Re: [ferret_users] ave vs. sum and ngd

[Peter Szabo <szabpet83@xxxxxxxxx>]

Sorry, i have my data on a default calendar time axis.
So the ave must work:

i use weights 31,30,31, but this equals to the @sum/3 value.

yes? list (1.74*31+8.74*30+15.34*31)/92
             VARIABLE : (1.74*31+8.74*30+15.34*31)/92
          8.605
Peter

On Mon, Jun 22, 2009 at 4:16 PM, Peter Szabo <szabpet83@xxxxxxxxx> wrote:

Martin,
as i am dealing with climatology, all the months are equally important, so i should use the @sum/3 for a spring average.
Thanks, Peter

---------- Forwarded message ----------
From: Martin Schmidt <martin.schmidt@xxxxxxxxxxxxxxxxx>
Date: Mon, Jun 22, 2009 at 12:28 PM
Subject: Re: [ferret_users] ave vs. sum and ngd
To: Peter Szabo <szabpet83@xxxxxxxxx>


Peter, the answer is most probably the following example:

yes? list (1.74*31/2+8.74*30+15.34*31/2)/61
           VARIABLE : (1.74*31/2+8.74*30+15.34*31/2)/61
        8.638

The values at the edges are counted with half weights. However, I am not sure about the remaining difference.

Greetings,
Martin

Peter Szabo wrote:

Dear Users,
i know there is a difference between @ave and @sum, @ngd. It is not really a
question, but still want to post it.

I list 3 timesteps:
yes? list/nohead temp[l=16:18]
 16-MAR-1962 / 16:   1.74
 16-APR-1962 / 17:   8.74
 16-MAY-1962 / 18:  15.34

Here is the real average for it. The mean i guess.
yes? list/nohead temp[l=16:18@sum]/3
         8.605

The (weighted) "average" is slightly different. Though if there is no
missing value, they should be the same.
yes? list/nohead temp[l=16:18@ave]
         8.643

Why is that weightening in @ave? The 3 timesteps are equivalent for me. I
should just use the sum and dividing then?
If i use @var later on (which is also a weighted variance), should i use
@ave then?

Thanks, Peter