NAD83(Original) Transformation with Heights

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

NAD83(Original) Transformation with Heights

Rob Skelly
Hello List,

I'm trying to convert some LiDAR data from NAD83(Original), with orthometric heights based on CGVD28, to WGS84 with ellipsoidal heights, and to NAD83(CSRS) with orthometric heights based on CGVD2013.

I've been verifying my output using NRCAN's online GPS-H tool (http://webapp.geod.nrcan.gc.ca/geod/tools-outils/gpsh.php), and not getting good results.

To begin with, I use las2las to convert my data to WGS84, using parameters:

--a_srs '+proj=utm +zone=12N +ellps=GRS80 +geoidgrids=HT2_0.gtx'
--t_srs EPSG:4326

Then I go from WGS84 to NAD83(CSRS) with the following:

--a_srs EPSG:4326
--t_srs +proj=utm +zone=12N +geoidgrid=CGG2013n83.gtx +ellps=GRS80 +datum=NAD83

Now, I don't actually have a decent way of verifying my output, other than with GPS-H, and the disagreement seems significant. For example, the original elevation at 494618.88mE, 6306815.69mN is 469.95m, and I get an ellipsoidal height at -111.088356°W, 56.90501°N of 444.52m. GPS-H claims that the orthometric height at that position (for HT2_0, the closest approximation to CGVD28) should be 470.004m, a difference of 54cm.

I realize there are a lot of points where this process could be going wrong, or that it's not going wrong at all. I wonder if anyone can give me any insight or fill in any obvious holes in my reasoning. If I've left out any pertinent information, let me know.

Note that I could just use GPS-H to process the points, but I need to automate the conversion of a lot of data, and establish a reusable process to do so. And, I'd like to use this opportunity to learn about proj!

Thanks,
Rob




_______________________________________________
Proj mailing list
[hidden email]
http://lists.maptools.org/mailman/listinfo/proj
Reply | Threaded
Open this post in threaded view
|

Re: NAD83(Original) Transformation with Heights

support.mn
Rob Skelly [[hidden email]] kirjoitti:

>
> Hello List,
>
> I'm trying to convert some LiDAR data from NAD83(Original), with
> orthometric heights based on CGVD28, to WGS84 with ellipsoidal heights, and
> to NAD83(CSRS) with orthometric heights based on CGVD2013.
>
> Now, I don't actually have a decent way of verifying my output, other than
> with GPS-H, and the disagreement seems significant. For example, the
> original elevation at 494618.88mE, 6306815.69mN is 469.95m, and I get an
> ellipsoidal height at -111.088356°W, 56.90501°N of 444.52m. GPS-H claims
> that the orthometric height at that position (for HT2_0, the closest
> approximation to CGVD28) should be 470.004m, a difference of 54cm.
>
> I realize there are a lot of points where this process could be going
> wrong, or that it's not going wrong at all. I wonder if anyone can give me
> any insight or fill in any obvious holes in my reasoning. If I've left out
> any pertinent information, let me know.
>

ok... "exactly"...

1) have you verified that the 54 cm is not within the limits of the
difference between HT2_0 and CGVD28?? if that is so then there is
no error

2) another source of errors might be the path you go to that elevation..
make sure the selected path (down and up) does not generate that error?

3) make sure that "las2las" and GPS-H do the same assumptions

4) remember that for later proj.4 versions you have to declare both
source and destination datums to make sure that it calculates
anything

5) how large was the horizontal difference?

6) etc etc ..

regards: Janne.


_______________________________________________
Proj mailing list
[hidden email]
http://lists.maptools.org/mailman/listinfo/proj
Reply | Threaded
Open this post in threaded view
|

Re: NAD83(Original) Transformation with Heights

Rob Skelly
In reply to this post by Rob Skelly
> have you verified that the 54 cm is not within the limits of the difference between HT2_0 and CGVD28?? if that is so then there is no error

Well, it turns out I made a math error: the difference is ~5cm, not 54cm. That's a small error, but what we really want is for the outputs to match.

> another source of errors might be the path you go to that elevation..

The HT2_0 file is the grid shift file, so that's the mechanism proj is using to get from the original orthometric heights to the ellipsoidal heights. The HT2_0 file uses geographic coordinates, so proj must be doing the horizontal transformation first, then the vertical shift -- so if the horizontal coordinates are off, the elevation will be off.

So 5cm is obviously a small error, but it disagrees with GPS-H...

> make sure that "las2las" and GPS-H do the same assumptions

My working hypothesis was that, since GPS-H is using HT2_0 with the NAD83(CSRS) reference frame, and our original data uses NAD83(Original), there should be a +towgs84 argument in my original -> WGS84 transformation, because WGS84 has changed between those two reference frames. However:

> how large was the horizontal difference?

My transformation from the original horizontal coordinates to WGS84 and GPS-H's transformation are almost identical. Even if there is a reference frame difference, the difference in shift values should be negligible.

So... we have just discovered that the grid shift files may be misaligned by 8' to the East. Shifting them by that much provides almost perfect results, at least for our study area.

Thanks for your time!
Rob



_______________________________________________
Proj mailing list
[hidden email]
http://lists.maptools.org/mailman/listinfo/proj