Assuming the 10m resolution of the DEM was too coarse I used g.region to
change the resolution to 1m, then ran r.resamp.interp with method=bicubic.
This was followe by r.shade.relief. The resulting map looks like a
basin-and-range province, with mountains running north-south in parallel.
That's not what the basin and surrounding area looks like.
I've no idea what I'm doing incorrectly that I cannot produce a proper
shaded relief map. I can provide command lines, input DEM and output map if
any of this helps. The maps can be provided as screenshots if that works.
Assuming the 10m resolution of the DEM was too coarse I used g.region to
change the resolution to 1m, then ran r.resamp.interp with method=bicubic.
This was followe by r.shade.relief. The resulting map looks like a
basin-and-range province, with mountains running north-south in parallel.
That's not what the basin and surrounding area looks like.
Run r.slope.aspect on the interpolated map, and check for grid-like
artifacts.
How good is the underlying data? If it's noisy, it may need to be
filtered before interpolating.
Also: which version of GRASS are you using? There was a bug in the
cubic interpolation code, but that was fixed quite a long time ago.
On Tue, Jan 5, 2010 at 2:32 AM, Rich Shepard <rshepard@appl-ecosys.com> wrote:
On Tue, 5 Jan 2010, Glynn Clements wrote:
Run r.slope.aspect on the interpolated map, and check for grid-like
artifacts.
Glynn,
Did this. Yes, there is a grid-like appearance to the resulting map.
this sounds like a resolution mismatch. If you want to significantly change
the resolution of a raster DEM, use
- r.resamp.interp - Resamples raster map layers to a finer grid using
interpolation.
- r.resamp.stats - Resamples raster map layers to a coarser grid using
aggregation.
How good is the underlying data? If it's noisy, it may need to be
filtered before interpolating.
How do I check for noisy data? I have all the basic information about the
source data and derived maps.
Using r.shaded.relief gives a good visual impression. Or create a polar diagram
of the aspect map:
> Run r.slope.aspect on the interpolated map, and check for grid-like
> artifacts.
Glynn,
Did this. Yes, there is a grid-like appearance to the resulting map.
> How good is the underlying data? If it's noisy, it may need to be
> filtered before interpolating.
How do I check for noisy data? I have all the basic information about the
source data and derived maps.
One trick is to create a filtered map with e.g. r.neighbors
method=average or r.mfilter.fp, and subtracting the original map from
the resulting map. This will help to highlight any artifacts
introduced by processing of the source data.
Any such processes should be performed at the map's native resolution,
to ensure that artifacts aren't introduced by nearest-neighbour
resampling.
One thing which can introduce grid-like artifacts is nearest-neighbour
resampling (including re-projection) at near-unity scale factors.