[Gvsig_english] BIG problems with reprojecting AGD66 to GDA94
Juan Lucas Dominguez Rubio
jldominguez at prodevelop.es
Tue Dec 1 20:55:40 CET 2009
Hello, Simon.
The GSB grid files I know are usually said to have an error of only a few centimeters. Something like: "In 95% of the cases, the error in the transformation will be smaller than 5 cm".
After doing some zoom in and out with the example seen in the video, I believe that the error for that little shapefile is well below one milimeter (too good to be true?), so I suspect that the points I extracted from your DBF were actually a subset of the points that were used to create the grid file. In other words, the grid needs not perform any interpolation because it happens to know exactly the output value for every input coordinates. The little divergence might be caused by the unavoidable roundings performed by the computer. This does not mean the grid is not working. Simply means that the error seen in the video does not give an idea of the grid's accuracy. This is only a hypothesis, but I wouldn't be surprised if you find an error of, for example, 5 cm if you do a test with a new well-known point.
By the way, did you hear about this last summer (or was it winter?):
http://www.cosmosmagazine.com/news/2875/massive-quake-moves-new-zealand-closer-australia
Maybe we should not take those grids so seriously, haha.
Regards,
Juan Lucas Domínguez Rubio
---
Prodevelop SL, Valencia (España)
Tlf.: 96.351.06.12 -- Fax: 96.351.09.68
http://www.prodevelop.es <http://www.prodevelop.es/>
---
________________________________
De: Simon Cropper [mailto:scropper at botanicusaustralia.com.au]
Enviado el: mar 01/12/2009 13:52
Para: gvsig_internacional at listserv.gva.es
CC: Juan Lucas Dominguez Rubio
Asunto: RE: Re: [Gvsig_english] BIG problems with reprojecting AGD66 to GDA94
Juan Lucas,
Do you know I have a range of programs that just don't like odd names or paths, especially those with spaces, so I name all project directories and files so there are no spaces present at all. Avoids problems.
On receipt of your email it dawned on me that gvSIG may be one of those programs. Since I avoid spaces myself I would not have encountered this problem except in getting a file from a third party (I was unclear whether gsb files could be renamed, so left the file as is).
I reran the tutorial you provided after renaming the gsb file "A66National" and guess what, it worked. Add a space in the grid file name and it does not work. Remove all spaces at it works again. Funnily enough, the program did not care about '(' or ')' or '.', just spaces. Go figure...
In regards to seeing "NT v2 Grids..." I reloaded gvSIG+Sextante on a new machine and this was not in the new install, "Grids" appeared in the drop down box the same as seen in your video.
So case closed. This whole episode boiled down to a glitch in the file opening routine. On further investigation I found that gvSIG tolerates spaces in directory names but not in filenames. So the lesson I have learnt here, apart from "gvSIG works with the Australia Tranformation Grid", is that "gvSIG requires all data files to have names WITHOUT any spaces".
Again, thanks for you help. I really appreciate it.
Cheers Simon
------- Original Message ------- On 1/12/2009 09:32 AM Juan Lucas Dominguez Rubio wrote:
Re: [Gvsig_english] BIG problems with reprojecting AGD66 to GDA94
Hello, Simon: I renamed the gsb file to 'a66_national.gsb' because I don't like white spaces, brackets,etc. I used gvSIG 1.1.2, not gvSIG 1.9. I have now tried 1.9 and I think it does not work, though a similar gsb grid for Spain does work. Another option (out of gvSIG) is using the FWTools package ( http://fwtools.maptools.org <http://fwtools.maptools.org/> ). It provides a series of text-mode tools to be started from a DOS-style window or terminal if you are using Linux. You can use the grid with it too. This worked for me: EPSG:4202 to EPSG:4283 ogr2ogr -f "ESRI Shapefile" -s_srs "+proj=longlat +ellps=aust_SA +nadgrids=./a66_national.gsb +wktext" -t_srs EPSG:4283 target_4283.shp source_4202.shp EPSG:4283 to EPSG:4202 ogr2ogr -f "ESRI Shapefile" -s_srs "+proj=longlat +ellps=GRS80 +towgs84=0,0,0,0,0,0,0 +nadgrids=./a66_national.gsb +wktext" -t_srs EPSG:4202 target_4202.shp source_4283.shp Shapefiles and grid need to be in the same folder. No
tice that the two last parameters are Regards, Juan Lucas Dom?nguez Rubio
--- Prodevelop SL, Valencia (Espa?a) Tlf.: 96.351.06.12 -- Fax: 96.351.09.68
http://www.prodevelop.es <http://www.prodevelop.es/>
---
De: gvsig_internacional-bounces at listserv.gva.es en nombre de Simon Cropper (Botanicus Australia Pty Ltd)
Enviado el: mar 01/12/2009 6:22
Para: Users and Developers mailing list
Asunto: Re: [Gvsig_english] BIG problems with reprojecting AGD66 to GDA94
Hi Juan Lucus,
I followed your instructions in gvSIG Version 1.9 (Build 1253) and it
does not work.
The only difference is that when I select transformation the following
drop down list appears -- notice that "NT v2 grid file" appears rather
than "Grids" as in your video. I only see 'Grids" if I use gvSIG Version
1.1.1 (Build 1022), but even so, if I repeat you tutorial it still does
not work.
Note also that the GRID file is named differently. The link I pointed to
allows me to download "A66 National (13.09.01).gsb". I notice that you
have "A66 National" in your video. Where did you get your Grid File?
None of the links I provided point to a "A66 National" file - maybe it
is in a different format. Also why is my version showing NT v2 format
grid file, and yours is not?
That fact that you can get it to work is encouraging. In the nutshell
the question is why is my setup different than yours.
Cheers Simon
Simon Cropper
Botanicus Australia Pty Ltd
PO Box 160, Sunshine, Victoria 3020.
P: 9311 5822. M: 041 830 3437.
mailto: scropper at botanicusaustralia.com.au
<mailto:scropper at botanicusaustralia.com.au> <mailto:scropper at botanicusaustralia.com.au%3E>
web: www.botanicusaustralia.com.au <http://www.botanicusaustralia.com.au> <http://www.botanicusaustralia.com.au%3e/>
Juan Lucas Dominguez Rubio wrote:
> Excuse me, Simon. The Australian GSB grid file does work in gvSIG:
>
> http://www.youtube.com/watch?v=eDfgMFlUnWo
> (full screen recommended)
>
> From EPSG:4202 to EPSG:4283: choose the left radio button (Layer SRS -
> EPSG:4202) as seen in the video.
> From EPSG:4283 to EPSG:4202: choose the right radio button (View SRS -
> EPSG:4202).
>
> Regards,
>
>
> Juan Lucas Dom?nguez Rubio
> ---
> Prodevelop SL, Valencia (Espa?a)
> Tlf.: 96.351.06.12 -- Fax: 96.351.09.68
> http://www.prodevelop.es <http://www.prodevelop.es/> <http://www.prodevelop.es/> <http://www.prodevelop.es/%3E>
> ---
>
> ------------------------------------------------------------------------
> *De:* gvsig_internacional-bounces at listserv.gva.es en nombre de Simon
> Cropper (Botanicus Australia Pty Ltd)
> *Enviado el:* lun 30/11/2009 1:36
> *Para:* Users and Developers mailing list
> *Asunto:* Re: [Gvsig_english] BIG problems with reprojecting AGD66 to
> GDA94
>
> Juan Lucas,
>
> Thanks for the feedback. So that others are able to follow the thread
> and have the outcome at their disposal I will repeat what has been
> stated so far both in this email and other that you have sent me.
>
> *OBSERVATION A*
>
> If you are using the reprojection tool and converting between systems in
> the same datum then the tool will work.
>
> This means if you are converting between AGD66 GEO <=> AGD66 AMG55 <=>
> AGD66 VicGrid there are no problems.
>
> This also applies for GDA94 GEO <=> GDA94 MGA55 <=> GDA94 VicGrid.
>
> *OBSERVATION B*
>
> It is necessary to 'transform' a dataset if you are changing between
> datums. That is, moving between AGD66 to GDA94 or back the other way.
>
> The regional transformation parameters for Victoria are published by the
> Intergovernmental Committee of Surveying and Mapping in the GDA94
> Technical manual Version 2.3, Amendment 1
> (http://www.icsm.gov.au/icsm/gda/gdatm/gdav2.3.pdf).
>
> The parameters *DO NOT WORK IF PLUGGED INTO THE REPROJECTION TOOL AS IS,
> YOU NEED TO CHANGE THE SIGN (+/-) OF VARIOUS VALUES DEPENDING ON YOUR
> DESIRED OUTCOME !* This is done by manually transforming the vector
> files by typing in the following values under the various situation
> specified.
>
> The following values can be used to _correctly reproject one file to
> another file_. Note you need to ensure the signs before the various
> values are correct.
>
> _*File2File: AGD66 => GDA94*_
> Dx = 119.353
> Dy = 48.301
> Dz = -139.484
> Rx = -0.415
> Ry = -0.26
> Rx = -0.437
> Sc = 0.613
>
> _*File2File: GDA94 => AGD66*_
> Dx = -119.353
> Dy = -48.301
> Dz = 139.484
> Rx = 0.415
> Ry = 0.26
> Rx = 0.437
> Sc = -0.613
>
>
> If you are _reprojecting of the fly_. The values are reversed.
>
> *_View Projection=AGD66; File Projection=GDA94_*
> Dx = 119.353
> Dy = 48.301
> Dz = -139.484
> Rx = -0.415
> Ry = -0.26
> Rx = -0.437
> Sc = 0.613
>
> _*View Projection=GDA94; File Projection=AGD66*_
> Dx = -119.353
> Dy = -48.301
> Dz = 139.484
> Rx = 0.415
> Ry = 0.26
> Rx = 0.437
> Sc = -0.613
>
>
> What is frustrating is that these values need to be entered every time
> you make a change. Very laborious and also prone to typographical errors.
>
> The GDA Technical Manual indicates that this 'rough' transformation has
> an accuracy of about 1 metre. The observations made by Juan Lucas
> supports this claim. I also found that in points I checked the values
> were <1m from the reference points created using the transformation grid
> and that the transformation work for latitude-longitude as well as
> northing-easting.
>
> *OBSERVATION C*
>
> Although a transformation grid is available, apparently in the correct
> format, this grid does not work with gvSIG.
>
> *OBSERVATION D*
>
> Although I am not happy that the transformation grid does not work as my
> reprojected points are not exactly the same as the reference points, the
> question is whether I am in any better position if I use ArcView (the
> pseudo-reference point at the moment). The answer is NO. I reprojected
> the same source using ArcView and found that points in the two
> reprojected files were <1cm apart. Not exactly the same but close enough
> that the gvSIG measure tool did not register the difference (although it
> did plot them in different locations if you zoomed in enough). I suspect
> that the two points differ because the ArcView procedure uses an earlier
> version of the Regional Transformation Parameters.
>
> *MY SUGGESTIONS*
>
> *Suggestion 1.* The reprojection tool should produce and error, or at
> least a warning, if the source file and destination files are not in the
> same datum letting the user know that it is necessary to select an
> appropriate transformation. The impression you get when selecting a
> source DATUM-COORDINATE combo and a destination DATUM-COORDINATE combo
> that the system is automatically sorting this issue out.
>
> *Suggestion 2.* Someone should look at the transformation grid produced
> by the Intergovernmental Committee of Surveying and Mapping and figure
> out why it does not work. gvSIG can't truly become an international
> package if files apparently in the correct format don't work, and that
> values (in this case rotation values published in the literature) have
> to have their signs reversed in gvSIG to get the transformation to work
> properly.
>
> *Suggestion 3.* OK, a work around has been found but it is not simple or
> quick. Manual values need to be entered, but these vary depending on the
> situation (see above), and they need to be entered every time you
> conduct a reprojection. It would be advantageous if manually entered
> data could be saved and subsequently picked from a list. This would
> speed up the process and avoid typographical errors resulting in the
> incorrect reprojection of data.
>
> Thanks Juan Lucas for your feedback on this issue.
>
> Cheers Simon
>
> Simon Cropper
> Botanicus Australia Pty Ltd
> PO Box 160, Sunshine, Victoria 3020.
> P: 9311 5822. M: 041 830 3437.
> mailto: scropper at botanicusaustralia.com.au
> <mailto:scropper at botanicusaustralia.com.au> <mailto:scropper at botanicusaustralia.com.au%3E>
> web: www.botanicusaustralia.com.au
>
> <http://www.botanicusaustralia.com.au/>> <http://www.botanicusaustralia.com.au/%3E%3E>
>
>
>
> Juan Lucas Dominguez Rubio wrote:
> > Hello, Simon.
> >
> > I have done a test with a view in EPSG:4202 and two shapefiles (a
> > target shapefile in EPSG:4202 and a source shapefile in EPSG:4283)
> > extracted from your DBF. After adding the target shapefile (same
> > projection as the view) I have added the source shp trying a variety
> > of transformations. When you add it to the view, gvSIG lets you choose
> > among eleven available "EPSG transformations" between those two EPSG
> > codes. The one that gives me the best result is the first one:
> >
> > Transformation code: 1278
> > Name: AGD66 to GDA94 (1)
> > (we are actually doing the inverse transformation (GDA94 to AGD66) but
> > the algorithm seems to be aware of this and does it well)
> >
> > I have measured the distance between the obtained coordinates
> > (transformed source shp) and the expected coordinates (target
> > shp) for 5 random points and I get these errors:
> >
> > 1.2 m
> > 0.9 m
> > 1.57 m
> > 0.88 m
> > 0.96 m
> >
> > I have also looked into some pdf files and I have found these
> > parameters (AGD66 to GDA94) for Victoria and NSW:
> >
> > http://www.icsm.gov.au/icsm/gda/gdatm/gdav2.3.pdf
> >
> > Dx = -119.353
> > Dy = -48.301
> > Dz = 139.484
> > Rx = -0.415
> > Ry = -0.26
> > Rx = -0.437
> > Sc = -0.613
> >
> > Since we are doing the inverse transfomation, we should reverse the
> > sign for each parameter, but I don't get good results.
> >
> > I have also read somewhere that in Australia and the US, the rotation
> > parameters have the inverse meaning than in Europe (big ???), so I
> > tried these parameters (reversing the ones above except for the
> > rotation parameters):
> >
> > Dx = 119.353
> > Dy = 48.301
> > Dz = -139.484
> > Rx = -0.415
> > Ry = -0.26
> > Rx = -0.437
> > Sc = 0.613
> >
> > And indeed, I get significantly smaller errors for those 5 points:
> >
> > 0.7 m
> > 0.15 m
> > 0.75 m
> > 0.8 m
> > 0.25 m
> >
> > Those 7 parameters can be entered with the "manual transformation"
> > option. I didn't get the grid to work either, sorry, although the
> > format seems to be one supported by gvSIG. And I have not done any
> > tests with non-geographical SRS.
> >
> > Regards,
> >
> > Juan Lucas Dom?nguez Rubio
> > ---
> > Prodevelop SL, Valencia (Espa?a)
> > Tlf.: 96.351.06.12 -- Fax: 96.351.09.68
> > http://www.prodevelop.es <http://www.prodevelop.es/> <http://www.prodevelop.es/> <http://www.prodevelop.es/%3E>
> <http://www.prodevelop.es/> <http://www.prodevelop.es/%3E>
> > ---
> >
> > ------------------------------------------------------------------------
> > *De:* gvsig_internacional-bounces at listserv.gva.es en nombre de Simon
> > Cropper (Botanicus Australia Pty Ltd)
> > *Enviado el:* s?b 28/11/2009 7:36
> > *Para:* Users and Developers mailing list
> > *Asunto:* [Gvsig_english] BIG problems with reprojecting AGD66 to GDA94
> >
> > Hi Guys,
> >
> > I am having problems getting the reproject tool to work as expected with
> > Australian Projections/Datums.
> >
> > If I create a shapefile in AGD66 then reproject the file to GDA94 using
> > the reproject tool, the points are moved but only ~20m rather than 204
> > metres northeast of the original position. If I use the resulting GDA94
> > file in the AGD66 View and tell the program, it reprojects on the fly
> > and the points overlap.
> >
> > This shows (a) that the reproject on the fly facility works, and (b) I
> > appear to have chosen the wrong projection.
> >
> > The problem is I have tried reprojecting northing-eastings,
> > latitude-longitude, in EPSG and ESRI format but can't get the correct
> > reprojection. I have been using EPSG:4202 for AGD66 geographicals,
> > EPSG:4283 for GDA94 geographicals, EPSG:20255 for AGD/ AMG zone 55 and
> > EPSG:28355 for GDA94/MGA zone 55. I have also tried ESRI: 28355 and
> > ESRI:20255 with no luck.
> >
> > I have even downloaded the transformation grid but not been able to get
> > this to work either.
> >
> > Can anyone help?
> >
> > Attached is a list of points in a table that are released by the local
> > government apparently created by FME. This has the comparable
> > coordinates in 4 commonly used projections for 160 distinct points
> > across Victoria.
> >
> > Can anyone get the correct setting and procedure to get these points to
> > superimpose when reprojecting on the fly and for the reproject facility
> > to work correctly?
> >
> > I have also included some useful links to help the more industrious and
> > knowledgeable to figure out what is wrong.
> >
> >
> http://www.land.vic.gov.au/Land/lcnlc2.nsf/LinkView/9B33746B3D77040F4A256A250008E70A5B923E73477D6D65CA256E5F0013D73B
> >
> > http://www.icsm.gov.au/icsm/gda/gdatm/
> >
> > http://www.icsm.gov.au/icsm/gda/gdatm/gdav2.3.pdf
> >
> > Any help would be appreciated as I am trying to tick this off my list of
> > things gvSIG can do, and was surprised to see when resulting reprojected
> > files were compared to external reference documents they did not align.
> >
> > Any help would be greatly appreciated.
> > --
> >
> > Cheers Simon
> >
> > Simon Cropper
> > Botanicus Australia Pty Ltd
> > PO Box 160, Sunshine, Victoria 3020.
> > P: 9311 5822. M: 041 830 3437.
> > mailto: scropper at botanicusaustralia.com.au
> > <mailto:scropper at botanicusaustralia.com.au> <mailto:scropper at botanicusaustralia.com.au%3E>
> > web: www.botanicusaustralia.com.au
> >
> > <http://www.botanicusaustralia.com.au/>> <http://www.botanicusaustralia.com.au/%3E%3E>
> >
> > ------------------------------------------------------------------------
> >
> > _______________________________________________
> > Gvsig_internacional mailing list
> > Gvsig_internacional at listserv.gva.es
> > http://listserv.gva.es/cgi-bin/mailman/listinfo/gvsig_internacional
> >
> _______________________________________________
> Gvsig_internacional mailing list
> Gvsig_internacional at listserv.gva.es
> http://listserv.gva.es/cgi-bin/mailman/listinfo/gvsig_internacional
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> Gvsig_internacional mailing list
> Gvsig_internacional at listserv.gva.es
> http://listserv.gva.es/cgi-bin/mailman/listinfo/gvsig_internacional
>
_______________________________________________
Gvsig_internacional mailing list
Gvsig_internacional at listserv.gva.es
http://listserv.gva.es/cgi-bin/mailman/listinfo/gvsig_internacional
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://listserv.gva.es/pipermail/gvsig_internacional/attachments/20091201/7928e677/attachment.htm
More information about the Gvsig_internacional
mailing list