[CF-metadata] stations and trajectories

From: Jonathan Gregory <j.m.gregory>
Date: Wed, 8 Jun 2005 17:37:53 +0100

Dear John

Stations
> If you want to store multiple trajectories in the file, CF doesnt
> explicitly say how to do that. The obvious solution might be to add a
> trajectory dimension. This has the limitation that you must store the
> same number of points for each trajectory.

Trajectories
> The current standard has the limitation that all stations have the same
> number of observations (time-series) and time coordinates. Mesonet
> station data, for example, is often variable length with arbitrary time
> stamps.

In both cases, if the timeseries or trajectories have different lengths, is
there any reason why you should not put them in separate variables with
different time dimensions and coordinates? NetCDF-4 will help this by allowing
more than one unlimited dimension.

> There is no explicitly specified way to describe station information, eg
> a station id, description, etc.

There is a method to associate such labels with stations in 6.1 of the
standard. Standardised station IDs such as WMO station numbers could be used
if that's appropriate (although use of codes makes the file less self-
describing, so I think names are better).

Best wishes

Jonathan
Received on Wed Jun 08 2005 - 10:37:53 BST