In my area of surveying practice, perhaps the most frequent misunderstanding that I encounter involves the proper application of NGS produced hybrid geoid height models, or even what a hybrid geoid height model is.
Because I was so frequently asked about this, I decided to prepare a paper on the topic that I could use to respond, rather than reply individually each time; sort of a FAQ document. But, before I went to that effort, I did a little legwork, and as is often the case, the NGS already had me covered.
My research led me to an article in the fall 2014 issue of the Minnesota Surveyor written by David Zenk, NGS Northern Plains Regional Geodetic Advisor. The original article can be accessed here, starting on page 16.
I extracted the article into a nice clean copy, which I have now distributed hundreds of times. It has proven very useful in explaining what a hybrid geoid height model is, and why specific versions thereof are paired with specific realizations of NAD83.
It has also caused some panic when surveyors realize that they have inappropriately applied an incorrect geoid model with a specific realization of NAD83 on previous projects and now wonder about how to “fix” the error. Unless the project is still active and in an early enough phase, and depending on the risks, I usually direct them to make sure that what they did was well documented so that it could be duplicated or corrected in the future, if needed.
Although this article was published when GEOID12A was the current version available, the math and science behind it applies to GEOID12B as well. Great, the world is all good – when working in the NAD83(2011) realization, GEOID12B is the only correct choice. Until….
For the first time that I am aware of, we now have two NGS produced hybrid geoid height models that are both derived from the same realization of NAD83 and the orthometric heights derived therefrom are not the same in many areas of the country. How can this be?
Well, it’s all your fault if you submitted observations to the NGS as part of the ongoing GPS on Bench Marks campaign, which was heightened specifically for the development of GEOID18 a few years ago.
As explained in Zenk’s article, a hybrid geoid height model is the mathematical connection between ellipsoid heights and orthometric heights. The mathematical connection exists only at locations that have both leveled NAVD88 heights and GPS/GNSS observed ellipsoid heights, thus generating the geoid height or “separation.” Between these locations, a gravimetric geoid model is used to interpolate the separation between the ellipsoid and geoid. In some areas of the country, either because of the lack of leveled bench marks, or the sparseness of GPS observations submitted to the NGS, the gravimetric model must interpolate across significant distances and areas. Not the best solution, but it was the best we had, and will have, until the new GRAV-D based geoid is produced for the NSRS Modernization project in a few years.
Because of the great success of the GPS on Bench Marks campaign, enhanced by the excellent interactive tools provided by the NGS to target high-value locations for observations, GEOID18 is a significantly improved version of GEOID12B. Some misfits were discovered, but for the most part, the realized improvements are from having many more points where the geoid height has been measured, thus serving as constraints to pin the gravimetric model to, providing a more refined mathematical relationship between NAD83 ellipsoid heights and the NAVD88 datum. Improvements in the gravimetric geoid model also contributed by improving the interpolation accuracy.
Okay, got it. When working in the NAD83(2011) realization, GEOID18 is the only correct choice. Oops, didn’t I already say that?
It seems that the tendency has been to use the “latest and greatest” geoid model, without considering the consequences. My recommendation has been – if you established control and derived orthometric heights for a project using GEOID12B, and that project is still active, DO NOT switch to GEOID18 now. In the cases I have been involved with, doing so has induced a two- to three-centimeter geoid bias in orthometric heights between points computed using 12B and 18. You would see that in certain observations (though may not understand it), but you likely would never detect it doing topographic work.
But, I don’t just expect peers to accept my word. They shouldn’t, at least not without some quantification. That is when I direct them to the NGS interactive geoid height calculators for the respective geoid models. The calculator for GEOID18 can be accessed here.
The calculator for 12B can be accessed here.
Previous versions can be found through links on the GEOID home page here.
To view a map based depiction of the differences between GEOID 18 and 12B, and the marks that were included or discarded, check this out.
These calculators are computing N in the equation H=h-N. H, the orthometric height, is fixed to the NAVD88 datum. The ellipsoid height, h, is fixed to the ellipsoid. We measure N when we do GPS on BM observations. These calculators compute the N for where these direct measurements have not been performed or not possible to be collected.
I know, the best way to access the NAVD88 datum is to incorporate leveled bench marks into your control as vertical constraints. However, there are large regions of California, and elsewhere where those leveled bench marks simply do not exist or are unreliable because of vertical instability. In those areas, the only way to access the vertical datum is via the coupling of a hybrid geoid model with GPS/GNSS measured ellipsoid heights.
All you ever wanted to know about GEOID18 and more is here.
In summary, GEOID18 is the best available tool, when working in NAD83(2011), for doing that. Unless it isn’t.