we handle non-bufr data, like IMS snow cover GRIB and tcvitals text files?

[frolovsa]

@HenryWinterbottom-NOAA take a look at the proposed directory tree structure for the clean bucket. your input would be appreciated.

[frolovsa]

how will we handle non-bufr data, like IMS snow cover GRIB and tcvitals text files?

I would imagine we would store it in something like reanalysis/snow/ims/yyyy/mm/grib

The addition of the last directory is new. NASA doesn't have it. I figure we might needed it in case we need to store data in several formats. I also imagine we would want to store auxiliary data like png images with data coverage in each file, etc.

[HenryRWinterbottom]

how will we handle non-bufr data, like IMS snow cover GRIB and tcvitals text files?

@jswhit I would suggest we have a sub-directory structure as follows for the respective file types above.

GRIB: /grib

TCVitals: /syndat -- this is the NCEP convention and will allow us to differentiate between synoptic TC-vitals and/or the hourly CIMSS derived TCVs (if we wish to maintain this capability)

Non-bufr: /obs -- I think this is generic enough to be used for all non-standard file types; we could potentially do something like /obs_nc if we wanted to further partition into file types (i.e., netcdf).

[HenryRWinterbottom]

how will we handle non-bufr data, like IMS snow cover GRIB and tcvitals text files?

I would imagine we would store it in something like reanalysis/snow/ims/yyyy/mm/grib

The addition of the last directory is new. NASA doesn't have it. I figure we might needed it in case we need to store data in several formats. I also imagine we would want to store auxiliary data like png images with data coverage in each file, etc.

@frolovsa I like this idea. Segregating the file format types is the best path forward IMHO. Otherwise the buckets could potentially become what we have now on BDP (a mess).

[HenryRWinterbottom]

Approved.