[GRASS5] projection-transformation

GRASS GRASS developer
1

The resulting UTM/international/eur50 datum location looks like this:
- GLOBE DEM 1km and GSHHS match rather good
- boundaries are shifted (why? - I didn't supervised their work)

Markus, I am glad that somebody else got this problem too!
Bill and I had to import data for various locations from arc/info into UTM
but also into state plane,
(usually shape files) and we had the vector data often shifted
compared to raster data - the difference is equivalent to
the difference in datums used in US, so at least in one case Bill
was able to re-project them through different location to get them right.
However the input data were supposed to be in the same projection.
I did not have time to figure out whether I am doing something wrong
but I am suspicious that there might be a bug - I haven't seen
people complaining about it though, so it may be my mistake
(or your student's) too - check your datums.

Helena

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

2

On Fri, Dec 22, 2000 at 08:59:41AM -0600, Helena Mitasova - staff wrote:

>The resulting UTM/international/eur50 datum location looks like this:
> - GLOBE DEM 1km and GSHHS match rather good
> - boundaries are shifted (why? - I didn't supervised their work)

Markus, I am glad that somebody else got this problem too!
Bill and I had to import data for various locations from arc/info into UTM
but also into state plane,
(usually shape files) and we had the vector data often shifted
compared to raster data - the difference is equivalent to
the difference in datums used in US, so at least in one case Bill
was able to re-project them through different location to get them right.
However the input data were supposed to be in the same projection.
I did not have time to figure out whether I am doing something wrong
but I am suspicious that there might be a bug - I haven't seen
people complaining about it though, so it may be my mistake
(or your student's) too - check your datums.

Hi Helena, hi all,

maybe it's a bug... (but where?). I made some further tests with my
dataset by analysing a point:

The west corner of the "norderney" island (one of those islands in the
picture I have sent the URL) has following points in the map (d.where):

UTM:
377261 5953152 (GSHHS dataset from v.in.gshhs [new, to be submitted soon])
377946 5953426 (GLOBE DEM 1km, generally matching to GSHHS quite well)

379592 5952535 (Students vector data from ARC/INFO)

I took a 1:100000 map of that area and got the coordinates in Gauss-Krüger
system (tmerc, bessel):
2575400 5953800 GK

These I transformed with m.proj to UTM to compare them with d.where results:
377375 5953157 UTM

It seems that the ARC/INFO vector data are shifted:
> 2km to east
> 500m to south

-----

Then I looked back into the original lat/long database:
The shifts are identical. So it's not a *.proj problem.

May there be a bug in m.in.e00 which I used for data import?

Markus

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

3

Markus Neteler wrote:

May there be a bug in m.in.e00 which I used for data import?

Well... m.in.e00 doesn't translate anything. It uses the
scale factor in projection info if it is not 1.0. A bug
here may lead to a false scale (and a non uniform shift)

Did you look at the original e00 file ? (take in dig_att the
attribute of the vector, and use it to search through
the arc section of the e00 file : you should find the same
lat/long values than in the grass vector file).

--
Michel WURTZ - DIG - Maison de la télédétection
               500, rue J.F. Breton
               34093 MONTPELLIER Cedex 5

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

4

Looking back at Markus' gif, I see that the digitized lines on the
northwest side of the image are offset to the east of the shoreline
data. The digitized lines on the east side of the image are offset to
the south-southeast of the shoreline data.

That doesn't look to me like a scaling problem or a linear offset, it
looks like a rotation. Odd. The easiest explanation for a rotation is
if the map that your students digitized in arcinfo was incorrectly
registered.

Roger Miller
Lee Wilson and Associates

Michel Wurtz wrote:

Markus Neteler wrote:

> May there be a bug in m.in.e00 which I used for data import?

Well... m.in.e00 doesn't translate anything. It uses the
scale factor in projection info if it is not 1.0. A bug
here may lead to a false scale (and a non uniform shift)

Did you look at the original e00 file ? (take in dig_att the
attribute of the vector, and use it to search through
the arc section of the e00 file : you should find the same
lat/long values than in the grass vector file).

--
Michel WURTZ - DIG - Maison de la télédétection
               500, rue J.F. Breton
               34093 MONTPELLIER Cedex 5

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

5

Hi All,

Since r.proj changed (beta9) I have had the general (totally subjective)
impression that projections of globe dem maps fit somewhat better than
those done with the old r.proj, but are shifted 0.5-1 resolution units to
the north and 0-0.5 resolution units to the east compared to gs coastline.
This happens in many different projections I have tried (laea, sweden-rt,
finland-ykj, norway-utm33). Your URL and the numbers below support those
observations.

I wonder where this smallish mismatch comes from; could be datum, could be
bug in r.proj or both.

But if this is a bug it's probably unrelated to the much larger mismatch
between the two vector maps.

Morten

On Fri, 22 Dec 2000, Markus Neteler wrote:

maybe it's a bug... (but where?). I made some further tests with my
dataset by analysing a point:

The west corner of the "norderney" island (one of those islands in the
picture I have sent the URL) has following points in the map (d.where):

UTM:
377261 5953152 (GSHHS dataset from v.in.gshhs [new, to be submitted soon])
377946 5953426 (GLOBE DEM 1km, generally matching to GSHHS quite well)

379592 5952535 (Students vector data from ARC/INFO)

I took a 1:100000 map of that area and got the coordinates in Gauss-Krüger
system (tmerc, bessel):
2575400 5953800 GK

These I transformed with m.proj to UTM to compare them with d.where results:
377375 5953157 UTM

It seems that the ARC/INFO vector data are shifted:
> 2km to east
> 500m to south

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'