[GRASS-user] r.horizon problems with large DEM
John Tate
john.tate at ntlworld.com
Thu Dec 3 11:42:54 EST 2009
Hi Joshua,
On Tuesday 01 December 2009 14:14:07 Markus Neteler wrote:
> Joshua,
>
> On Mon, Nov 30, 2009 at 5:50 PM, Joshua Campbell
>
> <jcampbell.geo at gmail.com> wrote:
> > Markus,
> >
> > Thanks for the input. I reviewed the r.sun page and unfortunately r.sun
> > doesn't provide the specific horizon angle provided by r.horizon. My goal
> > is not to compute solar irradiance.
>
> ok.
>
> > I was able to run r.horizon on a subset that was ~4900x9900. The DEM I'm
> > trying to use in 30000x78000 (but is has quite a bit of null values).
> >
> > Do you know if there is pixel dimension limit to r.horizon?
>
> I suspect that you have hit the 2GB limit but...
Hit a similar limit with v.outlier but that was probably just over c.1100x1100
All I could do was divide my area into 70 regions.
>
> > I am using the 64-bit package available through the Ubuntu site -- would
> > it make a difference if I built the code from source?
>
> ... if it is *really* a 64 bit binary that should not matter.
> If 32bit, then there is the --enable-largefile flag to configure
> the GRASS source code before compilation.
>
> > Do anyone know of a script that would iteratively set a region, compute
> > r.horizon (with a buffer), move the region, and r.patch the resulting
> > horizon files?
> > If anyone is interested in writing this script, we could discuss
> > compensation (contact directly at jcampbell.geo at gmail.com)
>
> Sounds good :) Unfortunately I have no time...
>
> best
> Markus
Still trying to learn the scripting thing myself, the 70 regions was a bore
and I don't want to have to do that again. So sorry, can't help. Though
regions and r.patch is the way when scripting and memory limits are a problem.
Have you tried using a mask so the NULL values are ignored? Does that help? or
mean less regions?
John
> _______________________________________________
> grass-user mailing list
> grass-user at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/grass-user
>
More information about the grass-user
mailing list