Re: Vertical averaging using varying depths

Hi Dennis + other users,
I have solved the memory problem by selecting my area of interest (instead of using the global data) after realizing that I can use the function: center_finite_diff to calculate gradients on non-global data. This way, Dennis' previous suggestion works. Thanks a million!
Hyacinth
 

> Date: Thu, 24 Jan 2013 07:18:26 -0700
> From: shea@ucar.edu
> To: hyacinth.1@hotmail.com
> CC: ncl-talk@ucar.edu
> Subject: Re: Vertical averaging using varying depths
>
> variable > 2GB are not an issue for NCL 6.0.0 and higher
>
> If you are encountering a memory error, it is likely that
> you are exceeding your operating system memory
>
> On 1/24/13 4:22 AM, Hyacinth Nnamchi wrote:
> >
> > Hi Dennis,
> > A more fundamental constraint is the size of my data: 624x29x193x256 and 624x193x256, which exceed 2G by far even before I create additional arrays. I'll therefore have to subset at some fixed levels and average directly. Thanks for your repeated hints, they may be useful for me elsewhere.
> > Hyacinth
> >
> >> Date: Wed, 23 Jan 2013 10:10:47 -0700
> >> From: shea@ucar.edu
> >> To: hyacinth.1@hotmail.com
> >> CC: ncl-talk@ucar.edu
> >> Subject: Re: Vertical averaging using varying depths
> >>
> >> The denominators should have a 2nd argumenty
> >>
> >> dim_sum_n(dz) ==> dim_sum_n(dz , 0)
> >> dim_sum_n(dz4) ==> dim_sum_n(dz4, 1) ;
> >>
> >> On 01/23/2013 09:09 AM, Hyacinth Nnamchi wrote:
> >>> Thanks for your response Dennis, I'll try out your suggestion.
> >>>
> >>> Hyacinth
> >>>
> >>>
> >>>
> >>> > Date: Mon, 21 Jan 2013 11:26:15 -0700
> >>> > From: shea@ucar.edu
> >>> > To: hyacinth.1@hotmail.com
> >>> > CC: ncl-talk@ucar.edu
> >>> > Subject: Re: Vertical averaging using varying depths
> >>> >
> >>> >
> >>> > There can not be function for everything.
> >>> > You can always write your own.
> >>> > Also, is the (x,y,z,t) fortran ordering???
> >>> > Is 'z' one dimensional or 4 dimensional?
> >>> >
> >>> > let: dz=layer_thickness,
> >>> > q=variable, fortran(x,y,z,t)=>ncl(t,z,y,x) ; 0,1,2,3
> >>> >
> >>> > z_zavg = dim_sum_n(q*dz, 1)/dim_sum_n(dz) ; q and dz (x,y,z,t)
> >>> > --
> >>> > dz4 = conform(dz,z,1)
> >>> > z_zavg = dim_sum_n(q*dz4, 1)/dim_sum_n(dz4) ; q(x,y,z,t), dz(z)
> >>> >
> >>> > On 1/21/13 10:00 AM, Hyacinth Nnamchi wrote:
> >>> > >
> >>> & gt; > Hi users,
> >>> > > I want to calculate the vertical average of a 4D (x,y,z,t) ocean
> >>> variable. The problem is that the depths will have to vary: I want to
> >>> use the thermocline depth (already calculated, x,y,t) as the base at
> >>> each grid point, for the vertical vertical averaging. Is there a
> >>> way/function to do this ncl?
> >>> > > Thanks in advance.
> >>> > > Hyacinth
> >>> > >
> >>> > >
> >>> > >
> >>> > >
> >>> > > _______________________________________________
> >>> > > ncl-talk mailing list
> >>> > > List instructions, subscriber options, unsubscribe:
> >>> > > http://mailman.ucar.edu/mailman/listinfo/ncl-talk
> >>> > >
> >>>
> >>>
> >>> _______________________________________________
> >>> ncl-talk mailing list
> >>> List instructions, subscriber options, unsubscribe:
> >>> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
> >>>
> >>
> >
> >
> >
> >
> > _______________________________________________
> > ncl-talk mailing list
> > List instructions, subscriber options, unsubscribe:
> > http://mailman.ucar.edu/mailman/listinfo/ncl-talk
> >
                                               

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Thu Jan 24 07:27:42 2013