Re: regrid issue

From: Xin Xi <xxi_at_nyahnyahspammersnyahnyah>
Date: Thu Aug 15 2013 - 13:18:59 MDT

Hi Mary,
Did you run the script using all the files I uploaded? It takes me more
than 5 minutes to finish; and it uses about 15GB memory, which causes it to
crash sometimes. I run it on a ubuntu 64bit and 8-core i7 cpu. I wonder if
there is an efficient and less-costly way to do it.

On Thu, Aug 15, 2013 at 1:54 PM, Mary Haley <haley@ucar.edu> wrote:

> Hi Xin,
>
> I'm looking at your script now. It takes about 1 minute to run on my Mac
> (wall clock time). Is this about what you're seeing?
>
> I think I might be able to improve the speed by cleaning up some of the do
> loops.
>
> --Mary
>
> On Aug 14, 2013, at 1:18 PM, Xin Xi wrote:
>
> > Hi Mary,
> > I have attached the script. The data files uploaded to server.
> > The interpolation is for 6 time slices, as a test, while I actually need
> to do it for much more (200+) files. Source grid is very fine (1km) and has
> a size of 20x30 degree; destination grid is 40km resolution on a
> curvilinear (WRF) grid.
> > I wonder if there is a fast method for the interpolation.
> > Thanks!
> > Xin
> >
> >
> >
> > On Wed, Aug 14, 2013 at 2:47 PM, Mary Haley <haley@ucar.edu> wrote:
> > Hi Xin,
> >
> > Without seeing your script, I'm not sure if the problem is with creating
> the weights, or applying the weights.
> >
> > Exactly how big are your original and destination grids? Are you only
> doing it for one lat/lon grid, or do you have another dimension, like level
> and/or time?
> >
> > Did you set the ESMF "PrintTimings" option to True, so you could see
> where it is spending its time?
> >
> > You can send me a script and I'll see if I can immediately tell what the
> problem is. Otherwise, I may ask for your data too, if you are able to
> provide that.
> > See:
> >
> > http://www.ncl.ucar.edu/report_bug.shtml#HowToFTP
> >
> > --Mary
> >
> >
> > On Aug 13, 2013, at 4:17 PM, Xin Xi wrote:
> >
> > > Hi,
> > > I am regridding a 2d dataset of 1-km resolution and 20 degx30 deg size
> at sinusoidal projection (MODIS LAI product) to a WRF domain of 40-km
> resolution. I use ESMF conserved method, but it takes lots of memory and
> time. Any suggestions on making it more efficient in NCL?
> > > Thanks,
> > > Xin
> > > _______________________________________________
> > > ncl-talk mailing list
> > > List instructions, subscriber options, unsubscribe:
> > > http://mailman.ucar.edu/mailman/listinfo/ncl-talk
> >
> >
> >
> >
> > --
> > Xin Xi
> > School of Earth and Atmospheric Sciences
> > Georgia Institute of Technology
> > 311 Ferst Drive, Atlanta, GA 30332-0340
> > <modis-lai.ncl>_______________________________________________
> > ncl-talk mailing list
> > List instructions, subscriber options, unsubscribe:
> > http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>
>

-- 
*Xin Xi
School of Earth and Atmospheric Sciences
Georgia Institute of Technology
311 Ferst Drive, Atlanta, GA 30332-0340*

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Thu Aug 15 13:19:19 2013

This archive was generated by hypermail 2.1.8 : Wed Sep 04 2013 - 10:27:57 MDT