Dear Roy

The number_of_observations and status_flag are problematic because they may
apply to several variables with different unmodified standard names and it's
not desirable to duplicate the information. That leaves detection_minimum and
standard_error, which are more like anomaly in principle. I am inclined to
think that standard_error might be better described as a cell_method, since
it's a statistical operation that refers to how the values of the cell might
vary (over repeated measurements).

I agree that it would be natural to describe anomaly as a modifier if we
need to make a special provision for anomalies, as David suggested. In fact
this has been discussed before and it would fit the convention, I think. My
view is that if there are only half-a-dozen use-cases, it's not a problem to
have standard names for them including "anomaly". I think that's better
because to me it seems more robust to have the information glued together in
the standard name, and it avoids adding to the convention.

Best wishes

Jonathan

----- Forwarded message from rmendelss gmail <[email protected]> -----

> Date: Fri, 1 Apr 2016 10:38:49 -0700
> From: rmendelss gmail <[email protected]>
> To: Jonathan Gregory <[email protected]>
> CC: [email protected]
> Subject: Re: [CF-metadata] Standard attributes for anomaly
> X-Mailer: Apple Mail (2.3124)
> 
> Hi Jonathan:
> 
> 
> > On Mar 31, 2016, at 5:28 AM, Jonathan Gregory <[email protected]> 
> > wrote:
> > 
> > (a) in practice the standard name modifiers have caused
> > complications, discussed in trac tickets, which makes it a less attractive
> > mechanism, and
> 
> Before I respond further, can you give more detail on this, what are the 
> problems and why these problems would be relevant to anomalies.
> 
> Thanks,
> 
> -Roy
> 

----- End forwarded message -----
_______________________________________________
CF-metadata mailing list
[email protected]
http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata

Reply via email to