At the 95% confidence level no state (or territory) has a 1-2 cm geoid uncertainty with GEOID12B. In general even in states where the model performs at its best NGS still estimates the uncertainty at 3-5 cm level. If you look at the GEOID12B tool https://www.ngs.noaa.gov/cgi-bin/GEOID_STUFF/geoid12B_prompt1.prl and enter any random lat/long the utility will provide the geoid height and the uncertainty estimate at the 95% level. Regrettably NGS does not provide that figure for published points, you have to perform the computation yourself. I would highly recommend relaying on NGS's estimates.
base9geodesy, post: 455351, member: 7189 wrote: At the 95% confidence level no state (or territory) has a 1-2 cm geoid uncertainty with GEOID12B. In general even in states where the model performs at its best NGS still estimates the uncertainty at 3-5 cm level. I.
I'll take your word for it. I was repeating what I thought I was told by an NGS guy who would be the most likely to know. I do know this: When we PP a 6 hour session on 1st order stability A marks holding the broadcast ellipsoidal heights from our RTN I don't recall a time when we were more than 3 cm from published datasheet elev computed using Geoid 12 A or B. More often than not its less than 2cm. If you go to 2nd and 3rd order and less stability marks all bets are off as Bill's data bears out. I'm fairly well convinced just because of what my baby blues are telling me the lower quality marks aren't the best way to access NAVD88 without height mod. leveling. Most of them were established 80 years ago after all.
BTW, we usually hit the 1st order post marks at less than 3 cm as well.
Note that you should not confuse the Ortho Ht +/- value on an OPUS report with the Geoid12B uncertainty reported at the link Dave gives. They are very different animals, and I think the +/- value reported by OPUS is total bunk. We had a https://surveyorconnect.com/community/threads/opus-height-uncertainty.329248/&apos ;">discussion on this forum quite a while back on that subject that didn't resolve the issue in my mind.
https://www.ngs.noaa.gov/OPUS/getDatasheet.jsp?PID=NJ0546&style=modern&apos ;">One of my shares (done for early practice, of no other importance) has Ortho Ht +/- 0.011 m but the Geoid12B uncertainty given at that point is 0.060 m at 95%. Pretty hard to see the relationship.
I'm unsure if you are referencing my #22 post. If you are we never use OPUS to PP static observations. We use the PP ortho height computed using the Geoid to the datasheet ortho height and ellipsoid height if given. Using RTN Rinex files gives us more freedom and better results in the adjustment than CORS does because the RTN stations intervals are smaller than the CORS intervals. NGS sees no wisdom in including all RTN stations and we do.
Update: I have an OPUS (rapid) solution for the height of a stability B mark 3.3 miles from the culvert that measured -44 mm. I take that as evidence that the culvert hadn't settled.
Diff = (Ellip ??Geoid12B) - OrthoLev
[PRE]Diff, mm Stability Type Nearest Geoid12B pt
+1 B Large abutment (Reset) 16 km
+4 C Poured post 60 km
+11 C Poured post 0
-15 C Poured post 44 km
-32 C culvert headwall 43 km
-43 C culvert headwall 44 km
-44 D Medium culvert 38 km
-51 B Depot building 38 km[/PRE]
I feel much better about my data and the stability of the marks now.?ÿ I used the experimental geoid to calculate
Diff17 = Ellip(GPS) - xGeoid17B - OrthLev(NAVD88) - constant
and for some choice of constant all the differences are within a range of +/- 9 mm.
That small difference would reflect (within a constant) the sum of C&GS leveling error + mark settling + GPS error + tilt between geoids.?ÿ
That it comes out with so small a range is nearly a miracle.?ÿ I take it as meaning my measurements confirm the new geoid well within expected accuracy.?ÿ It also means much of the differences in my previous post(s) must be due to Geoid12B modeling error in the area between HARN stations.