[GRASS-user] r.shaded.relief visual banding

Hello all,
Are there known issues with "banding" in the output of the r.shaded.relief command? When I say banding, I mean the generation of regularly occurring, narrow vertical and horizontal lines. I've just created a large shaded relief map of the northern Pacific coasts of the USA and Canada from the STRM 90m v.4 Arc/Ascii files. At ~380 m intervals, ~30 m bands have been generated that crisscross the output. I'd like to know if this is a consequence of the STRM 90 data set (notably 3DEM appears to have no similar trouble with the same data), or an artifact of the shaded relief command in GRASS. I can attach an image if this would help the diagnosis.

Much appreciated,
Patrick Dolan

H Patrick,

I have had banding issues like that with r.shaded.relief when the resolution of the computational region is set higher than the resolution of the input elevation data. You can set this with ‘g.region res=’.

Cheers,
Stuart

On Thu, Nov 18, 2010 at 12:14 AM, Patrick Dolan <pdolan39@gmail.com> wrote:

Hello all,
Are there known issues with “banding” in the output of the r.shaded.relief command? When I say banding, I mean the generation of regularly occurring, narrow vertical and horizontal lines. I’ve just created a large shaded relief map of the northern Pacific coasts of the USA and Canada from the STRM 90m v.4 Arc/Ascii files. At ~380 m intervals, ~30 m bands have been generated that crisscross the output. I’d like to know if this is a consequence of the STRM 90 data set (notably 3DEM appears to have no similar trouble with the same data), or an artifact of the shaded relief command in GRASS. I can attach an image if this would help the diagnosis.

Much appreciated,
Patrick Dolan


grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user

Thanks, Stuart, but that doesn’t appear to be the problem.

Patrick

On 10-11-18 12:51 AM, Stuart Gralton wrote:

H Patrick,

I have had banding issues like that with r.shaded.relief when the resolution of the computational region is set higher than the resolution of the input elevation data. You can set this with ‘g.region res=’.

Cheers,
Stuart

On Thu, Nov 18, 2010 at 12:14 AM, Patrick Dolan <pdolan39@gmail.com> wrote:

Hello all,
Are there known issues with “banding” in the output of the r.shaded.relief command? When I say banding, I mean the generation of regularly occurring, narrow vertical and horizontal lines. I’ve just created a large shaded relief map of the northern Pacific coasts of the USA and Canada from the STRM 90m v.4 Arc/Ascii files. At ~380 m intervals, ~30 m bands have been generated that crisscross the output. I’d like to know if this is a consequence of the STRM 90 data set (notably 3DEM appears to have no similar trouble with the same data), or an artifact of the shaded relief command in GRASS. I can attach an image if this would help the diagnosis.

Much appreciated,
Patrick Dolan


grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user

Dylan, I appreciate the advice and the examples. I've succeeded in removing the "banding" but only though converting the a small subsection of the DEM into a point vector (with r.to.vect) and then back to a dem (with r.surf.rst).

It's a working, though suboptimal solution as it takes my Gen 1 MacBook (2 GB of DDR2 SDRAM and 2 Ghz Core) several hours to create the vector and then recreate a 1 by 1 degree DEM (the SRTM 90 tiles are 5 by 5 degrees-which I subdivide or suffer GRASS' code 12 error).

Cheers,
Patrick

On 10-11-17 10:48 AM, Dylan Beaudette wrote:

On Wednesday, November 17, 2010, Patrick Dolan wrote:

Hello all,
Are there known issues with "banding" in the output of the
r.shaded.relief command? When I say banding, I mean the generation of
regularly occurring, narrow vertical and horizontal lines. I've just
created a large shaded relief map of the northern Pacific coasts of the
USA and Canada from the STRM 90m v.4 Arc/Ascii files. At ~380 m
intervals, ~30 m bands have been generated that crisscross the output.
I'd like to know if this is a consequence of the STRM 90 data set
(notably 3DEM appears to have no similar trouble with the same data), or
an artifact of the shaded relief command in GRASS. I can attach an image
if this would help the diagnosis.

Much appreciated,
Patrick Dolan

Hi Patrick,

I haven't used the SRTM data, but I have seen this happen when working with
the USGS 10m NED. It seems like it has something to do with the projection, as
illustrated in this somewhat dated demonstration I put together in order to
figure out what was going on:

http://169.237.35.250/~dylan/gdalwarp/image_matrix.html

Here are some related messages from many years ago:
http://www.osgeo.org/pipermail/grass-user/2006-May/034257.html

I should probably update the original analysis based on some of Maciek's
comments in that thread, and changes that have taken place since 2006.

Cheers,
Dylan

--
Dylan Beaudette
Soil Resource Laboratory
http://casoilresource.lawr.ucdavis.edu/
University of California at Davis
530.754.7341

Looking into a different direction: how come you noticed ~30m bands
while SRTM is 3 arc sec (~90m at the equator)? Have you reprojected
and resampled the DEM? I use this banding in the output of
r.shaded.relief to check if the reprojection/resampling settings are
ok.

OTOH, SRTM v4 elevation sometimes deviates substantially, far more
than by the original SRTM vertical error, from SRTM v2.1, and the SRTM
v4 grid geometry is weird, v4 tiles won't fit into the same
computational region because of subcell shifting, shifted by 0 to 1.5
arc sec and everything in between against each other.

my 0.2c

Markus M

On Wed, Nov 17, 2010 at 5:14 PM, Patrick Dolan <pdolan39@gmail.com> wrote:

Hello all,
Are there known issues with "banding" in the output of the r.shaded.relief
command? When I say banding, I mean the generation of regularly occurring,
narrow vertical and horizontal lines. I've just created a large shaded
relief map of the northern Pacific coasts of the USA and Canada from the
STRM 90m v.4 Arc/Ascii files. At ~380 m intervals, ~30 m bands have been
generated that crisscross the output. I'd like to know if this is a
consequence of the STRM 90 data set (notably 3DEM appears to have no similar
trouble with the same data), or an artifact of the shaded relief command in
GRASS. I can attach an image if this would help the diagnosis.

Much appreciated,
Patrick Dolan
_______________________________________________
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user

Patrick Dolan wrote:

Are there known issues with "banding" in the output of the
r.shaded.relief command? When I say banding, I mean the generation of
regularly occurring, narrow vertical and horizontal lines. I've just
created a large shaded relief map of the northern Pacific coasts of the
USA and Canada from the STRM 90m v.4 Arc/Ascii files. At ~380 m
intervals, ~30 m bands have been generated that crisscross the output.
I'd like to know if this is a consequence of the STRM 90 data set
(notably 3DEM appears to have no similar trouble with the same data), or
an artifact of the shaded relief command in GRASS. I can attach an image
if this would help the diagnosis.

First, ensure that the region matches the map. nearest-neighbour
resampling will introduce artifacts in anything which calculates
derivatives.

Second, ensure that the source data is good. The simplest way to do
this is to run r.slope.aspect on the map and visually examine the
resulting slope map.

--
Glynn Clements <glynn@gclements.plus.com>