Re: [GRASS-user] Reprojecting to x,y grid
On 2020-11-19 at 01:18 -08, Moritz Lennert wrote... > Maybe because of changes linked to proj6 support ? Markus Metz will > know. Yes I found this yesterday. Here is the v.proj recent commits https://github.com/OSGeo/grass/commits/master/vector/v.proj And I'm guessing it was this commit that introduced the behavior change https://github.com/OSGeo/grass/commit/ce56f30a6871356dac0db49c5faa850941e3#diff-fa03e53897bb203999fbd1b78e611058b754555626b19744b8c6f71f95c0955a -k. ___ grass-user mailing list grass-user@lists.osgeo.org https://lists.osgeo.org/mailman/listinfo/grass-user
Re: [GRASS-user] Reprojecting to x,y grid
On 18/11/20 15:01, Ken Mankoff wrote: Hello, On 2020-11-17 at 10:06 -08, Ken Mankoff wrote... I've successfully worked in rotated pole coordinates (e.g., [1]) but because I don't know what (x,y) coord contains the pole, I haven't figured out how to adapt this to the rotated pole situation as in [1]. Anyway, this isn't a rotated pole... I was incorrect - this is a rotated pole coordinate system. When I run the following code on GRASS 7.4, I get the attached graphic. The key part here is that the rotated pole is at (lon,lat) = (-200,18). grass -c EPSG:4326 Gnorm # Import something into the normal location v.import input=~/data/Zwally_2012/sectors output=Z cat << EOF > ./Gnorm/PERMANENT/PROJ_INFO name: General Oblique Transformation datum: wgs84 towgs84: 0.000,0.000,0.000 proj: ob_tran o_proj: latlon ellps: wgs84 a: 6378137.00 es: 0.0066943800 f: 298.2572235630 lat_0: 0.00 lon_0: 180.00 o_lat_p: 18.0 o_lon_p: -200.0 EOF # rotated_pole:grid_north_pole_latitude = 18. ; # rotated_pole:grid_north_pole_longitude = -200. cat << EOF > ./Gnorm/PERMANENT/PROJ_UNITS unit: degree units: degrees meters: .0174532925 EOF grass -e -c EPSG:4326 Grot grass ./Grot/PERMANENT v.proj location=Gnorm input=Z g.region vector=Z d.mon start=wx0 d.vect Z d.grid 1:0:0 The above code run in GRASS 7.4.0 on a clean Ubuntu 18.04 VM produces the attached graphic. However, when I run the exact same code in GRASS 7.8.4 it does not work. The reproject/rotate does not occur, and Greenland remains at 62-82 N, and 22-72 W. The contents of the "Gnorm" folders is appears identical. The md5sum of DEFAULT_WIND, MYNAME, PROJ_EPSG, PROJ_INFO, and PROJ_UNITS are all the same. I can do the work in GRASS 7.4 in a VM but I'd prefer to do it on the main development machine in the latest GRASS. Does anyone have any idea why this feature stopped working? Maybe because of changes linked to proj6 support ? Markus Metz will know. Moritz Moritz ___ grass-user mailing list grass-user@lists.osgeo.org https://lists.osgeo.org/mailman/listinfo/grass-user
Re: [GRASS-user] Reprojecting to x,y grid
Hello, On 2020-11-17 at 10:06 -08, Ken Mankoff wrote... > I've successfully worked in rotated pole coordinates (e.g., [1]) but > because I don't know what (x,y) coord contains the pole, I haven't > figured out how to adapt this to the rotated pole situation as in [1]. > Anyway, this isn't a rotated pole... I was incorrect - this is a rotated pole coordinate system. When I run the following code on GRASS 7.4, I get the attached graphic. The key part here is that the rotated pole is at (lon,lat) = (-200,18). grass -c EPSG:4326 Gnorm # Import something into the normal location v.import input=~/data/Zwally_2012/sectors output=Z cat << EOF > ./Gnorm/PERMANENT/PROJ_INFO name: General Oblique Transformation datum: wgs84 towgs84: 0.000,0.000,0.000 proj: ob_tran o_proj: latlon ellps: wgs84 a: 6378137.00 es: 0.0066943800 f: 298.2572235630 lat_0: 0.00 lon_0: 180.00 o_lat_p: 18.0 o_lon_p: -200.0 EOF # rotated_pole:grid_north_pole_latitude = 18. ; # rotated_pole:grid_north_pole_longitude = -200. cat << EOF > ./Gnorm/PERMANENT/PROJ_UNITS unit: degree units: degrees meters: .0174532925 EOF grass -e -c EPSG:4326 Grot grass ./Grot/PERMANENT v.proj location=Gnorm input=Z g.region vector=Z d.mon start=wx0 d.vect Z d.grid 1:0:0 The above code run in GRASS 7.4.0 on a clean Ubuntu 18.04 VM produces the attached graphic. However, when I run the exact same code in GRASS 7.8.4 it does not work. The reproject/rotate does not occur, and Greenland remains at 62-82 N, and 22-72 W. The contents of the "Gnorm" folders is appears identical. The md5sum of DEFAULT_WIND, MYNAME, PROJ_EPSG, PROJ_INFO, and PROJ_UNITS are all the same. I can do the work in GRASS 7.4 in a VM but I'd prefer to do it on the main development machine in the latest GRASS. Does anyone have any idea why this feature stopped working? Thanks, -k. ___ grass-user mailing list grass-user@lists.osgeo.org https://lists.osgeo.org/mailman/listinfo/grass-user
Re: [GRASS-user] Reprojecting to x,y grid
Hi Markus, On 2020-11-17 at 14:40 -08, Markus Neteler wrote... > On Tue, Nov 17, 2020 at 7:06 PM Ken Mankoff wrote: >> I have a data set provided on an (x,y) grid, where each coordinate >> does have a paired (lon,lat). I'd like to work on the native (x,y) >> grid, and bring some other geo-referenced data there. How do I do >> this? > > Here 2.5 wild guesses: > > If it is about "pushing around" a raster map (no squeezing): > - https://grass.osgeo.org/grass78/manuals/r.region.html > > If a rotation is needed: > - https://grass.osgeo.org/grass7/manuals/addons/i.rotate.html > > If it is about a real transform: > - convert the raster matrix to its vector center point representation; then > - https://grass.osgeo.org/grass78/manuals/v.transform.html > - rasterize again Thanks for getting me thinking down this path. This isn't about the raster. I want to work in the raster XY location. I want to reproject a vector there. I know the (lon,lat) of every (x,y). I think maybe I need to solve this outside of GRASS - I can take the (lon,lat) of each point of the geo-referenced vector, and look up the nearest (lon,lat) in the XY location, and set those as the new coordinates for the vector. I'm not yet sure what the difference is between that method and using v.rectify. I would build the control points using the same method: Take the (lon,lat) of each point of the geo-referenced vector, look up the nearest (lon,lat) in the XY location, from there jump to the (x,y) coord in the XY location, and use those as my control points. I then rectify (un-project?) the vector to the XY coords. Is the only difference that the second method would "fill in" points that I don't map between (lon,lat) and (x,y), while the first method requires that I update every point? Thanks, -k. ___ grass-user mailing list grass-user@lists.osgeo.org https://lists.osgeo.org/mailman/listinfo/grass-user
Re: [GRASS-user] Reprojecting to x,y grid
On Tue, Nov 17, 2020 at 7:06 PM Ken Mankoff wrote: > > Hi GRASS list, > > I have a data set provided on an (x,y) grid, where each coordinate does have > a paired (lon,lat). I'd like to work on the native (x,y) grid, and bring some > other geo-referenced data there. How do I do this? Here 2.5 wild guesses: If it is about "pushing around" a raster map (no squeezing): - https://grass.osgeo.org/grass78/manuals/r.region.html If a rotation is needed: - https://grass.osgeo.org/grass7/manuals/addons/i.rotate.html If it is about a real transform: - convert the raster matrix to its vector center point representation; then - https://grass.osgeo.org/grass78/manuals/v.transform.html - rasterize again But again, just some almost random ideas. best, Markus ___ grass-user mailing list grass-user@lists.osgeo.org https://lists.osgeo.org/mailman/listinfo/grass-user