In [msg=280342]another thread[/msg] Gavin Schrock wrote:
> There is little advantage over very-short baseline RTK but in longer baselines there can be significant advantages. The application of ultra-rapid orbits over broadcast is one of them. Some network even apply tighter clock/orbit products than the default JPL ones.
It seems to me that the biggest error source in single-base RTK outside the user's control is the broadcast ephemeris, so I'm intrigued by the above statement. My understanding of the IGS ultra-rapid orbit is that the first half is computed while the last half is predicted. Is the ultra-rapid prediction based on the computed half, or is it the same as the broadcast orbit?
By The Time You Are In The Office, IGU Is Calculated
The ultra rapid orbit is calculated 4 times per day at: 03, 09, 15 and 21 UTC. Each time the projected is reformulated from the calculated. Typically in the igunnnn_18.sp3 the predicted part is so close to the calculated that you cannot see the difference in your results if you wait 1/2 day for the igr.
Unless I am working very late in the day or into next GPS day, if I have an OPUS control point, I will accept _18 for OPUS and my post processing final product.
Paul in PA
By The Time You Are In The Office, IGU Is Calculated
By the time you're in the office the IGU isn't going to help with RTK, unless the predicted half is based on the computed half. If the predicted half is the same as the broadcast orbit, then there's no advantage to using the IGU in RTK solutions.
Broadcast Orbit Is Estimated For 24 Hours Then Broadcast
The farther from midnight the greater the drift.
If you have RTK software that can download igu and use the predicted orbit closest to the calculated it will indeed improve accuracy.
RTK works off a known fixed point. All RTK software allows you to adjust that fixed base point to a post processed OPUS position and move all the rover coordinates along with the base.
Have I got the gist of your question?
Paul in PA
Broadcast Orbit Is Estimated For 24 Hours Then Broadcast
> If you are talking about the results right-then-and-there then the ultra rapid available to the RTN has a slight advantage over the broadcast that a standalone base-rover is getting from the sats.
This is what I'm talking about, true real-time positioning.
I just looked at the IGS products page, and see that the predicted accuracy of the ultra-rapid is 5 cm, versus 1 m for the broadcast. That's a big difference; I would think it would confer more than a slight advantage. Even if the broadcast errors are randomly distributed, the SV count is so low that I wouldn't think that the errors would reliably cancel out.
Broadcast Orbit Is Estimated For 24 Hours Then Broadcast
Differential processing makes this a lot less important than it seems.
Broadcast Orbit Is Estimated For 24 Hours Then Broadcast
> Differential processing makes this a lot less important than it seems.
I've always assumed that the orbit prediction was the primary source of the 1ppm distance error associated with single-base RTK. Do you attribute it more to the short data sets, even when integers are correctly fixed?
Broadcast Orbit Is Estimated For 24 Hours Then Broadcast
I think you are right about that, Jim. My comment probably wasn't helpful to your thread topic. I am still wrapping my mind around how precision GPS works. Because the processor is determining the difference in cycles between base and rover of a signal from a satellite, the exact location of the satellite is not important. And in fact, the actual distance between receiver and satellite is never actually known (PPP being a possible exception). I believe that just as the old rule of thumb for error induced by error in base position is 1 ppm for every 10 meter error on the ground that there is a 1 ppm error for every 10 meter error in satellite position. But I think this is 1 ppm for the differential range between base and rover for that particular satellite and not necessarily the the final solution itself.
Hi Jim,
The orbit and clock errors for differential RTK are very minimal as long as your baselines are sub-50km.
A rule of thumb for how orbit errors affect differential position is given in UNB's guide to GPS positioning:
Error in baseline = error in orbit * baseline length / distance to satellite
So for a typical user scenario with a 20,000 m baseline, and a 1m broadcast ephemeris error and a range to satellite of 20,000,000 m we get an error in baseline of ~ 1 mm.
The most problematic error source for single baseline RTK outside the user's control is still the troposphere (particularly if you base station and rover are at significantly different elevations) and the ionosphere. Although we can remove the ionosphere using dual frequency receivers, most RTK algorithms don't do this as it makes ambiguity resolution very difficult. Instead the ionosphere is estimated, which can be tricky if the ionosphere is behaving unexpectedly.
One benefit from running with more precise orbit products is that the fault tolerance is also much more stringent.
Hope this helps!
Trah
:good:
> Hope this helps!
Yes, it helps a lot. Thanks!
