[GRASS-user] Re: Strange r.slope.aspect output of Aster dem

[Carlos Grohmann]

Leonidas, you're facing an issue that arises when you reproject a grid (in
your case the DEM)
from a geographic coordinate system (lat-long) to UTM or other planar
coordinate system.
When you reproject, the cells will shift and skew a bit, and this will
create those 'fishnet'
lines in your projected raster (you can also see them in a shaded relief).

I discussed it briefly in a IJGIS paper, and to me, the option is to
convert your original (latlong) data
to vector points, then project the points to UTM (or other) and then
interpolate the vector points
to create a new raster in the new coordinate system. This will eliminate
those artifacts.

I also remember that Dylan Baudette had a page showing the outputs from
gdalward, and you could see the artifacs there as well, but it seems that
page is offline...

Here is the paper:

Grohmann, C.H., Steiner, S.S., 2008. SRTM resample with Short Distance-Low
Nugget Kriging.
International Journal of Geographical Information Science, 22 (8):895-906.
direct link (no subscription needed):
http://igc.usp.br/pessoais/guano/downloads/Grohmann_Steiner_2008_IJGIS_uncorrected_proof.pdf


best

Carlos





On Thu, Mar 15, 2012 at 11:01, Glynn Clements <glynn at gclements.plus.com>wrote:

>
> leonidas wrote:
> > Thank you for your answers!
>
> > I did g.region rast=aster_dem_ggrs and then I run again r.slope.aspect
> but
> > the problem still remains.
> >
> > Here is the whole process.
> > 1. I download the tiles from
> http://www.gdem.aster.ersdac.or.jp/search.jsp
> > 2. Merge them: *gdalwarp ASTG* mosaic.tif*
> > 3. Reproject the raster to greek grid: *gdalwarp -s_srs EPSG:4326 -t_srs
> > EPSG:2100 mosaic.tif dem_temp.tif*
>
> This is the problem. Resampling introduces artifacts which affect the
> gradient calculation. Ideally, you should calculate the slope and
> aspect on the original data, then re-project the result. Failing that,
> either use the -r switch to select a more suitable resampling method
> (such as cubic, cubicspline or lanczos) or filter the data with e.g.
> r.mfilter or r.neighbors prior to running r.slope.aspect.
>
> --
> Glynn Clements <glynn at gclements.plus.com>
> _______________________________________________
> grass-user mailing list
> grass-user at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/grass-user
>



-- 
Prof. Carlos Henrique Grohmann - Geologist D.Sc.
Institute of Geosciences - Univ. of São Paulo, Brazil
---
http://www.igc.usp.br/pessoais/guano
http://www.igc.usp.br/openstereo<http://www.igc.usp.br/index.php?id=openstereo>
http://lattes.cnpq.br/5846052449613692 (CV)
---
Twitter: @CarlosGrohmann
http://carlosgrohmann.tumblr.com/
________________
Can’t stop the signal.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/grass-user/attachments/20120315/97873027/attachment.html