Caro-COOPS.org Discussion Board

jcothran · Joined: 24 Feb 2025 Posts: 176 Location: Columbia, SC

Hi Charlton,

The reasons which come to mind for using CF-1.0/Seacoos compliant netCDF instead of ASCII are as follows:

- standard time and spatial dimension references and conversion libraries
- existing applications such as ncBrowse which allow visualization of the netCDF file online directly
- existing applications such as the DODS netCDF servers which allow query and subsetting of the netCDF file online directly
- data validation in conversion to format itself
- compressed storage (binary over text) - not really currently an issue, but potentially one as archiving is addressed
- whatever else is listed at Unidata(the keeper's of netCDF) http://my.unidata.ucar.edu/content/software/netcdf/guidec/guidec-6.html#HEADING6-0

For a particle format where time and the number of particles can vary I would suggest something along the following format

<row_id - relational database only>
time
particle_id
lat
long
<depth - optional>
<various metadata fields related to this particle and it's display>

If the number of particles was the same throughout the fileset, then you could treat each particle as it's own dimension and maintain a regular array - but since the number of particles is changing, all particles are folded into the 'particle_id' field which is regular while ranging over all the possible particles.

We can always supply simple conversion programs/tools as well to facilitate model ASCII output to a chosen netCDF format.

Jeremy