The NGS Gravity Slope Verification Study GSVS14 line of exhaustively measured monuments went within a few miles of some older monuments that had both GPS and leveled data.?ÿ I was curious how well they fit, and can't explain the differences I see.
I computed GPS_EllipHt - Geoid12B - leveled_NAVD88 for a couple of those older points, and as expected the differences are small since they are only errors of fit from the development of Geoid12B.?ÿ Then I computed the same measure for the GSVS14 points within a very few miles of the older ones. ?ÿ
I also computed the formula using xGeoid17B+constant.?ÿ There is a big difference between geoids because they don't intend to represent the same thing, but I just wanted to look at the consistency with a chosen geoid.
The differences in this measure between various points (NOT geoids) can be attributed?ÿ(I think) to a) inaccuracy in GPS measurement of El Ht, b) movement of the old monuments since leveled, c) very slight inaccuracies of the geoid model slope over those few miles. The absolute accuracy of the geoid model is irrelevant to the comparison.
I expected to see differences on the order of a couple cm, which I do find in the eastern end of the line, but got 4 and 5 cm differences at these points further west.
|
PID |
Desig |
GPS_ellip -12B -lev, mm |
GPS_ellip -17B -lev +const, mm |
|
NL0104 |
U 88 (CBN) |
+2 |
41 |
|
DP4523 |
GSVS 069 |
-30 |
17 |
|
DP4522 |
GSVS 068 |
-31 |
9 |
|
DP4521 |
GSVS 067 |
-39 |
1 |
|
DP4520 |
GSVS 066 |
-39 |
0 |
Another one:
|
MH0406 |
U 33 (CBN) |
+10 |
69 |
|
DP4558 |
GSVS 104 |
-45 |
15 |
|
DP4557 |
GSVS 103 |
-46 |
14 |
I'm assuming the GSVS work was as near perfect as humanly possible, so does that mean the old monuments are that far off??ÿ
What am I missing?
?ÿ
These are results I will generally see by using raw Ellipsoid heights with a geoid model applied from CORS/OPUS and comparing them to NAVD88.
As you can see except for the CBN points there is a uniform shift from GPS to Bench Mark you have 16mm in the data using 12B and 17mm using 17B. This is typical and to be expected. What is interesting is how close the 17B numbers are using raw data.
Of course this data is a good example why I've never used OPUS or CORS for elevation data, I always use elevation control applied the geoid to the elevation to generate the Ellipsoid height, by doing that with this data you will be within 8mm for the Bench Marks, that's very good for GPS.
I'm not sure what you mean by questioning the accuracy of the Bench Marks, I would hold them before GPS numbers, unless a new level?ÿrun is done between them?ÿthere will be no answer for that question.
GPS is great to find disturbed Bench Marks, you will know something is up when you see one .5' or 1' out of position.
I would hold the Marks over GPS numbers. Whenever we have run levels between marks they will almost always work, there are some we could tell from GPS that were disturbed but your numbers sure don't reflect that.
I'm not comparing GPS derived versus leveled.?ÿ I'm comparing the GPS derived of two mark and leveled of those marks for consistent differences.?ÿ However good or bad the geoid model is overall, I don't care for this exercise.
If both GPS measurements are good, both leveling operations are good, and the geoid SLOPE over a few miles is good, my differences should be small.?ÿ That is true for a couple marks further east.?ÿ
Note that I added a constant to the 17b numbers so they weren't big, as it is IGS not NAD.
If I can find some time, I'll take a look at the gravity part, and see what interesting part it is playing in the process.?ÿ The 17b numbers were derived from the Colorado area( i i read the procedures and processes correctly) for the maximum difficulties interpolating the model due to the regional uplift and different rock densities.?ÿ Not sure if I will help, or just go slightly more crazy.?ÿ Whenever I use GPS, I routinely accept that it will never be any more accurate than +/- 1cm unless occupation times are ridiculously long, and a gravity tide adjustment is applied too.?ÿ Damn geophysicists ruined nearly everything for me........
?ÿGSVS11 is in Texas, GSVS14 in Iowa, and GSVS17 in Colorado.?ÿ I think (not certain) the GSVS data was just used to check xGeoid17b?ÿ but not to constrain it.?ÿ?ÿ
It's not easy to translate between Geoid12b and xGeoid17b due to the difference between IGS and NAD83.?ÿ I let NGS handle that, and just extract their numbers.
This is a bit old be quite informative:
?ÿ
Hey Bill,
?ÿ
I have a suspicion that to find the numbers you're looking for,?ÿ you'll need to apply a tidal adjustment to squeeze in the final accuracy you seek. Tides and the time domain play out in the smaller numbers with regard to the precise measurement at all locations.?ÿ I'm still looking for a couple good references that might help corral this and give some additional insight.?ÿ I aspire to have time to chase these things like you're doing someday.
Here's an interesting article to go with my blathering:
?ÿ
?ÿ
?ÿ
I was once told that OPUS takes earth tide effects into account.?ÿ I don't have an immediate reference for that.?ÿ I'm a long way from the oceans, so any ocean tide effects would be the same for any CORS that would be used here so should be taken care of.
I was once told that OPUS takes tidal effects into account.?ÿ I don't have an immediate reference for that.
They Should be, and I'm pinging my geophysicists to get more and better information regarding this, specifically the GSVS part.?ÿ We never used the Bleeding edge of the reference frames because they always show a shift from all the previous data and results.?ÿ As expected. Slow to be accepted though.