[CF-metadata] featureType attribute (was CF-1.6 DSG clarification: time series & lat/lon coordinates)

From: Jonathan Gregory <j.m.gregory>
Date: Tue, 3 Dec 2013 21:06:03 +0000

Dear Nan

> Does the presence of a featureType attribute indicate that a file
> uses the DSG
> "machinery" and should therefore follow the guidelines of limited axes that
> are spelled out in chapter 9?

featureType can have only those values which are shown in Table 9.1. Each
value means that the data has a particular geometry, as shown in the table.
Your data
> float seatemp(time, depth, lat, lon)
does not have one of those geometries. In words, it isn't apparently a set
of timeseries, or a set of profiles, or any other of the possibilities. It's
a variable with four independent spatiotemporal axes. This is the "usual"
type of gridded domain which CF has always supported. In the discrete
sampling geometries the spatiotemporal axes aren't all independent. Therefore I
don't think you can use the featureType attribute with your data as it stands.

If the data was
> float seatemp(station, time)
it would look like a collection of timeseries, so you could give it
featureType="timeseries".

Best wishes

Jonathan
Received on Tue Dec 03 2013 - 14:06:03 GMT