Hey Seth! You’re not the only one. This has been a big discussion in this forum as long as I’ve been here. I felt like I was crazy coming in here talking about ground level-looped elevations and how things are done in surveying and construction in the United States.
“The definition of NAVD 88 uses the Helmert orthometric height, which calculates the location of the geoid (which approximates MSL) from modeled local gravity. The NAVD 88 model is based on then-available measurements, and remains fixed despite later improved geoid models.”
This is straight off the web. Note the two statements “approximates” and “then-available”. The ellipsoid and geoid are mathematical best guesses so when you add that to the previous statements you get something that is not in check with reality. The fact is that in the US almost every construction site has a coordinate system of its own. It may appear to be State Plane coordinates, but very rarely is a one-to-one transformation from GNSS data.
Probably one of the biggest factors in this would be CAD. It’s very easy for things to get slightly shifted or rotated to a point that you wouldn’t even know it unless you brought in a one-to-one transformed piece of GNSS data like Google Earth or a drone orthomosaic. Another thing that messes with surface coordinate systems is the scale factor and the condition of the actual control which in our case is usually property pins and a couple of vertical benchmarks. Property pins go missing and get moved and our vertical benchmarks are traversed in from other benchmarks which they themselves are not even first generation from NAVD.
In construction in particular we do a process called localization which is sometimes called transformation or calibration. Basically uses the monumentation as it exists on the ground and best fits it to the theoretical positions from a plat or CAD file. This instantaneously creates a new coordinate system relative only to that construction site. It’s not uncommon to have horizontal scale factors of 100 ft and vertical adjustments of up to 5 ft based on the difference in orthometric heights versus real ground elevations.
At the end of the day you are working with theoretical information against ground calculated information that is in a constant system of digital manipulation so there are a lot of factors as to why control and coordinate systems between two different parties don’t match. It’s the age-old surveying discussion.