How to tranform to Irish Transverse Mercator, with Malin Head Datum

[harrytwomey]

so i was trying to mimic the same process for irishgrid and i am getting results that are off 50 in elevation, i used the same tiff that paulo so kindly provded or the itm transormation, as there in not change in heights. also getting a small error the the position, but i imaging this is because of the height error.

[Paulo]

The problem is that MalinHead geoid gives geoid height relative to WGS84 or Irenet95 datum (i.e. difference between elipsoidal height (WGS84) and Malin head geoid)...

but the Irish grid is referred to geodetic datum of 1965 which has its referenc elipsoid  shifted relative to WGS84 (delX,delY,delZ[482.5,-130.6,564.6])....

So you would need geoid file giving difference between elipsoidal ht referrred to 1965 datum and geoid....

[harrytwomey]

Code: [Select]

47_EP-11-31756_0022_0107.JPG 621513.954020 666505.171191 208.568732

So I'm double checking this and cheked the same photo
here are mi results, slightly different than yours
 

Code: [Select]

47_EP-11-31756_0022_0107 621513.9718114218 666505.21073729824 208.56900652348543
As i move away from the central meridan thing get worse
In the west of the country the error grows to, 470127.308062 835225.845397 128.810508 from photoscan, 470127.315490,835225.795441,128.810718

Whats interesting is that the standard coordinate systems seem to give the error too
maybe this is a bug?

[harrytwomey]

So you would need geoid file giving difference between elipsoidal ht referrred to 1965 datum and geoid....

Yes i see this now, while trying different combination the heights for the supplied itm and irish grid are different, whereas the gridinquest application give the same heights, so i had come to this idea

So currently the undulation grid paulo provided for ITM corrections is the difference between the IRENET95 geodectic datum/ the Ellipsoid GRS 1980 and the Malin Head Vertical Datum
And to make the same work for Irish Grid i need to make a .gri/.grd file for the difference between the TM65 geodectic Datum/ the Ellipsoid Airy Modified 1849 and the Malin Head vertical Datum

if i understnad correctly?

[Paulo]

That is correct...

For a point P, having orthometric height H (above MSL or Malin Head geoid), the geoid height relative to WGS84 is given by:

N_wgs84 = h_wgs84 - H where h_wgs84 is elipsoidal height relative to WGS84

and the geoid height or undulation relative to TM65 datum is:

N_tm65 = h_tm65 - H where h_tm65 is elipsoidal height relative to TM65 datum

so it follows that:

N_tm65 = N_wgs84 + h_tm65 - h_wgs84

[harrytwomey]

Hello again

I finally had a chance to do this, but i got stuck at the import the geoid.

I created a grid of points with both meta-shape and Gridinquest and subtracted the height values that gives a gird of differences.

put that into global-mapper gives a lovely looking picture but when i export a goetiff or .grd file fails to be accepted by metashape

Im not sure if theres something wrong with the file or the format or both??

[Paulo]

Hi HarryTwomey,

I think you exported your tif as 16 bit integer samples instead of 32 bit float....

See capture from GM export GeoTiff ...

[Paulo]

I was able to create a compound CS for Irish Grid  / TM65 incorporating a Geoid (MSL) tif file,
see attached screen copy and prj...

The tif geoid file can be downloaded from following link https://drive.google.com/open?id=1vgw78FXENHhYiVcZ1Sgd6f7cA1b4rZVx

Hope this can be useful,

PS. seems that geoid file must be referenced in GCS (TM65) and not in PCS (Irish Grid / TM65) in order to be correctly dealt with by MetaShape... as your original grid was referenced in Irish Grid, that was probably the issue...

[Paulo]

Another example with 10 points distributed at 100 km intervals all over Ireland...

[jdulak]

Hi Paulo,

I have a project where client requir a DEM in Irish Grid with a geoid OSGM02.

My project is in WGS84 and whenever I export something to IG with standard agisoft setting then I have a differance between gcp's and DEM heights around 2.7 m.

In turn when I am using your geoid IG65 the diffrance between gcp's and exported points has approx. 15 m.

I do not know what I am doing wrong.

I am using a geoid for ITM created by you and it is working correctly however I have some seriously issues with IG..

Maybe do you have any clue whats it can be wrong?

Regards,
Jakub

[Paulo]

Hi Jakub,

I think this is same problem previously cited:

For a point P, having orthometric height H (above MSL or Malin Head geoid), the geoid height relative to WGS84 is given by:

Nwgs84 = hwgs84 - H where hwgs84 is elipsoidal height relative to WGS84

and the geoid height or undulation relative to TM65 datum is:

Ntm65 = htm65 - H where htm65 is elipsoidal height relative to TM65 datum

so it follows that:

Ntm65 = Nwgs84 + htm65 - hwgs84

The geoid file OSGM02 is referenced to ETRS89 datum with geocentric spheroid GRS80 or WGS84 while Irish grid is referenced to TM65 datum with non geocentric spheroid Airy Modified 1849 with offset to WGS84 defined by 7 parameter similarity  TOWGS84[482.5,-130.6,564.6,-1.042,-0.214,-0.631,8.15]

so you have to correct the OSGM02 geoid values by difference between ellipsoid heights for 2 spheroids as:

Ntm65_IrishGrid = Netrs89_osgm02 + htm65 - hetrs89_wgs84

PS which OSGM02 file are u using? Because the one I downloaded from geoids page coveres only Britain and not Ireland..

[Paulo]

Jakub,

I think you should use Irenet95/ITM + OSGM02 as your target CRS from WGS84 and so avoid using old datums as TM65/Irish grid with inaccurate coordinate transformations...

I downloaded geoid OSGM02_RoI.txt from http://www.isgeoid.polimi.it/Geoid/Europe/Great_Britain/Great_Britain_OSGM02.html and created a Compound RS incorporating Irenet95 / ITM with this geoid...

This should solve your problem... Attached is geoid tif file  created in compressed rar along with prj file...

Hope this can help,

[jdulak]

Paul,

Thank you for your quick respond.

I used this osgm02 geoid to check how it works.

Like I know a difference between osgm02 and osgm15 shouldn't be bigger than approx. 40mm for area which i am working right now and in my example there is a couple of meters.

Please see attached print screens.


Regards,

[Paulo]

Hi Jakub,

i looked at difference between 2 Geoid Models and got following result...

Majority of points have difference  between -0.067 m and 0.067 m...

and in 2nd attachment. the difference in elevation for point placed at ITM origin is less than 3 mm between 2 Geoids (OSGM02 and Malin head or OSGM15)

So if you have a point in particular then i can check the difference...

Looking at your 2 DEMs, it seems that they were not georeferenced with same accuracy.... Maybe a problem with your camera reference locations not having same accuracy between the 2 surveys... Were they done with DJI drone? cause often there is an issue with drone reference altitude...