Problem georeferencing images in UTM

classic Classic list List threaded Threaded
13 messages Options
Reply | Threaded
Open this post in threaded view
|

Problem georeferencing images in UTM

Carlos Henrique Grohmann de Carvalho-2
Hello all

I have two pngs of topographic maps from southeastern Brazil (a zip file with images and points is available here: https://www.dropbox.com/s/mexeh4be46pgsk6/images_georef.zip?dl=0)

I'm trying to georeference them, but with partial success only. In this image (https://www.dropbox.com/s/37xxqp4fkwpdyqh/utm_error.png?dl=0) you can see that the X coordinate is fine, but the Y coordinate of the two maps won't match.

Both maps use South American Datum 1969. I selected points near the edges, and used the UTM values. 

Interestingly, if I use Lat/Long, and place the points at the corners, using decimal degrees, the maps align perfectly.


Any help is appreciated


--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 

http://orcid.org/0000-0001-5073-5572
________________
Can’t stop the signal.

_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
Reply | Threaded
Open this post in threaded view
|

Re: Problem georeferencing images in UTM

Nicolas Cadieux
Hi,
Disable projection on the fly.  Make sure the project is in the same CRS as your maps.
Nicolas

Le 17 mai 2018 à 12:14, Carlos Henrique Grohmann de Carvalho <[hidden email]> a écrit :

Hello all

I have two pngs of topographic maps from southeastern Brazil (a zip file with images and points is available here: https://www.dropbox.com/s/mexeh4be46pgsk6/images_georef.zip?dl=0)

I'm trying to georeference them, but with partial success only. In this image (https://www.dropbox.com/s/37xxqp4fkwpdyqh/utm_error.png?dl=0) you can see that the X coordinate is fine, but the Y coordinate of the two maps won't match.

Both maps use South American Datum 1969. I selected points near the edges, and used the UTM values. 

Interestingly, if I use Lat/Long, and place the points at the corners, using decimal degrees, the maps align perfectly.


Any help is appreciated


--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 

http://orcid.org/0000-0001-5073-5572
________________
Can’t stop the signal.
_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user

_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
Reply | Threaded
Open this post in threaded view
|

Re: Problem georeferencing images in UTM

Carlos Henrique Grohmann de Carvalho-2
Thanks Nicolas, but the misalignment still persists. OTF is off, project has same CRS as data (epsg 29192). There is a ~400m difference in the N-S direction.

Carlos

On Thu, May 17, 2018 at 1:28 PM, Nicolas Cadieux <[hidden email]> wrote:
Hi,
Disable projection on the fly.  Make sure the project is in the same CRS as your maps.
Nicolas

Le 17 mai 2018 à 12:14, Carlos Henrique Grohmann de Carvalho <[hidden email]> a écrit :

Hello all

I have two pngs of topographic maps from southeastern Brazil (a zip file with images and points is available here: https://www.dropbox.com/s/mexeh4be46pgsk6/images_georef.zip?dl=0)

I'm trying to georeference them, but with partial success only. In this image (https://www.dropbox.com/s/37xxqp4fkwpdyqh/utm_error.png?dl=0) you can see that the X coordinate is fine, but the Y coordinate of the two maps won't match.

Both maps use South American Datum 1969. I selected points near the edges, and used the UTM values. 

Interestingly, if I use Lat/Long, and place the points at the corners, using decimal degrees, the maps align perfectly.


Any help is appreciated


--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 

http://orcid.org/0000-0001-5073-5572
________________
Can’t stop the signal.
_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user



--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 

http://orcid.org/0000-0001-5073-5572
________________
Can’t stop the signal.

_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
Reply | Threaded
Open this post in threaded view
|

Re: Problem georeferencing images in UTM

stevejking
When using the georeferencer some of the methods require at least 3 points,
even though you might think it not necessary. Otherwise you can get unusual
results.
Steve



--
Sent from: http://osgeo-org.1560.x6.nabble.com/QGIS-User-f4125267.html
_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
Reply | Threaded
Open this post in threaded view
|

Re: Problem georeferencing images in UTM

Nicolas Cadieux
In reply to this post by Carlos Henrique Grohmann de Carvalho-2
Hi,

I am looking at your maps now. If I understand, it works with lat long but not the UTM Coordinates.  It could be a mistake in the utm grid.  Perhaps one utm grid is in NAD27 and the other in NAD83.  That could explain the shift.  
I will look at it.

Nicolas

Le 17 mai 2018 à 12:42, Carlos Henrique Grohmann de Carvalho <[hidden email]> a écrit :

Thanks Nicolas, but the misalignment still persists. OTF is off, project has same CRS as data (epsg 29192). There is a ~400m difference in the N-S direction.

Carlos

On Thu, May 17, 2018 at 1:28 PM, Nicolas Cadieux <[hidden email]> wrote:
Hi,
Disable projection on the fly.  Make sure the project is in the same CRS as your maps.
Nicolas

Le 17 mai 2018 à 12:14, Carlos Henrique Grohmann de Carvalho <[hidden email]> a écrit :

Hello all

I have two pngs of topographic maps from southeastern Brazil (a zip file with images and points is available here: https://www.dropbox.com/s/mexeh4be46pgsk6/images_georef.zip?dl=0)

I'm trying to georeference them, but with partial success only. In this image (https://www.dropbox.com/s/37xxqp4fkwpdyqh/utm_error.png?dl=0) you can see that the X coordinate is fine, but the Y coordinate of the two maps won't match.

Both maps use South American Datum 1969. I selected points near the edges, and used the UTM values. 

Interestingly, if I use Lat/Long, and place the points at the corners, using decimal degrees, the maps align perfectly.


Any help is appreciated


--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 

http://orcid.org/0000-0001-5073-5572
________________
Can’t stop the signal.
_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user



--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 

http://orcid.org/0000-0001-5073-5572
________________
Can’t stop the signal.

_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
Reply | Threaded
Open this post in threaded view
|

Re: Problem georeferencing images in UTM

Nicolas Cadieux
In reply to this post by Carlos Henrique Grohmann de Carvalho-2
Hi,
The utm grid is way off or we have the wrong epsg code.  If you geofence using 2 points (linear algo)  using a service like google earth, you will see the distortion.  The long lat coordinate are probably ok but I did not test them.

You can either use the long lat or use more points.  With 4 points and using the thin plate spline, I got good results.
Nicolas

Le 17 mai 2018 à 12:42, Carlos Henrique Grohmann de Carvalho <[hidden email]> a écrit :

Thanks Nicolas, but the misalignment still persists. OTF is off, project has same CRS as data (epsg 29192). There is a ~400m difference in the N-S direction.

Carlos

On Thu, May 17, 2018 at 1:28 PM, Nicolas Cadieux <[hidden email]> wrote:
Hi,
Disable projection on the fly.  Make sure the project is in the same CRS as your maps.
Nicolas

Le 17 mai 2018 à 12:14, Carlos Henrique Grohmann de Carvalho <[hidden email]> a écrit :

Hello all

I have two pngs of topographic maps from southeastern Brazil (a zip file with images and points is available here: https://www.dropbox.com/s/mexeh4be46pgsk6/images_georef.zip?dl=0)

I'm trying to georeference them, but with partial success only. In this image (https://www.dropbox.com/s/37xxqp4fkwpdyqh/utm_error.png?dl=0) you can see that the X coordinate is fine, but the Y coordinate of the two maps won't match.

Both maps use South American Datum 1969. I selected points near the edges, and used the UTM values. 

Interestingly, if I use Lat/Long, and place the points at the corners, using decimal degrees, the maps align perfectly.


Any help is appreciated


--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 

http://orcid.org/0000-0001-5073-5572
________________
Can’t stop the signal.
_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user



--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 

http://orcid.org/0000-0001-5073-5572
________________
Can’t stop the signal.

_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
Reply | Threaded
Open this post in threaded view
|

Re: Problem georeferencing images in UTM

Carlos Henrique Grohmann de Carvalho-2
Thanks for the input Nicolas

These are official maps so the utm grid is ok. I figured that I was using a linear transformation, with the option to just create the world file, which works for control points at the corners (lat/long), since the boundaries of the map is defined by latitude/longitude. With UTM, this simple transformation won't work, but TPS will give the correct results.

Carlos


On Thu, May 17, 2018 at 7:28 PM, Nicolas Cadieux <[hidden email]> wrote:
Hi,
The utm grid is way off or we have the wrong epsg code.  If you geofence using 2 points (linear algo)  using a service like google earth, you will see the distortion.  The long lat coordinate are probably ok but I did not test them.

You can either use the long lat or use more points.  With 4 points and using the thin plate spline, I got good results.
Nicolas

Le 17 mai 2018 à 12:42, Carlos Henrique Grohmann de Carvalho <[hidden email]> a écrit :

Thanks Nicolas, but the misalignment still persists. OTF is off, project has same CRS as data (epsg 29192). There is a ~400m difference in the N-S direction.

Carlos

On Thu, May 17, 2018 at 1:28 PM, Nicolas Cadieux <[hidden email]> wrote:
Hi,
Disable projection on the fly.  Make sure the project is in the same CRS as your maps.
Nicolas

Le 17 mai 2018 à 12:14, Carlos Henrique Grohmann de Carvalho <[hidden email]> a écrit :

Hello all

I have two pngs of topographic maps from southeastern Brazil (a zip file with images and points is available here: https://www.dropbox.com/s/mexeh4be46pgsk6/images_georef.zip?dl=0)

I'm trying to georeference them, but with partial success only. In this image (https://www.dropbox.com/s/37xxqp4fkwpdyqh/utm_error.png?dl=0) you can see that the X coordinate is fine, but the Y coordinate of the two maps won't match.

Both maps use South American Datum 1969. I selected points near the edges, and used the UTM values. 

Interestingly, if I use Lat/Long, and place the points at the corners, using decimal degrees, the maps align perfectly.


Any help is appreciated


--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 

http://orcid.org/0000-0001-5073-5572
________________
Can’t stop the signal.
_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user



--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 

http://orcid.org/0000-0001-5073-5572
________________
Can’t stop the signal.



--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 

http://orcid.org/0000-0001-5073-5572
________________
Can’t stop the signal.

_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
Reply | Threaded
Open this post in threaded view
|

Re: Problem georeferencing images in UTM

listsen

I haven't been able to open the dataset but going through the discussion I would like to offer some assistance since I believe there are some misconceptions here:

First, the data has nothing to do with the NAD27 / NAD83 datum's since these are North American datum's and they don't apply I believe in South America. Moreover, the separation between NAD27 to NAD83 is about 100m. I think Carlos mention something like being 400m off.

Another point of confusion is about the projection of Google Earth. Google is using a Web Mercator projection based on a perfect sphere. This has nothing to do with the UTM that has already incorporate the Earth Flattening values into the Northing and Easting equations.

If I find some time tomorrow, I will come back and give the proper equations showing the difference between Google's projection and UTM.

-0x906


On 2018-05-17 19:53, Carlos Henrique Grohmann de Carvalho wrote:

Thanks for the input Nicolas
 
These are official maps so the utm grid is ok. I figured that I was using a linear transformation, with the option to just create the world file, which works for control points at the corners (lat/long), since the boundaries of the map is defined by latitude/longitude. With UTM, this simple transformation won't work, but TPS will give the correct results.
 
Carlos
 

On Thu, May 17, 2018 at 7:28 PM, Nicolas Cadieux <[hidden email]> wrote:
 
Hi,
The utm grid is way off or we have the wrong epsg code.  If you geofence using 2 points (linear algo)  using a service like google earth, you will see the distortion.  The long lat coordinate are probably ok but I did not test them.
 
You can either use the long lat or use more points.  With 4 points and using the thin plate spline, I got good results.
Nicolas

Le 17 mai 2018 à 12:42, Carlos Henrique Grohmann de Carvalho <[hidden email]> a écrit :

Thanks Nicolas, but the misalignment still persists. OTF is off, project has same CRS as data (epsg 29192). There is a ~400m difference in the N-S direction.
 
Carlos

On Thu, May 17, 2018 at 1:28 PM, Nicolas Cadieux <[hidden email]> wrote:
 
Hi,
Disable projection on the fly.  Make sure the project is in the same CRS as your maps.
Nicolas

Le 17 mai 2018 à 12:14, Carlos Henrique Grohmann de Carvalho <[hidden email]> a écrit :

Hello all
 
I have two pngs of topographic maps from southeastern Brazil (a zip file with images and points is available here: https://www.dropbox.com/s/mexeh4be46pgsk6/images_georef.zip?dl=0)
 
I'm trying to georeference them, but with partial success only. In this image (https://www.dropbox.com/s/37xxqp4fkwpdyqh/utm_error.png?dl=0) you can see that the X coordinate is fine, but the Y coordinate of the two maps won't match.
 
Both maps use South American Datum 1969. I selected points near the edges, and used the UTM values. 
 
Interestingly, if I use Lat/Long, and place the points at the corners, using decimal degrees, the maps align perfectly.
 
 
Any help is appreciated

 
--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 
 
http://orcid.org/0000-0001-5073-5572
________________
Can't stop the signal.
_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user


 
--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 
 
http://orcid.org/0000-0001-5073-5572
________________
Can't stop the signal.


 
--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 
 
http://orcid.org/0000-0001-5073-5572
________________
Can't stop the signal.

_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user



_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
Reply | Threaded
Open this post in threaded view
|

Re: Problem georeferencing images in UTM

Kirk Schmidt
In reply to this post by Carlos Henrique Grohmann de Carvalho-2

Hi Carlos:

I georeferenced 2 screen shots of your maps and obtained good results.  I would have a close look at the coordinates of your ground control points and make sure you selected the utm grid intersection as a control point versus the corner of the mapsheet which is a lat/long grid. It is easy to make a simple mistake

Kirk



On 5/17/2018 8:53 PM, Carlos Henrique Grohmann de Carvalho wrote:
Thanks for the input Nicolas

These are official maps so the utm grid is ok. I figured that I was using a linear transformation, with the option to just create the world file, which works for control points at the corners (lat/long), since the boundaries of the map is defined by latitude/longitude. With UTM, this simple transformation won't work, but TPS will give the correct results.

Carlos


On Thu, May 17, 2018 at 7:28 PM, Nicolas Cadieux <[hidden email]> wrote:
Hi,
The utm grid is way off or we have the wrong epsg code.  If you geofence using 2 points (linear algo)  using a service like google earth, you will see the distortion.  The long lat coordinate are probably ok but I did not test them.

You can either use the long lat or use more points.  With 4 points and using the thin plate spline, I got good results.
Nicolas

Le 17 mai 2018 à 12:42, Carlos Henrique Grohmann de Carvalho <[hidden email]> a écrit :

Thanks Nicolas, but the misalignment still persists. OTF is off, project has same CRS as data (epsg 29192). There is a ~400m difference in the N-S direction.

Carlos

On Thu, May 17, 2018 at 1:28 PM, Nicolas Cadieux <[hidden email]> wrote:
Hi,
Disable projection on the fly.  Make sure the project is in the same CRS as your maps.
Nicolas

Le 17 mai 2018 à 12:14, Carlos Henrique Grohmann de Carvalho <[hidden email]> a écrit :

Hello all

I have two pngs of topographic maps from southeastern Brazil (a zip file with images and points is available here: https://www.dropbox.com/s/mexeh4be46pgsk6/images_georef.zip?dl=0)

I'm trying to georeference them, but with partial success only. In this image (https://www.dropbox.com/s/37xxqp4fkwpdyqh/utm_error.png?dl=0) you can see that the X coordinate is fine, but the Y coordinate of the two maps won't match.

Both maps use South American Datum 1969. I selected points near the edges, and used the UTM values. 

Interestingly, if I use Lat/Long, and place the points at the corners, using decimal degrees, the maps align perfectly.


Any help is appreciated


--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 

http://orcid.org/0000-0001-5073-5572
________________
Can’t stop the signal.
_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user



--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 

http://orcid.org/0000-0001-5073-5572
________________
Can’t stop the signal.



--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 

http://orcid.org/0000-0001-5073-5572
________________
Can’t stop the signal.


_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user

-- 
Kirk Schmidt, MScF, BScF, RPF
General Manager
Nortek Resource Solutions Inc.
RR # 1
Thorburn, NS
B0K 1W0
Tel (902) 922.3607
Email: [hidden email]
Web: www.nortekreources.com

_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
Reply | Threaded
Open this post in threaded view
|

Re: Problem georeferencing images in UTM

Nicolas Cadieux
In reply to this post by listsen


Le 18 mai 2018 à 00:28, <[hidden email]> <[hidden email]> a écrit :

I haven't been able to open the dataset but going through the discussion I would like to offer some assistance since I believe there are some misconceptions here:

First, the data has nothing to do with the NAD27 / NAD83 datum's since these are North American datum's and they don't apply I believe in South America. Moreover, the separation between NAD27 to NAD83 is about 100m. I think Carlos mention something like being 400m off.

Another point of confusion is about the projection of Google Earth. Google is using a Web Mercator projection based on a perfect sphere. This has nothing to do with the UTM that has already incorporate the Earth Flattening values into the Northing and Easting equations.

If I find some time tomorrow, I will come back and give the proper equations showing the difference between Google's projection and UTM.

-0x906


On 2018-05-17 19:53, Carlos Henrique Grohmann de Carvalho wrote:

Thanks for the input Nicolas
 
These are official maps so the utm grid is ok. I figured that I was using a linear transformation, with the option to just create the world file, which works for control points at the corners (lat/long), since the boundaries of the map is defined by latitude/longitude. With UTM, this simple transformation won't work, but TPS will give the correct results.
 
Carlos
 

On Thu, May 17, 2018 at 7:28 PM, Nicolas Cadieux <[hidden email]> wrote:
 
Hi,
The utm grid is way off or we have the wrong epsg code.  If you geofence using 2 points (linear algo)  using a service like google earth, you will see the distortion.  The long lat coordinate are probably ok but I did not test them.
 
You can either use the long lat or use more points.  With 4 points and using the thin plate spline, I got good results.
Nicolas

Le 17 mai 2018 à 12:42, Carlos Henrique Grohmann de Carvalho <[hidden email]> a écrit :

Thanks Nicolas, but the misalignment still persists. OTF is off, project has same CRS as data (epsg 29192). There is a ~400m difference in the N-S direction.
 
Carlos

On Thu, May 17, 2018 at 1:28 PM, Nicolas Cadieux <[hidden email]> wrote:
 
Hi,
Disable projection on the fly.  Make sure the project is in the same CRS as your maps.
Nicolas

Le 17 mai 2018 à 12:14, Carlos Henrique Grohmann de Carvalho <[hidden email]> a écrit :

Hello all
 
I have two pngs of topographic maps from southeastern Brazil (a zip file with images and points is available here: https://www.dropbox.com/s/mexeh4be46pgsk6/images_georef.zip?dl=0)
 
I'm trying to georeference them, but with partial success only. In this image (https://www.dropbox.com/s/37xxqp4fkwpdyqh/utm_error.png?dl=0) you can see that the X coordinate is fine, but the Y coordinate of the two maps won't match.
 
Both maps use South American Datum 1969. I selected points near the edges, and used the UTM values. 
 
Interestingly, if I use Lat/Long, and place the points at the corners, using decimal degrees, the maps align perfectly.
 
 
Any help is appreciated

 
--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 
 
http://orcid.org/0000-0001-5073-5572
________________
Can't stop the signal.
_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user


 
--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 
 
http://orcid.org/0000-0001-5073-5572
________________
Can't stop the signal.


 
--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 
 
http://orcid.org/0000-0001-5073-5572
________________
Can't stop the signal.

_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user



_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
Reply | Threaded
Open this post in threaded view
|

Re: Problem georeferencing images in UTM

Nicolas Cadieux

Hi,

NAD27 does touch the northern part of South America but you are correct in saying that it does not extend in Brazil.  I wrote the comments when I was still trying to figure out where the maps were in the world.  I often get problems with NAD27 vs NAD83 grid in Canada as many maps have been printed with the errors in NAD27 military grid reference system.  In those cases, error in much bigger than the standard 30 to 100 metres error so that was my working hypotheses.   People assume that adjoining maps use the same CRS and that is often not the case in remote areas.

I was reprojecting the Google map image to the local CRS in order to figure out the distortion and not using the  WGS 84 Pseudo Merc coordinates. It's important, like to say to know that there are differences between the web merc and the various  WGS 84 realizations.  I would like to see the numbers out of curiosity if you have them. I often wonder why these projections were invented.

BTW, the problem was solved when we used the TPS transformation and not the linear transformation. 

Cheers,

Nicolas

        


On 2018-05-18 3:52 PM, Nicolas Cadieux wrote:


Le 18 mai 2018 à 00:28, <[hidden email]> <[hidden email]> a écrit :

I haven't been able to open the dataset but going through the discussion I would like to offer some assistance since I believe there are some misconceptions here:

First, the data has nothing to do with the NAD27 / NAD83 datum's since these are North American datum's and they don't apply I believe in South America. Moreover, the separation between NAD27 to NAD83 is about 100m. I think Carlos mention something like being 400m off.

Another point of confusion is about the projection of Google Earth. Google is using a Web Mercator projection based on a perfect sphere. This has nothing to do with the UTM that has already incorporate the Earth Flattening values into the Northing and Easting equations.

If I find some time tomorrow, I will come back and give the proper equations showing the difference between Google's projection and UTM.

-0x906


On 2018-05-17 19:53, Carlos Henrique Grohmann de Carvalho wrote:

Thanks for the input Nicolas
 
These are official maps so the utm grid is ok. I figured that I was using a linear transformation, with the option to just create the world file, which works for control points at the corners (lat/long), since the boundaries of the map is defined by latitude/longitude. With UTM, this simple transformation won't work, but TPS will give the correct results.
 
Carlos
 

On Thu, May 17, 2018 at 7:28 PM, Nicolas Cadieux <[hidden email]> wrote:
 
Hi,
The utm grid is way off or we have the wrong epsg code.  If you geofence using 2 points (linear algo)  using a service like google earth, you will see the distortion.  The long lat coordinate are probably ok but I did not test them.
 
You can either use the long lat or use more points.  With 4 points and using the thin plate spline, I got good results.
Nicolas

Le 17 mai 2018 à 12:42, Carlos Henrique Grohmann de Carvalho <[hidden email]> a écrit :

Thanks Nicolas, but the misalignment still persists. OTF is off, project has same CRS as data (epsg 29192). There is a ~400m difference in the N-S direction.
 
Carlos

On Thu, May 17, 2018 at 1:28 PM, Nicolas Cadieux <[hidden email]> wrote:
 
Hi,
Disable projection on the fly.  Make sure the project is in the same CRS as your maps.
Nicolas

Le 17 mai 2018 à 12:14, Carlos Henrique Grohmann de Carvalho <[hidden email]> a écrit :

Hello all
 
I have two pngs of topographic maps from southeastern Brazil (a zip file with images and points is available here: https://www.dropbox.com/s/mexeh4be46pgsk6/images_georef.zip?dl=0)
 
I'm trying to georeference them, but with partial success only. In this image (https://www.dropbox.com/s/37xxqp4fkwpdyqh/utm_error.png?dl=0) you can see that the X coordinate is fine, but the Y coordinate of the two maps won't match.
 
Both maps use South American Datum 1969. I selected points near the edges, and used the UTM values. 
 
Interestingly, if I use Lat/Long, and place the points at the corners, using decimal degrees, the maps align perfectly.
 
 
Any help is appreciated

 
--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 
 
http://orcid.org/0000-0001-5073-5572
________________
Can't stop the signal.
_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user


 
--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 
 
http://orcid.org/0000-0001-5073-5572
________________
Can't stop the signal.


 
--
Prof. Carlos Henrique Grohmann
Institute of Energy and Environment - Univ. of São Paulo, Brazil
- Digital Terrain Analysis | GIS | Remote Sensing - 
 
http://orcid.org/0000-0001-5073-5572
________________
Can't stop the signal.

_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user




_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user


_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
Reply | Threaded
Open this post in threaded view
|

Re: Problem georeferencing images in UTM

Andre Joost
Am 18.05.18 um 22:30 schrieb Nicolas Cadieux:
> Hi,
>
> NAD27 does touch the northern part of South America but you are correct
> in saying that it does not extend in Brazil.

Note that +datum=NAD27 redirects to the US/Canadian grid shift files
which do not even cover Central America.

See
https://gis.stackexchange.com/questions/124322/reprojecting-between-nad27-and-wgs-84-part-iii 
for a visualization.

Greetings,
Andre Joost

_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
Reply | Threaded
Open this post in threaded view
|

Re: Problem georeferencing images in UTM

listsen
The UTM Easting and Northing are the following:

X = a*L ,
Y = a * tan^-1(sin(F)) - a*e * tan^-1(e*sin(F))

For the Web Mercator the same equations are:

X = a*L,
Y = a * tan^-1(sin(F))

What's going on here?  The Web Mercator as a projection is treating the
Earth as being a perfect sphere.  On the other hand, the UTM is making
use of the Earth's flattening value "e".  As you can see at the Web
Mercator Northing equation, the flattening is zero (e=0) and thus, only
the first part of the Northing equation produces a value.

[Sources: "Implementation Practice Web Mercator Map Projection"
http://earth-info.nga.mil/GandG/wgs84/web_mercator 
/(U)%20NGA_SIG_0011_1.0.0_WEBMERC.pdf]




On 5/19/2018 3:32 AM, Andre Joost wrote:

> Am 18.05.18 um 22:30 schrieb Nicolas Cadieux:
>> Hi,
>>
>> NAD27 does touch the northern part of South America but you are correct
>> in saying that it does not extend in Brazil.
>
> Note that +datum=NAD27 redirects to the US/Canadian grid shift files
> which do not even cover Central America.
>
> See
> https://gis.stackexchange.com/questions/124322/reprojecting-between-nad27-and-wgs-84-part-iii 
> for a visualization.
>
> Greetings,
> Andre Joost
>
> _______________________________________________
> Qgis-user mailing list
> [hidden email]
> List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
> Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user

_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user