# [gradsusr] difference between field and single point values for the same defined expression

Jeff Duda jeffduda319 at gmail.com
Sat Apr 30 15:52:53 EDT 2016

```My only guess is that since the differences are so small, Grads just
truncates them to zero.

Jeff Duda

On Fri, Apr 29, 2016 at 2:36 PM, Helio Camargo <heliocamargo at gmail.com>
wrote:

>
> I´m performing the following expressions over time 1 of a 50 ensemble
> member run:
>
> 'set t 1 last'
> 'set e 1 last'
> 'define ens = tpratesfc.1 '
>
> 1) 'define smens = sum(ens,e=1,e=50)'
> 2) 'define 'smensmsk =  maskout(sum(ens,e=1,e=50),sum(ens,e=1,e=50)-0)'
> 3) 'define 'smensmskcnst
>
> 'set t 1'
>
> For some reason, for t=1, expression 1, yelds small negative values
> (smens.gif, attached). Expressions 2 and 3 should set undef to these
> negative values and set 0 for these undef grid points respectively. Looking
> at the figures smensmsk.gif and smensmskcnst.gif, these operations seem to
> work perfectly
>
> And setting lon/lat to a point where the above operations should
> influence, is consistent with the figures above:
> Lon = 298.442  Lat = -28.8304
>
> *ga-> d smens*
> *Result value = -8.33652e-19*
>
> *ga-> d smensmsk*
> *Result value = -9.99e+08*
>
> *ga-> d smensmskcnst*
> *Result value = 0*
>
> ** differences also show expected result*
> *> operation 2 - operation 1*
> ga-> d smensmsk-smens
> Result value = -9.99e+08
>
> *> operation 3 - operation 2*
> ga-> d smensmskcnst-smensmsk
> Result value = -9.99e+08
>
> *> operation 3 - operation 1*
> ga-> d smensmskcnst-ens
> Result value = 1.9073e-20
>
> However, making FIELD  diferences between defined expression 3 and defined
> expression 1 results "Constant Field=0" (diff_smensmskcnstsmens.gif).
>
> Can anyone explain why this is happening?
>
> --
> Helio
>
> _______________________________________________
>
>

--
Jeff Duda