7

I've read things all over the internet which claim that NAD83 and WGS84 are effectively equivalent (with minor ellipsoid differences). However, if I do geodetic (WGS84) -> ECEF -> geodetic (NAD83) conversions, using only the WGS84 and GRS80 ellipsoid parameters and no Helmert transformation, I can't replicate the results of government sponsored online converters like this one. My geodetic/ecef conversions are done following EPSG note 7.2 and have been rigorously tested for correctness.

Thus my questions are:

  • Is a Helmert transform required to convert WGS84 to NAD83?
  • Are there canonical or official coefficient values for doing so? I know there are variants of NAD83 and WGS84 over time, but I'm interested in coefficients that would be used today (I think that means NAD83 (2011) and WGS 84 (G1674)?)
6

Short answer is yes. You need to convert from your coordinates' native reference frame to the NAD83 realization's reference frame, then possibly add an adjustment. In the case of NAD83(CSRS), which I use, there are 3 steps:

  1. Native RF -> ITRF96(1997.0)
  2. ITRF96 (1997.0) -> NAD83(CSRS)
  3. Grid shift

Steps 1 and 2 require the Helmert transformation. Step 3 uses interpolated shifts from a raster to apply adjustments due to tectonic motion and isostatic rebound.

NAD83 is intended to be locked to the North American plate so its realization must rotate with respect to the reference frame. In CSRS this accounted for partly by the final reference frame transformation and partly by the grid shift. (It's fairly difficult to find complete information on this online -- when I was writing software to transform point clouds, I had some generous help from the NRCAN scientist who's responsible for it.)

The IERS publishes transformation parameters for going between various ITRF realizations: here.

I got the ones for CSRS from The Evolution of NAD83(CSRS) In Canada, by Mike Craymer (2006). This page has the transformation parameters between 2011 and ITRF. (This one has a spreadsheet with many transformations.)

  • I think I've run into all those sites before, but you answer definitely helps put the pieces together. Do I correctly infer that, for recent WGS84, the 'Native RF -> ITRFXX' is a 0 parameter transform? – Nicolas Holthaus Jan 25 '16 at 17:38
  • Yeah, each recent iteration of WGS84 is supposed to be interchangeable with the contemporary ITRF, but between iterations there's a transform, especially if you're starting with the original Doppler version. Of course, for a lot of applications, these transformations are pretty small and most people ignore them. (In the field I'm working in at the moment, most people have no idea it's even a thing...) – Rob Skelly Jan 25 '16 at 17:52
  • You've got a mostly Canadian-specific answer, but the OP uses NAD83 (2011) which is a US realization. Some of the US transformations come from here. – mkennedy Jan 25 '16 at 18:02
  • @mkennedy -- My link didn't actually have the information I thought it did. Substituted yours. – Rob Skelly Jan 25 '16 at 18:13
  • Actually, it did on the NAD83 tab of the excel spreadsheet--please include yours again (I've added it as a favorite!). – mkennedy Jan 25 '16 at 18:15
1

There is the paper "Transformations between NAD83 (2011) and WGS84 (G1674)" from the US NGS detailing the process of converting NAD83 data to WGS84 data.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.