BTW, base9 was never the Chief Geodesist but Chief "Geodetic Surveyor." Big difference.
GeeOddMike - I stand corrected. I knew that too. Still, I give great validity to what he shares with us here and elsewhere.
spledeus, post: 424678, member: 3579 wrote: Find the document that states the OPUS solution Ortho Height is VIVD09. Note the geoid issue and that it has little to no relevance as your clients' house is more than the 95%error above the bfe.
I agree that the ellipsoid- geoid separation is irrelevant given the poster's statement that the property at issue is "...100 feet.." above water level. I am surprised at the uncertainty in GEOID12B in this area. I would not use heights from OPUS for any work given this level of uncertainty.
The absolute uncertainty in GEOID12B would be a problem, but wouldn't the geoid slope be pretty accurate? So if he used the data sheet on the reference tidal station and OPUS at his location, he could calculate a more accurate height above the tidal station.
All FEMA maps I know of are based on area NAVD88 data, this of course is irrespective of OPUS which gives an estimation of NAVD88. I never have and would not give out an OPUS number for a FEMA map and therein probably is where FEMA's issue is with the number.
If you are saying all bench marks/physical data monuments on St. John are destroyed, that is a real problem to get on VIVD. The OPUS report doesn't give much confidence for the ortho height; are you sure there isn't any vertical local control available? Usually they are shown on the FIRM map or on FIRM maps nearby. It wouldn't make any difference where they are on St John, the island is small enough that a couple anywhere on the island would apply.
Dear Larry Best,
Perhaps I can help. I'm Dennis Milbert, Ph.D. I retired as the NGS
Chief Geodesist in 2004. My predecessor was Bill Strange, and my
successor was Dr. Dru Smith.
The GEOID12B model in the region of the Virgin Islands provides for
direct conversion between NAD83(2011)ellipsoid heights and VIVD09
orthometric heights. This is documented in the GEOID12B "readme"
file, g2012Brme.txt:
https://www.ngs.noaa.gov/GEOID/GEOID12B/g2012Brme.txt
Quoting:
"The GEOID12B model is a hybrid geoid model, as it is modified from a
gravimetric model to fit GPS ellipsoid heights on leveled bench marks.
The GEOID12B model refers to a GRS80 type ellipsoid, centered at the
NAD83 origin. It supports direct conversion between NAD83 GPS
ellipsoidal heights and NAVD88, GUVD04, NMVD03, PRVD02, VIVD09, and
ASVD03 orthometric heights."
The OPUS output you provided expresses both NAD83(2011) and IGS08
ellipsoidal heights. Thus, the datum conversion to VIVD09 would
simply be a geoid height correction:
Datum conversion: H = h-N
+h: NAD_83(2011)(EPOCH:2010.0000) ellipsoid height = (+) -41.198(m)
-N: GEOID12B geoid height = (-) -42.243(m)
---------------------------------------------------------------------
=H: VIVD09 orthometric height = + 1.045(m)
The good news is that your 2-hr OPUS used 95% of its obs and returns
a peak-to-peak of 2.2 cm. The bad news is that the NGS estimate of
geoid error at that location 3.798 meters (95%). I find the geoid
error estimate a little breathtaking. But, I admit I never computed
a hybrid model for the PR/VI region, either. The magnitude of the
error estimate there takes away the utility one normally expects
from a hybrid geoid model.
If you require more than 4 meters accuracy for VIVD09 ortho, then I
recommend you find a VIVD benchmark somewhere that is GPS-able.
Perform a GPS tie between the benchmark and your point of interest.
Then apply a differential datum conversion:
delta H = delta h - delta N
-or-
H2 = (h2-h1) - (N2-N1) + H1 <---- (the VIVD09 bench)
In essence, you are applying the geoid slope against the ellipsoidal
height slope to get an orthometric height slope.
This is how we did it with NAVD88 and GEOID90 back in the day before
hybrid-type geoid models. Even with GEOID96, we still had
gravity-only models for Alaska and PR/VI, and needed to use
differential methods in those regions.
Hope this helps.
Finding a bench mark on a island where the tide is only about a foot and the hills are steep really isn't a problem if +/- 1 foot is good enough. It's documenting it for FEMA that's the problem. And the bench marks that GeeOddMike referenced above are in a cluster, they are there, they agree with OPUS and with the ocean. It's the single horizontal GPS control point on the island that's destroyed. Funny thing is that the 2 previous LOMA's I did was before I had GPS. So I ran trig levels from the ocean, stated so on the forms and they passed just fine. So maybe it's the GPS that has messed me up so bad. Or maybe it's the roll of the dice with the FEMA agent processing the application.
Dennis Milbert's post just came in. Thank you very much. When I get to talk with FEMA again I'm confident I can get this sorted out, maybe by referencing this thread on Surveyor Connect.
There's no way there's a 3.789 error here. The bench mark is just a few feet from the ocean. And I have compared my OPUS results on the opposite side of the island with the ocean also. I have some idea of Geoid, Ellipsoidal and Orthometric terms, but how much difference can there be on a 10 mile long island?
Since you have GPS and bench marks, hold the bench marks, apply the geiod model to the ELEVATIONS on the bench mark, locate the house and forget OPUS for the vertical, since FEMA is basing their BFE on VIVD then it's the basis for their maps; if you can locate a bench, apply the model and check into another bench or two that's the procedure to use, throw away all that OPUS stuff. It's only for a rough check and is may or may not be interesting to you to know the differences but it's NOT the official datum for your purposes.
FEMA will be pass your info right through if you sate that you use the data that created the map.
As far as how does the Geoid check, my guess it probably very well if you use the bench marks as a base, it's easy to find out, it takes me about 15 minutes from my office to check into two of them.
I volunteer to man any GPS station on the beach of St John for as long as it takes to solve this issue to everyone's satisfaction.
😎
While I did emphasize the large uncertainty in the GEOID12B value, I did not characterize it as error. In fact, looking at the independently determined VIVD09 height and OPUS-derived NAD83 ellipsoid height (see prior posts) we can compute what the geoid-ellipsoid separation should be.
Using the relationship h - H - N = 0, we rearrange to solve for N yielding h - H = N.
H (VIVD09) from data sheet = 1.079 m
h (NAD83) from OPUS. = -41.198 m
N = -41.198 - 1.079 =. -42.277 m
The output from GEOID12B = -42.243 m.
The difference between the GEOID12B heights, published v computed above) is 0.034 m.
The reported uncertainty is obviously problematic but should not be considered equivalent to the error in the reported value.
As the possibiliity exists that GEOID12B included a GPS on BM value at the point during the creation of the hybrid model, I also took a look at the gravimetric model USGG2012 (shown above). Using the IGS08 ellipsoid height from the OPUS solution and the USGG value we obtain:
IGS08 h = -43.077
USGG N = -44.479
Using h - N = H, H = 1.402 m.
What does this mean? Like others, I recommend differential GPS from local bench marks rather than OPUS solutions for height determinations in your error. While the information above does show the formal errors of the GEOID12B are likely inflated, the reported uncertainty is a problem.
Cheers,
>> "...how much difference can there be on a 10 mile long island?"
A good question. To explore this I extracted GEOID12B into
a sub-grid enclosing St. John:
LAT min= 18.2833333333333 max= 18.3833333333333
LON min= 295.166666666667 max= 295.350000000000
nrows = 7 ncols = 12 (84 grid points)
which is a 6' x 11' rectangle (a small region for geoid work).
min= -42.62940 m max= -42.14720 m
ave= -42.2589641298567 m
std= 9.585922449693816E-002 m (9.6 cm -- one sigma)
We see peak-to-peak variation of half a meter, and an RMS
about the mean of 9.6 cm. Note, this is the full geoid model
variation. One can expect the geoid model error to be much less.
I can not defend the NGS error estimate of 3.798 meters (95%).
There is not that much geoid "signal" in the region.
For what it's worth, I'm sure Dr. Dan Roman used least-squares
collocation with a multi-model covariance function to establish
the error estimates. But, error propagation can be very tricky
in inhomogenous domains, such as the 3 distinct islands comprising
the VIVD09 datum. To be sure, the GEOID12B error estimates here
need some TLC.
Recognizing a bad error estimate does not mean that the GEOID12B
model in the area is itself necessarily good. We need more evidence.
The post above by GeeOddMike finding a mere 3.4 cm difference between
a GPS derived ortho and a leveled VIVD09 ortho gives some relief.
Ideally, we would like a number of truth points (GPS on benches)
scattered around St. Johns -- which essentially re-invents an
error prediction grid.
Differential GPS from local bench marks is easier .
All the best,
Larry Best - Regrettably FEMA and a number of other federal agencies often show heights in the island areas of the U.S. as either NGVD 29 or NAVD 88. They know enough about geodetic datums to be dangerous - unless the some great deity is your party chief you can't level from the conterminous U.S. to Puerto Rico and VI (or the islands in the Pacific). The vertical datum in VI is defined by the LMSL heights at the primary bench marks at the long term NOAA tide gauges on each island. Myself and Dr. Dru Smith wrote an article about the VIVD09. If you would like a copy just send me an e-mail (base9geodesy@gmail.com)
The NGS database lists 44 First-Order VIVD09 bench marks on St. John. That being said neither those nor using an OPUS solution are probably in the same vertical datum of the FIRMs. Unless they have been redone since they were originally published by FEMA in 2007 then they are not in VIVD09. I was working with FEMA and the then VI governor to try and get the leveling completed to be used specifically for new mapping - FEMA in their infinite wisdom decided to move forward before any of the VIVD leveling had been done. When I asked them what they used for vertical control since none was published by NGS they provided me their Flood Insurance Study which lists the "bench marks." They are all stations in the NGS database that NGS (nor anyone else) ever leveled to. They obtained the heights from some form of photogrammetry project conducted by the Corps of Engineers. No definition of the origin of heights is provided in the documentation. The study lists 4 marks on St. John:
TV0477 (SOLOMAN) -- 724.3921 ft
TV0471 (DIT) -- 144.0636 ft
TV0441 (LAGOON) -- 5.0619 ft
TV0474 (SUGAR) -- 159.6254 ft
If the heights shown to four decimal places doesn't beg a question or two, maybe the fact that station TV0474 (SUGAR) is an old stone sugar mill fan with no top and can't be leveled to or from won't come as a surprise either - another example of how little FEMA understands geodesy and surveying. Since none of these marks are in the VIVD leveling we have no idea of how the "datum" FEMA defined for the FIRMs does or does not match VIVD09.
base9geodesy, post: 424971, member: 7189 wrote: Larry Best - Regrettably FEMA and a number of other federal agencies often show heights in the island areas of the U.S. as either NGVD 29 or NAVD 88. They know enough about geodetic datums to be dangerous - unless the some great deity is your party chief you can't level from the conterminous U.S. to Puerto Rico and VI (or the islands in the Pacific). The vertical datum in VI is defined by the LMSL heights at the primary bench marks at the long term NOAA tide gauges on each island. Myself and Dr. Dru Smith wrote an article about the VIVD09. If you would like a copy just send me an e-mail (base9geodesy@gmail.com)
The NGS database lists 44 First-Order VIVD09 bench marks on St. John. That being said neither those nor using an OPUS solution are probably in the same vertical datum of the FIRMs. Unless they have been redone since they were originally published by FEMA in 2007 then they are not in VIVD09. I was working with FEMA and the then VI governor to try and get the leveling completed to be used specifically for new mapping - FEMA in their infinite wisdom decided to move forward before any of the VIVD leveling had been done. When I asked them what they used for vertical control since none was published by NGS they provided me their Flood Insurance Study which lists the "bench marks." They are all stations in the NGS database that NGS (nor anyone else) ever leveled to. They obtained the heights from some form of photogrammetry project conducted by the Corps of Engineers. No definition of the origin of heights is provided in the documentation. The study lists 4 marks on St. John:
TV0477 (SOLOMAN) -- 724.3921 ft
TV0471 (DIT) -- 144.0636 ft
TV0441 (LAGOON) -- 5.0619 ft
TV0474 (SUGAR) -- 159.6254 ftIf the heights shown to four decimal places doesn't beg a question or two, maybe the fact that station TV0474 (SUGAR) is an old stone sugar mill fan with no top and can't be leveled to or from won't come as a surprise either - another example of how little FEMA understands geodesy and surveying. Since none of these marks are in the VIVD leveling we have no idea of how the "datum" FEMA defined for the FIRMs does or does not match VIVD09.
WOW !
base9geodesy, post: 424971, member: 7189 wrote: Larry Best - Regrettably FEMA and a number of other federal agencies often show heights in the island areas of the U.S. as either NGVD 29 or NAVD 88. They know enough about geodetic datums to be dangerous - unless the some great deity is your party chief you can't level from the conterminous U.S. to Puerto Rico and VI (or the islands in the Pacific). The vertical datum in VI is defined by the LMSL heights at the primary bench marks at the long term NOAA tide gauges on each island. Myself and Dr. Dru Smith wrote an article about the VIVD09. If you would like a copy just send me an e-mail (base9geodesy@gmail.com)
The NGS database lists 44 First-Order VIVD09 bench marks on St. John. That being said neither those nor using an OPUS solution are probably in the same vertical datum of the FIRMs. Unless they have been redone since they were originally published by FEMA in 2007 then they are not in VIVD09. I was working with FEMA and the then VI governor to try and get the leveling completed to be used specifically for new mapping - FEMA in their infinite wisdom decided to move forward before any of the VIVD leveling had been done. When I asked them what they used for vertical control since none was published by NGS they provided me their Flood Insurance Study which lists the "bench marks." They are all stations in the NGS database that NGS (nor anyone else) ever leveled to. They obtained the heights from some form of photogrammetry project conducted by the Corps of Engineers. No definition of the origin of heights is provided in the documentation. The study lists 4 marks on St. John:
TV0477 (SOLOMAN) -- 724.3921 ft
TV0471 (DIT) -- 144.0636 ft
TV0441 (LAGOON) -- 5.0619 ft
TV0474 (SUGAR) -- 159.6254 ftIf the heights shown to four decimal places doesn't beg a question or two, maybe the fact that station TV0474 (SUGAR) is an old stone sugar mill fan with no top and can't be leveled to or from won't come as a surprise either - another example of how little FEMA understands geodesy and surveying. Since none of these marks are in the VIVD leveling we have no idea of how the "datum" FEMA defined for the FIRMs does or does not match VIVD09.
It looks like Larry has a job in front of him, FEMA is going to want to tie his elevations to the map basis, the map basis is incomplete to put it mildly and OPUS will be only a guess. St. Johns has some high priced development and there must be some agency that has good sea level data. They have expensive docks and ships all around there all the time. I would "get on" that if possible and tie as many of the bench marks as I can and see what I'm working with. I just billed mid five figures to get a Zone A property out of the flood plane and he was very happy to pay it ($700 a month flood insurance makes people happy to pay my bill). If the bench marks are close and your property is clearly way above, then state your basis as the "official" bench's and FEMA should be happy,,,,,good luck.
Use the existing elevation basis, apply the geoid model to it, then tie as much as you can of bench marks and sea level info. Who knows you can become the expert on St. Johns and use your knowledge to get all the FEMA work there. 😉
base9geodesy,
I don't see elevations on the data sheets. What am I missing?
About 5 years ago, I think, A NGS crew spent over a week running precise levels from the VIVD bench marks clustered in Lameshure Bay, up to the top of the island and then down into the town, Cruz Bay. They set lost of disks along the way, but I don't think anything was ever published. Just last week I found a NOS disk at about N18-18-59 W 64-46-31, no stamped ID. The place is lousy with markers, but no good data. I guess it's a popular place to send a crew to waste time, but there's no reason to stick your neck out publishing data.
Mr. Best ,
Go the NGS datasheet retrieval page by county - https://www.ngs.noaa.gov/cgi-bin/ds_county.prl . Under "Pick a State" scroll down and select US Virgin Islands, then select "Get County List." While USVI doesn't have counties it's an easy way to separate the datasheets by the three islands. In the list provided obviously select St John and under "Data Type Desired" select "Any Vertical Control." I suspect you have been selecting "GPS Sites Only." That would bring back the only one of the 40 bench marks on the island that has actually been GPSed to a high level of accuracy, the other 39 bench marks have been leveled only - those are the marks that set by the NGS field crew you referenced. The crew certainly did not waste time - I was the project manager for the VI and PR vertical datums and we had them working 6 days a week while they were in VI and I know they were long working days. We had a saying on Coast Survey/NGS field crews - "You'll like working for the federal government, you only have to work a half day - anything after 12 hours is all yours!" GeeOddMike can attest to the many very long days getting the job done.
Wow!! there is a lot of data available, use it, don't worry about OPUS, get on those bench's set for St Johns, no doubt that's what FEMA is looking for.
It would be most excellent if Mr. Best could determine the VIVD09 height for one or more of the three actual survey marks I posted previously that FEMA used as the foundation for the FIRMs (Station SUGAR is a joke). At least one would give as a good idea of what the datum difference is. I wanted to make those observations when I had the field crew down there but I barely had the funding to finish the leveling they did.
MightyMoe, post: 425124, member: 700 wrote: Wow!! there is a lot of data available, use it, don't worry about OPUS, get on those bench's set for St Johns, no doubt that's what FEMA is looking for.
It would be educational to occupy one or more of those benchmarks with your GPS and then compare the OPUS report to the datasheet.
I apologize for my remarks about publishing the data. I was looking at the interactive map that apparently isn't updated yet. I will look into those 4 old marks.
Mark Mayer, post: 425129, member: 424 wrote: It would be educational to occupy one or more of those benchmarks with your GPS and then compare the OPUS report to the datasheet.
Yes, I agree it would be educational and even something Larry may want to find out, but it's not what the FEMA map is based on, if it can be done without charging a client extra I don't see the harm, or maybe he would devote his own time to it, but that should be completely up to him. No doubt the bench marks supersede data from OPUS.
