[GRASS-user] raster slices to volume

[Hamish]

> > Tomas Lanczos wrote:
> >> I have contour maps converted to rasters (rock temperatures) in 
> >> different depths (-500, -1000, -1500, -2000, -2500, -3000, -4000, -6000 
> >> meters below sea level). What is the best way to interpolate them to one 
> >> 3d raster (volume)?

Hamish wrote:
> > for evenly spaced raster slices first set the 3D region, e.g.
> >   g.region t=-500 b=-6000 tbres=500
> > 
> > then use the r.to.rast3 module to make the 3D raster volume map.
> > 
> > but you would need to interpolate intermediate levels for missing slices.
> > For -3500 you could use r.series method=average, for 4500,5000,5500 you
> > would have to use r.mapcalc manually.
> > 
> > v.vol.rst could be used to interpolate from 3D points, but RST+contour
> > lines usually is prone to artifact problems do to the uneven spatial
> > distribution of data points.
> > 
> > see also the 3D help page on the wiki site:
> >   http://grass.gdf-hannover.de/wiki/Help_with_3D

Tomas wrote:
> I experimented a lot with this, Your way described below was the easiest 
> and fastest but the result is not very satisfying for me becasue the 
> produced "stripes" in the vertical slices (see the picture).

If you look at just the -500, -1000, -1500, -2000, -2500, -3000 slices you
have, with no r.series interpolation, do you see the stripe? If 0 is the top,
it looks like the -2500m values are misplaced??

Also if you are just doing a linear interpolation you can use r.series in
multiple passes instead of r.mapcalc, first calculate -5000 from average or
-4000 and -6000 maps, then take average of new -5000 and -4000 for -4500 map,
etc. Maybe less chance to go wrong then.

You can do a slide show of the slices, with d.slide.show, xganim, or the new
Tcl animation program in 6.3. Find the min/max range of all the maps and then
set & use common color rules for all maps.


> I also tried to extract points using r.random (3%), and interpolate using
> v.vol.rst but after 12 hours of computing and reaching approx. 20% I 
> gave it up. What is the best optimalization method for interpolation in 
> v.vol.rst?

Lower XY resolution? Or just leave it running over the weekend..


Hamish



      ____________________________________________________________________________________
Get easy, one-click access to your favorites. 
Make Yahoo! your homepage.
http://www.yahoo.com/r/hs