NGS releases Technical Report and Web Map for GEOID18
My first reading of the TR reveals it contains a good level of detail about how data, especially GPS on BM submissions, is evaluated. Not much on the improvements in geoid modeling theory; I guess that is reserved for journal submissions. The TR also discusses the deflection of the vertical model improvements.
Lots of good detail down to the state level. Geoid18 does not include Alaska or Pacific Islands but does include Puerto Rico and the USVI.
Direct link to TR:
Well worth reading IMO. There is even discussion of “jackknifing” as an analysis technique.
I was especially interested in the discussion of MG0388 in Figures 26 and 27, as that mark had my submissions. I sure wish someone else had done a share session on it, too.
As the report notes, it is a long way to any others submitted. I had checked Google Earth or in person for every disk over a 50 mile span and found only three existing and tolerably good for GPS. The next one south is almost certainly destroyed, the one after that is overgrown, and the two after that apparently destroyed. Any marks along an abandoned railroad remaining undisturbed tend to be on culverts or bridges over streams and now overgrown with trees. Going north, the next one is destroyed (my brother lost his picture to prove it out of place and now it's gone so we can't get it marked as destroyed). The one after that is vertically mounted so not eligible for GPSonBM, and the next one is not found. So they continue.
I was suspicious of my result because the comparison (GPS_Ellip - geoid - NAVD880) was significantly different from those I had done in other areas. Worrying that the old culvert had somehow moved, I also checked MG0386 the vertically mounted one and got a result between the two on MG0388. MG0388 doesn't have the best sky visibility, and my two sessions on it differed by 31 mm, not great repeatability, so I probably should have done a 3rd session, but didn't get around to it as it was further from home than most of my sessions.