[ferret_users] Re: decadal time axis

[Marco van Hulten <Marco.Hulten@xxxxxx>]

James, others—

Thanks, this solved my issue!

On 26 Sep 11:39 James Orr wrote:
> In response to your recent question on the ferret list, try something like the 
> following:
> 
> define axis/t="1-jul-1000":"1-jul-2000":1/units="years"/npoints=100/cal=noleap/edges tdecadal

I tried this, and indeed


yes? sh grid gdecadal
    GRID GDECADAL
 name       axis              # pts   start                end
 normal    X
 normal    Y
 normal    Z
 TDECADAL  TIME               100 r   01-JUL-1005 00:00    01-JUL-1995 00:00
yes? sh axis tdecadal
 name       axis              # pts   start                end
 TDECADAL  TIME               100 r   01-JUL-1005 00:00    01-JUL-1995 00:00
T0 = 01-JAN-1000
CALENDAR = NOLEAP
   Axis span (to cell edges) = 1000
yes? plot sssO12[d=1,i=1,j=100,k=1,gt=gdecadal]
 **ERROR: regridding: only @ASN regridding between calendar and
          non-calendar axes: SSSO12
yes? plot sssO12[d=1,i=1,j=100,k=1,gt=gdecadal@asn]
yes? 


The last command works and the figure shows something decent!

I guess that the @ASN is applied first to get rid of the (wrong) time
axis, and then the gdecadal grid is used (which has normal x, y and z
axes and tdecadal as the time axis).

For me, the essential points from James' solution are the 'npoints'
specification and taking the outer boundaries of the timespan.

—Marco