[ferret_users] problem with set grid q and redefining time-axis

[Peter Szabo <szabpet83@xxxxxxxxx>]

Hi Users,

i experienced something strange during trend analysis: when i do many linear regression along t axis one after an other (i use "set grid q") and define again and again the time grid, the timesteps of my FD var go wrong (actually looses its last timestep).

...
let p=t[gt=fd,l=1:40]
let q=fd[l=1:40]
set grid q

go regresst
!-------
let start_year = 1960
let nyears = 41
let indices = l[l=1:`nyears`]
let year = start_year + indices
def ax/un=days/t0=1-jan-1900/ed trueyear = DAYS1900(year,1,1)
...

my var, FD is like this:
yes? sh grid fd
    GRID (G003)
 name       axis              # pts   start                end
 LON       X                    1 r   19.2                 19.2

 LAT       Y                    1 r   47.4                 47.4
 normal    Z
 TRUEYEAR  TIME                40 i   02-JUL-1961 12:00    02-JUL-2000 00:00

after this i define trueyear,

I know i solved this without defining again (leaving) the "trueyear" time axis, but why these cant assort with each other?
Or is it better leaving the "set grid q" before "go regresst"? Is that really necessary?

Thanks,
Peter