All things constant. Good sky, geometry, etc...
How accurately is OPUS-RS going to report elevations. A couple tenths, a foot?
Just a general question.
Just some of the factors.
The Sat geometry is just a part of the equation.
Length of observation.
And HOW the geoid was modeled all factor in.
You can have a PRECISE elevation, with a poor geoid model, and still get better results with a local 3rd order bench mark.
My point is that with GPS, there are more factors, than with conventional leveling. And, each one is important.
Nate
look in your opus report. pay attention to the rms entries for ellipsoidal and ortho height quality. in the case below the normalized rms is o.364m and the height is ellipsoid 0.066m and ortho is 0.068m. for a detailed description of normalized rms, see some ngs powerpoints of "opus". i believe it is a polar peak to peak value. see also quality indicator and observations used definitions. you will be good to go after that
NGS OPUS-RS SOLUTION REPORT
========================
USER: DATE: June 14, 2010
RINEX FILE: ccv-060p.08o TIME: 17:39:25 UTC
SOFTWARE: rsgps 1.35 RS40.prl 1.60 START: 2008/02/29 15:23:40
EPHEMERIS: igs14685.eph [precise] STOP: 2008/02/29 16:21:44
NAV FILE: brdc0600.08n OBS USED: 2088 / 5085 : 41%
ANT NAME: LEIAX1202GG NONE QUALITY IND. 24.23/ 26.52
ARP HEIGHT: 1.2035 NORMALIZED RMS: 0.364
REF FRAME: NAD_83(CORS96)(EPOCH:2002.0000) ITRF00 (EPOCH:2008.16301)
X: 1109632.665(m) 0.024(m) 1109631.951(m) 0.024(m)
Y: -4836476.868(m) 0.055(m) -4836475.419(m) 0.055(m)
Z: 3993860.101(m) 0.035(m) 3993859.985(m) 0.035(m)
LAT: 39 1 3.20031 0.014(m) 39 1 3.22948 0.014(m)
E LON: 282 55 18.26541 0.015(m) 282 55 18.24995 0.015(m)
W LON: 77 4 41.73459 0.015(m) 77 4 41.75005 0.015(m)
EL HGT: 45.503(m) 0.066(m) 44.209(m) 0.066(m)
ORTHO HGT: 77.341(m) 0.068(m) [NAVD88 (Computed using GEOID09)]
> All things constant. Good sky, geometry, etc...
>
> How accurately is OPUS-RS going to report elevations. A couple tenths, a foot?
>
> Just a general question.
also;
http://www.ngs.noaa.gov/CORS/Articles/SchwarzetalGPSSOL09.pdf
Nate had some pretty good comments, which I agree with.
I played around with OPUS_RS a few years ago, and again last year. Based on my cursory tests, I would say +/- 0.5 Feet under good conditions, BUT all of the factors that Nate alluded to above can (and will) push that UP or DOWN (it depends).
I suspect that OPUS_RS works MUCH better back East where there are MANY more CORS Sites, AND a [somewhat] less complex (bumpy) GEOID (Model). This GEOID issue is of course NOT a limitation unique to to OPUS_RS.
Out here in the Rocky Mountain area (and any other mountainous area), the GEOID Models are somewhat limited in the “accuracy” sense. There are obvious reasons for this condition, and when GRAV_D is completed in our area, I'll bet that things will tidy up quite a bit.
I have also noticed that the OPUS_RS Ellipsoid Heights are kinda twitchy in the areas that I have experimented in. I believe that this is primarily due to the “spotty” (less than ideal) CORS coverage out here. This is something that IS specific to OPUS_RS, and NOT something that I see in the OPUS_Static solutions.
All that said...I have NOT rerun my experiments since the Multiyear Adjustment was published, and that should help some (maybe a lot).
If you are interested, give me a call, and I'll outline my Redneck Science Experiment procedures, so you can try them in your neck of the woods. It's pretty simple, but it does take a little time.
(307-789-0291)
Loyal
Most of my 2-5 hour OPUS solutions are followed up with an OPUS-RS solution on the same point. My results are generally between 0.3' to 1.0' difference in the elevation.
While I've only used it a couple of times, I never do less than 4 hours processed through regular OPUS (or whatever it's named now) for elevations, paying attention to the peak-to-peak varations and stats/RMS. While that's not always practical for everyone to do, I certainly wouldn't trust or use OPUS-RS for any elevations.
...Redneck Science Experiment procedures...😉 That comment made my day!
In attending the Oregon GPS Users group meeting, Mark Armstrong-Oregon's NGS advisor, intimated that big changes are coming to OPUS this spring. Stay tuned.
Yep, my understanding is that OPUS-RS is going away AND the main OPUS engine will be replaced, there will be only one OPUS that operates differently and from my understanding MUCH better.
SHG
I have run it on 1st order NGS vert monuments to check its reliability in my area... runs .1'-.2'
I would suggest testing it in your area on some known points.
Jud's post (that just disappeared) brings up a good point, not just for OPUS, BUT for any GNSS derived orthometric heights using a geoid model, how is the relative accuracy vs. the absolute accuracy, the latter is almost entirely a function of how well the geoid model fits BM's in a given area, the relative accuracy of the geoid model may be fine once you apply an offset to the geoid model for your project area.
SHG
Here In The East I Am Consistently Under 0.020 Meters
Usually well under. For good elevations I go to 2 hours or a little beyond, but split the file, even if under 2 hours. I know enough about the OPUS-RS methodolgy to trust 2 1 hour positions over 1 2 hour position. I get very good OPUS-S positions and throw it away for the split file OPUS-RS positions.
I am very spoiled with 9-12 CORS within 200 Km of anywhere I go.
Paul in PA
More Than Just The East Coast
From looking at the CORS map, OPUS-RS should be good in every state East of the Mississippi River except, Maine, New Hampshire, Illinois, Wisconsin, a small part of Virginia and West Vitginia, some of South Carolina and more of Georgia. Plus good in Minnesota, Missouri, some of Oklahoma and 2/3 of Texas.
Paul in PA
Here In The East I Am Consistently Under 0.020 Meters
Paul, I have trimmed some data using note pad before to get under the 2 hour mark for Rapid static. Are you splitting an original file into 2 files?
Yes Using A Text Editor, Usually WordPad
Since it is easier than filtering it.
I may remove the first few minutes if noisy or till there are at least 6 clean satellites, then split a long file into 2 or sometimes 3 equal parts. Because I set up a project alone, I may have up to an hour till my last receiver is running. Sometimes I break down in reverse order, or may break down #2 to the end, leaving my primary for the longest time. If I got busy doing reconn or actual survey work, that may be longer than I required. My primary is most likely to be a 2 or 3 file OPUS-RS submission. My minimum on even the smallest project is 3 points.
I submit the longer file to OPUS-S then the smaller ones to OPUS-RS. Because I get an OPUS-S solution quickly I can insert my initial control positions into my post processing software to solve my other L-1 only positions. Sometimes the OPUS-RS may require additional cleaning or I may have to wait for the next day to get the CORS I want. Once I get all my OPUS-RS positions back with rapid orbit solutions I do the minor adjusting of my post process control points and do a final report.
I was using RSGPS, the software in OPUS-RS before the online beta version of OPUS-RS was available. Doing it yourself you can even solve with a 5 minute observation. Because OPUS-RS creates a single atmospheric correction value for a solution, breaking the file up allows different correction values if there are atmospheric changes during the observation. The initial beta OPUS-RS was accepting up to 4 hour files which because of the single correction, solutions were not as good as expected.
Paul in PA
+/- 0.25ft for a 45-60 minute OPUS-RS occupation here close to Seattle.
OPUS-RS elevations are really efficient when you level between them and combine everything in least squares.
I find that in my state the geoid09 model does fit SOME BM's within a few hundredths in a local area (county wide)using a small offset. This is averaging RTN however not OPUS-RS. Same concept as they are both tied to CORS. Those BMs that fit are the usually higher order with good stability. However what it demonstrates more in my opinion is how bad some of the BM's are that have been accepted as published without question for the past 80 years. Some will say no, Geoid 09 is too inaccurate at those marks. I have seen re-leveling disprove that theory often enough to believe the geoid model before the published "absolute" differences. That's how it seems to work out when there's no height mod. I have a question? What returns more truthful ground elevation differences? RTN adjusted to the national CORS or traditional survey methods tied to 3rd order 80 year old bench runs that adjust out 5 hundredths or more a mile between?