Dear Jonathan and Brian,
Apologies for the delay in carrying on this conversation, busy times.
I agree that we should consider this issue as analogous to that of the sigma
vertical levels. What we have is two new auxiliary coordinate variables,
forecast_reference_time and forecast_validity_time. They can be identified
by their standard names and should not cause a problem for software.
However, if they have standard_names other than "time" then some current
software will miss assigning the time dimension unless there is an axis="T"
attribute. Having said that, if there are two time dimensions then most
software will fall over anyway trying to understand forecast files.
For ultimate flexibility, the forecast_validity_time should have a units
attribute entirely independent of the forecast_reference_time units
attribute and is therefore a fully fledged time axis in its own right.
Do we want to encourage people who have one analysis and 3 forecast steps
(00,06,12 and 18hr time steps) to use the old system of just one linear time
dimension (which would lose valuable metadata regarding the relationship
between the time steps) or should they follow the 2 auxiliary coordinate
variable approach? I would suggest that in the long run they want to do the
latter but right now they would have more luck doing the former.
Another related issue brought up by a colleague the other day is encoding
ensemble simulations. It's a similar issue to forecasts in that you have
duplicate variables at the same space/time coordinates. Do you think it is
effectively the same issue as forecasts or do we need an extraneous
description for ensemble runs?
In summary, I like the standard names of forecast_reference_time (i.e.
analysis time) and forecast_validity_time. Just to raise it as a
possibility, these could also be "time forecast_reference" and "time
forecast_validity" in the same way as we discussed encoding errors and
uncertainties. It's probably not the right solution.
Sorry for more questions than answers (as usual!).
Kind regards,
Ag
> -----Original Message-----
> From: Jonathan Gregory [mailto:j.m.gregory at reading.ac.uk]
> Sent: 16 June 2003 10:33
> To: cf-metadata at cgd.ucar.edu
> Subject: Fwd: Re: [CF-metadata] Example of forecast data
>
>
> Dear Brian and Ag
>
> > On Fri, Jun 13, 2003 at 12:28:00PM +0100, Jonathan Gregory wrote:
> > > I would like
> > > to propose a new standard name of forecast_validity_time
> to use instead of
> > > plain time in this scheme.
> >
> > My initial reaction to this is not favorable. This is
> using the time axis
> > to express a property that belongs to the data variable.
> In other words,
> > the fact that the variable may contain forecast data is a
> property of the
> > data, not the time axis. A standard name already exists
> that allows us to
> > distinquish the valid time of the data from a reference
> time. It seems to
> > me that that's sufficient.
>
> If we have two auxiliary coordinate variables, one for
> forecast reference time
> and one for validity time, they have a similar status, and it
> seems natural to
> me to give them standard names of forecast_reference_time and
> forecast_validity_time. Both of these are more specific
> "varieties" of time,
> and being specific about both of them is more informative to
> the user. Why
> should one of them have the standard name of time when the
> other one doesn't?
>
> > > Can't we use simpler schemes?
> >
> > I assume that what we're discussing is what to recommend
> (via examples) in
> > the document.
>
> That's right. The scheme with an index dimension and
> auxiliary coordinate
> variables is a sensible use of an existing CF convention, and
> seems the best
> way to deal with forecasts having various combinations of
> forecast period
> and analysis time. My question is whether we should recommend
> it in simpler
> cases, where there is only one analysis time, or one validity
> time, involved.
>
> > > For Ag's case of a set of analyses I think I would use a
> single time axis with
> > > standard name forecast_reference_time instead of time, as
> it's more informative.
> >
> > This was my initial thought as well. But on further
> reflection, whether
> > data represents a forecast or an analysis is really a
> property of the data
> > itself and not of the time axis. So I would say that this
> data should use
> > a plain time axis which represents it's valid time, and
> that the fact that
> > it's analysis data should be more precisely indicated by
> the "source"
> > attribute which could include details of how the analysis
> was produced.
> > That said, using forecast_reference_time to identify the
> time axis is
> > certainly legal and would most likely be correctly
> interpreted by someone
> > using the data.
>
> Yes, I wondered about that argument too. In the context of
> other data which
> is concerned with forecasts, using the standard name of the
> time axis might
> seem more natural. In connection with forecasts, I think you
> can regard analysis
> and validity time as two kinds of time dimension. In that
> case, using the
> standard name to distinguish them is analogous to using it to
> distinguish
> vertical levels of pressure and sigma, say. Software which
> identifies a time
> axis by units or by the axis attribute would not have a
> problem with it. I
> think it's too restrictive to require a time axis to have a
> standard name of
> time. Since standard names are standardised, and the source
> attribute is not,
> using the standard name may also be more helpful when
> comparing data from
> different institutions.
>
> Best wishes
>
> Jonathan
> _______________________________________________
> CF-metadata mailing list
> CF-metadata at cgd.ucar.edu
> http://www.cgd.ucar.edu/mailman/listinfo/cf-metadata
>
Received on Fri Jun 27 2003 - 08:42:52 BST
