[GRASS-dev] weird boundary box appears around reprojected raster map

[Isaac Ullah]

I just tried reinstalling with newest GRASS 6.4 binary snapshot
(grass-6.4.svn-i686-pc-linux-gnu-11_10_2008.tar.gz), and trying r.proj with
a different map. I am trying to reproject from a WGS84, UTM Z36N location
(that I know is correct) to a wgs84 Longlat location I created
"interactively" in the shell with the "projection values" option at GRASS
start up. The first image linked below is the original map. You can see that
it is of a single watershed with all values outside the watershed as NULL,
and is the flowacc map produced by r.watershed. The second map shows the
result of running r.proj to reproject the map from the UTM location into the
latlong location. You can see that now not only is there a small one pixel
border at what was the extents of the original map layer, bu there is a
wierd "rainbow" of valued pixels forming around the edge of the watershed,
and two "rainbow" bars shooting off towards the nw and se corners. FYI, I
get the same effect whether I choose "nearest", "bilinear", or "bicubic" for
the interpolation procedure. IMO, something is fundamentally broken in
r.proj but I can't think of what could be causing such wierd errors other
than an interpolation problem...

Original Map:
http://www.public.asu.edu/~iullah/files/MEDLAND/original_map.png

Reprojected_map:
http://www.public.asu.edu/~iullah/files/MEDLAND/more_wierd_rproj_effects.png




On Tue, Nov 18, 2008 at 1:35 PM, Isaac Ullah <isaac.ullah at asu.edu> wrote:

> Sorry to have been unclear. It seems that the new "boundary" pixels are
> created at the what seems to be edge of the input map's regional extents. To
> clarify, there is no "boundary of pixels" (eg. "frame") in the original
> raster stream seg maps produced by r.watershed, but r.proj creates a one to
> two pixel wide "frame" that outlines the reprojected maps. to repeat, there
> is NO boundary in the origianl maps, but there IS a boundary after
> reprojection. Why is this happening?
>
>
> On Tue, Nov 18, 2008 at 11:59 AM, Glynn Clements <glynn at gclements.plus.com
> > wrote:
>
>>
>> Isaac Ullah wrote:
>>
>> > Hi all, I'm attching to png's of a wierd phenomenon that just occured to
>> me
>> > when reprojecting a couple of stream segment maps (produced by
>> r.watershed)
>> > from a UTM location ot a longlat location (so I can export to kml for
>> use in
>> > GE). The maps reprojected okay, but there is a 1-2 pixel wide "border"
>> > around the reprojected maps that is approximately the same size as the
>> > original map's boundary.
>>
>> What do you mean by "the original map's boundary"?
>>
>> If the source map has a border embedded in its data, it will get
>> re-projected along with the rest of the map.
>>
>> --
>> Glynn Clements <glynn at gclements.plus.com>
>>
>
>
>
> --
>
> Isaac I Ullah, M.A.
>
> Archaeology PhD Student,
> ASU School of Evolution and Social Change
>
> Research Assistant,
> Mediterranean Landscape Dynamics Project
> ***************************************************
> isaac.ullah at asu.edu
> ullah at archaeologist.com
>
> http://www.public.asu.edu/~iullah <http://www.public.asu.edu/%7Eiullah>
> ***************************************************
>



-- 

Isaac I Ullah, M.A.

Archaeology PhD Student,
ASU School of Evolution and Social Change

Research Assistant,
Mediterranean Landscape Dynamics Project
***************************************************
isaac.ullah at asu.edu
ullah at archaeologist.com

http://www.public.asu.edu/~iullah
***************************************************
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/grass-dev/attachments/20081118/3f99a928/attachment.html