I’m trying to interpolate a raster from a shapefile (points or contours). The shapefile is a 2ft contour lidar-derived elevation and I need a 10 ft resolution DEM. Obviously even on a dual core dual processor 64 bit workstation with 8 GB of ram I couldn’t complete the computation. Is there an automated way to slice the domain in smaller pieces, carry out the computation and piece them back together afterward? I’m thinking of writing a script but i was wondering if it has already been done.
Thanks
kapo coulibaly wrote:
I'm trying to interpolate a raster from a shapefile (points or contours).
The shapefile is a 2ft contour lidar-derived elevation and I need a 10 ft
resolution DEM.
Extracting contours discards a lot of information, results are more
detailed if a raster surface is interpolated straight from the LiDAR
points. Try any of the v.surf.* modules with the LiDAR points, not the
contours.
Obviously even on a dual core dual processor 64 bit
workstation with 8 GB of ram I couldn't complete the computation.
How many cells (rows, columns) are in the current region? Did you use
r.surf.contour? The -s flag might help to reduce memory requirements a
bit for very large datasets.
Markus M
Is there
an automated way to slice the domain in smaller pieces, carry out the
computation and piece them back together afterward? I'm thinking of writing
a script but i was wondering if it has already been done.
Thanks
_______________________________________________
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user
I tried v.surf.contour and after about 24 hrs of computation the progress bar was still nowhere to be seen. I’ll try to acquire the points from the county. But i converted the contours to points and tried the whole series of v.surf (rst, bspline, idw …etc) with the same result as i stated previously. But with 50 ft resolution I can complete the computation.
Here is the output for g.region to answer your question about nrows and ncolumns
north: 808710.758065
south: 736239.936599
west: 377975.676116
east: 465460.787903
nsres: 10.00149344
ewres: 9.99829849
rows: 7246
cols: 8750
cells: 63402500
Cheers
On Mon, Sep 20, 2010 at 10:37 AM, Markus Metz <markus.metz.giswork@googlemail.com> wrote:
kapo coulibaly wrote:
I’m trying to interpolate a raster from a shapefile (points or contours).
The shapefile is a 2ft contour lidar-derived elevation and I need a 10 ft
resolution DEM.
Extracting contours discards a lot of information, results are more
detailed if a raster surface is interpolated straight from the LiDAR
points. Try any of the v.surf.* modules with the LiDAR points, not the
contours.
Obviously even on a dual core dual processor 64 bit
workstation with 8 GB of ram I couldn’t complete the computation.
How many cells (rows, columns) are in the current region? Did you use
r.surf.contour? The -s flag might help to reduce memory requirements a
bit for very large datasets.
Markus M
Is there
an automated way to slice the domain in smaller pieces, carry out the
computation and piece them back together afterward? I’m thinking of writing
a script but i was wondering if it has already been done.
Thanks
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user
kapo coulibaly wrote:
I tried v.surf.contour
Please try r.surf.contour
and after about 24 hrs of computation the progress
bar was still nowhere to be seen. I'll try to acquire the points from the
county. But i converted the contours to points and tried the whole series of
v.surf (rst, bspline, idw ...etc) with the same result as i stated
previously. But with 50 ft resolution I can complete the computation.
Here is the output for g.region to answer your question about nrows and
ncolumns
...
rows: 7246
cols: 8750
cells: 63402500
Weird, that's only 63 million cells, shouldn't cause memory troubles.
With 8GB RAM, interpolations should be no problem (I tested with
several 100 million cells), but depending on the settings, can take
some time.
Markus M
Markus Metz wrote:
kapo coulibaly wrote:
> I'm trying to interpolate a raster from a shapefile (points or
> contours).
> The shapefile is a 2ft contour lidar-derived elevation and I need a 10
> ft
> resolution DEM.
Extracting contours discards a lot of information, results are more
detailed if a raster surface is interpolated straight from the LiDAR
points. Try any of the v.surf.* modules with the LiDAR points, not the
contours.
> Obviously even on a dual core dual processor 64 bit
> workstation with 8 GB of ram I couldn't complete the computation.
How many cells (rows, columns) are in the current region? Did you use
r.surf.contour? The -s flag might help to reduce memory requirements a
bit for very large datasets.
Markus M
> Is there
> an automated way to slice the domain in smaller pieces, carry out the
> computation and piece them back together afterward? I'm thinking of
> writing
> a script but i was wondering if it has already been done.
>
> Thanks
>
> _______________________________________________
> grass-user mailing list
> grass-user@lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/grass-user
>
>
I tried r.surf.contour but using the vector contours converted to raster, it didn’t work either. I’ll try the actual lidar point and get back to you. Maybe the contours leave wide empty spaces (it is a pretty flat terrain) and it makes computation more difficult.
Thanks.
On Mon, Sep 20, 2010 at 11:19 AM, Markus Metz <markus.metz.giswork@googlemail.com> wrote:
kapo coulibaly wrote:
I tried v.surf.contour
Please try r.surf.contour
and after about 24 hrs of computation the progress
bar was still nowhere to be seen. I’ll try to acquire the points from the
county. But i converted the contours to points and tried the whole series of
v.surf (rst, bspline, idw …etc) with the same result as i stated
previously. But with 50 ft resolution I can complete the computation.
Here is the output for g.region to answer your question about nrows and
ncolumns
…
rows: 7246
cols: 8750
cells: 63402500
Weird, that’s only 63 million cells, shouldn’t cause memory troubles.
With 8GB RAM, interpolations should be no problem (I tested with
several 100 million cells), but depending on the settings, can take
some time.
Markus M
Markus Metz wrote:
kapo coulibaly wrote:
I’m trying to interpolate a raster from a shapefile (points or
contours).
The shapefile is a 2ft contour lidar-derived elevation and I need a 10
ft
resolution DEM.
Extracting contours discards a lot of information, results are more
detailed if a raster surface is interpolated straight from the LiDAR
points. Try any of the v.surf.* modules with the LiDAR points, not the
contours.
Obviously even on a dual core dual processor 64 bit
workstation with 8 GB of ram I couldn’t complete the computation.
How many cells (rows, columns) are in the current region? Did you use
r.surf.contour? The -s flag might help to reduce memory requirements a
bit for very large datasets.
Markus M
Is there
an automated way to slice the domain in smaller pieces, carry out the
computation and piece them back together afterward? I’m thinking of
writing
a script but i was wondering if it has already been done.
Thanks
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user
hello friends....
I have been trying to convert a contour map to DEM map. First I opened
Spearfish Location and imported the contour map(this map is of Powai
city-Mumbai, India) using v.in.ogr and creating a new location by name IITBP
and having set the -o option for overriding projection. But while doing that
I got the following warning :
driver: dbf
database: $GISDBASE/$LOCATION_NAME/$MAPSET/dbf/
But the location got created successfully.
After this I exited Spearfish location and entered IITBP location. I just
saw the region settings by g.region -m and found:
n=19.1431633152778
s=19.1237958119444
w=72.9004772163889
e=72.9195980877778
nsres=107.19300143
ewres=100.58320921
rows=20
cols=20
cells=400
After this I plotted the contour map which got displayed properly on the
display window.(According to me importing the map was correct because it got
displayed properly).
Now inorder to use "r.surf.contour" I had to first convert it to raster map
using "v.to.rast".
After running v.to.rast I got the resulting raster map to be faded(may be
because of the resolution).Therefore I changed the Grid resolution of 2d to
"0.00001" using the g.region command and re-run the v.to.rast command. Now
when I plotted this raster file it was displayed on monitor perfectly. Now
keeping the same resolution but zooming a small portion of the raster map
(raster contour map) I ran the command "r.surf.contour". This command
executed successfully and the raster map is visible in the monitor. But when
i try to view this map using NVIZ it gives error saying: """"loading
failed"""
I am new to GRASS somebody please help me to solve this problem.
Thanks in advance......
--
View this message in context: http://osgeo-org.1803224.n2.nabble.com/High-resolution-raster-interpolation-tp5550605p5685236.html
Sent from the Grass - Users mailing list archive at Nabble.com.