What are the benefits? How do you implement this data with a LOCUS static
survey? Is this a method of tightening
your network or adjustment?
Differencing means what pertaining to
the precise ephemeris?

Thanks

Todd,

For everyday work tied to HARN or local control, I'd doubt you'd see any great benefit to waiting 1-2 weeks to process data with the precise ephemeris. For projects such as the O.R.G.I. experiment, I do think it would be good.

Problem is, AS 2.xxx can't use the precise ephemeris. I wish it could.

J.D.

First, it appears that AS2.XX can't use the precise without something called Doctor K (about $2000) if the posts about it are correct..

Brian posted on the upcoming release of the precise (inside of AS2.xx?)..but didn't say if they were going to include it as part of the package, or would still require additional funds to purchase it as a module... Here's what he posted..Maybe he'll sound off on how this is going...

Re: Precise ephemeris
Posted By Brian Ewing, PLS on 11/20/2001 at 8:44 AM
Precise ephemeris import will be implemented in a future version of Solutions. I can't give you a promise date, though.


I posted some information both on this side and the General Msg Board on testing that I have done.. Just search for ephemeris and you can catch up pretty quick...

I'll go farther and expound on the precise ephemeris and some misconceptions that I routinely hear..

1).I have dual frequency so I don't need them....Wrong. Wrong..Wrong...The precise is just as important to L1/L2 as it is to L1 only receivers...Dual's cannot remove the orbital error without the precise, just like the L1's...

2). It doesn't matter on short lines...Also not completely accurate...It does affect your statistics (such as RefVar, RMS and Ratio) and while these may not affect your actual station position, it does affect the way your least squares weights the vectors (or observations) in the adjustment.

3). They take too long to become available...True...At least for most surveyors, two weeks is much too long to wait on results...

4)I get an ephemeris when I download from my receiver....Yes, but it's the broadcast and not the precise orbital information...Each SV sends the exact same broadcast ephemeris (so it's based on the modelled system, rather than an SV by SV flight)..It's a pre-computed flight pattern (distance-direction) that's the best guess of where the SV is at any moment.In fact, this is what you use to compute tomorrow's sessions by way of your mission planning software.. I'd prefer that we called it the almanac vs an ephermeris...While not really correct, then at least others won't get confused with broadcast versus precise.
The precise ephemeris is the actual measured path of the SV during it's operation. This can only be done after the flight period has ended, and is done by a very sophisticated system of ground monitoring stations located around the globe. This data is compiled and the precise orbital information then is published in a variety of formats for downloading.

In closing (yes, it had to come at some point)...The almanac or broadcast is a good representation of the system, but cannot 'catch' the odd anomoly in the system..The precise monitors each SV independently and will catch the small error's that may only affect one of the SV's in your constellation.

TM

PS- We're spec'ing our new system to include the Precise...And since I've used GPS (before it actually worked), then you can take this for what it's worth.

And my wish list for AS...A link to automatically locate the precise (if available) from the web while processing data...much as OPUS does on the fly... (beat Trimble to the punch guys!!!!)

I question the assertion that the delay in high accuracy orbits is excessive. For most purposes, I expect that for most purposes (and < 20 KM baselines) the ultra rapid orbits (accurate at about the 25 cm level and posted twice daily) are adequate. Waiting ~17 hours for the rapid obits gets you down to about 5 cm.

See the link below for more details:

http://igscb.jpl.nasa.gov/components/prods.html

I used two weeks as the latency for publishing the precise...This is what the link you posted also shows...or maybe you were questioning if the wait is worth it, as opposed to using the broadcast as it..I agree in short lines that it hasn't shown any positional difference, but it has made some statistical difference, which could (if a large enough anomoly was involved)actually change the position by weighting the solution better..

This thread was started when I embarked on JD's quest to pursue the ultimate L1 baseline..I never used the precise much (except for longer lines and then only with duals)...but now I'm a believer, even on shorter lines...

Again...Thanks to JD and Shawn for a very enlightning journey thus far!

TM

I agree that results are improved with post-fit orbits. My question was whether most users would not see sufficient benefit with the ultra-rapid orbits. The latency of this product does not seem unreasonable to me.

Unfortunately, I am unfamiliar with most of the commercial adjustment packages and cannot address how (or if) they weight baseline solutions on the basis of orbit accuracy.

BTW, NIMA does deliver a precise orbit based on three-days of observations. They currently show an orbit for 2 January 2002 available for download.

ftp://164.214.2.65/pub/gig/pedata

Unlike the versions available elsewhere the NIMA orbits are in the enhanced SP3 format. I do not know whether the commercial packages to convert SP3 file to the EF18BIN format deal correctly with the enhanced format. The NGS utility SP3_EF18 does ...

An extract of the file is shown below:


/* NIMA, ST. LOUIS,MO.
/* EPHEMERIS COMPUTED FROM 3 DAYS OF DATA
/* CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC
/* CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC
* 2002 1 2 0 0 .00000000
P 1 24847.850315 1075.304001 9544.865129 209.806562
V 1 -11056.651365 4859.997350 28698.801861 .017442
P 2 15902.144204 -7959.227264 -18908.597139 -130.227256
V 2 -3106.158757 25600.581844 -13363.938512 -.058455



BTW a good approximation of the effect of orbit inaccuracy on baselines can be calculated by dividing the baseline length by the distance to the satellite then multiplying the result by the estimated orbit inaccuracy.

Running the numbers will show that the error on a 30 km baseline due to a 2.6 m orbit is about 4 mm, a ultra-rapid 0.25 m orbit less than a millimter and the rapid less than a tenth of a millimeter.

Orbits accurate at the 5 centimeter level should also help the baseline processors to determine integer unknowns.

The below applies for normal length baselines...
The post-fit orbits tend to improve statistics, but not actual positions (except the vertical which they do improve quite a bit)...This is what started this quest...Why wouldn't the post-fit affect both the same? I don't have a clue, but it certainly appears to tighten the vertical much more than the horizontal... Probably need a Mr. G to explain the reasons..

I was referring mainly to when you take the processed baseline into the LS adjustment(my package doesn't weight the orbital accuracy per se). The LS would use the more accurate RefVar-Ratio-RMS in the adjustment process. Better (and more accurate)apriori errors would be estimated.
We are really fine tuning at this point and it probably doesn't make a mm on any position...
But...what if it does...

My particular package can import about any of the versions, but I use the enhanced SP3 format. My package doesn't have to do any conversion (unless it's in the background)..

I suppose that this is one of those questions that asks "When is good enough, enough"...

Am I wrong, but are you referring to a single SV(and not total system variation) with your statistical variance of 4mm on the 2.6m orbital computation? With 5-10 SV's in the solution,then what do you think the whole mish-mash can amount to?

TM

and the plot thickens yet again...(and still learning)...Just call me Grasshopper

You are correct that the approximations I provided are for individual SV ranges. I included them to illustrate the negligible difference between the rapid and precise orbits (on short baselines).

BTW, a numeric comparison of the various orbit types (wrt the precise) is available at the University of New Brunswick web page at:

http://gauss.gge.unb.ca/grads/orbit/

I am still learning about GPS as well. It remains a moving target with new signals, improvements in equipment resolution, better models, different approaches to the positioning problem like the "Epoch by Epoch" approach of Bock, et al. See:

http://www.geodetics.com/NewWebSite/home.html

As far as the formal statistics are concerned, they are too optimistic in all the packages I've seen. That is why OPUS uses "peak-to-peak" errors to represent positional accuracy. The improvement resulting from the use of post-fit orbits is probably a by-product of the SV positions having been adjusted to fit together.

Hope this is responsive,

Don