Re: Ambiguous coordinates on time axis

[Ansley Manke <ansley.b.manke@xxxxxxxx>]

Hi David,
Here is a case where the command SET MODE DIAGNOSTIC really helps
you see what is happening.  The command in your Approach 1 is ambiguous
because Ferret attempts to find a context (region, grid and datasets) 
for the entire
command.  It uses the range from the grid of each variable, and since 
they have
different ranges in time can't be reconciled in a definitive way.  If 
you issue a
SET MODE DIAGNOSTIC before issuing the command, I think you'll see
that Ferret uses a 1-year region for the time averages of both data 
sets. If the
range of times in the two files did not overlap, you would have seen an 
error
message saying that the variable from the second set is not in the range of
the variable from the first set, and no result would be plotted.

When you specify time ranges explicitly,  Ferret knows to compute each 
average
over the specified range, and you will get the result you intended.

Ansley Manke

David Muhati wrote:

> Hi all,
> Came across this error message while calculating the difference 
> between myvar (variable) for 2 different periods for 1 year and 4years 
> respectively. Not sure what the best approach is of this two; NB: d=1 
> is the 1year period e.g 1997 and d=2 is the 4 year (1995 to 1998)
> Approach 1
> yes? fill myvar[d=1,l=@ave]- myvar[d=2,l=@ave]
> yes? go land
> error: ambiguous coordinates on time axis
> Approach 2
> yes? fill myvar[d=1,t="1-jan-1997":"31-dec-1997"@ave]- 
> myvar[d=2,t="1-jan-1995":"31-dec-1998"@ave]
> yes? go land
> Doesn't give an error message.
>
> So which is the best because the results of the two look different?
> David