How do you know which one to use or ban from being THE "reference satellite"? Any takers on this one?
Sorry JD, I just had to! (actually we discussed this over the weekend).

The "reference satellite" is used in double differencing when vector processing is done.

Usually, the highest elevation satellite is chosen automatically to be the reference satellite, since it will almost always have lower measurement noise Vs the lower elevation satellites.

I can see 2 situations whereby you would want to ban a particular satellite from being the reference:


(1) A high-elevation satellite happens to be partially blocked by trees or other obstructions. Any noise in this reference satellite's measurements will also be present in the double differences to the other satellites. Solutions usually switches automatically to another reference satellite if one is noisy, but on rare occasions the switch might not happen automatically.

(2) A very unlikely scenario--
a high elevation satellite was unhealthy during data collection, and was not flagged as unhealthy in the broadcast ephemeris i.e. Solutions does not know that the satellite was in fact unhealthy based on the Broadcast epehemeris information


Regards,

Richard

Thank you Richard for clearing that up for JD.
I learned something also (!)
Best Regards,
Dave

is THE reference sv noted as such, and if so how?

J.D.

p.s. By the way Dave, thanks for letting me help present your question. Don't worry though, I'm sure Jimbo will ask one for you sooner or later.

Modified By J.D. Billings on 8/21/2001 at 4:12 AM

Ok, the reference satellite can be seen in the residual plots for a given vector.

The reference satellite is located in a small box on the left hand side of the residual plot. The plot can be pulled up by right clicking the vector in the vectors tab, and selecting "View Residuals".

The reference SV should be displayed automatically in this residual view Window. You can turn its display off and on by right clicking the background of the residual plot.

Regards,

Richard

As Richard Phelan says one uses reference satellites in performing a double difference solution. One uses double differencing for computing the difference in coordinates between two GPS stations using carrier phase observations. One of the station is designated the reference station and has its coordinates input to the solution. By differencing between simultaneous observations of two satellites at a station one eliminates any error due to the station clock. Then taking these single differences and differencing them between stations (the reference station and the station being differentially positioned relative to it) at the same instant in time eliminates errors due to time differences between the clocks on different satellites.

Software can pick which satellite(s) to use as a reference satellite, or it can allow you to pick. Most surveyors use relatively short data spans and, again as Richard Phelan says, it it is common for the software to pick a single reference satellite that has a high elevation angle and use it throughout the observing session. I am more familiar with long (multi hour) sessions and being able to pick the reference satellite(s) I want to use. With long observing sessions it is not uncommon to use one satellite as the reference satellite for the first part of the observing session and another for the second part.

The choice of reference satellite(s) impacts the final answer in two primary ways. The noise in reference satellite observations impacts all observation double differences used in a solution. Also, using double differencing you do not solve for initial epoch constants for each satellite independently. Rather you solve for differences in constants for different satellites and different stations. The initial epoch constants for the reference satellite enters into all the initial epoch constant differences being solved for. So the choice of which satellite is used as the reference satellite impacts answers in two ways.

It is always interesting to perform different solutions using different reference satellites and see how much the answer changes. Hopefully not much. Presumably the "best" choice for reference satellite(s) will that which leads to the lowest solution residuals and makes fixing integer values the easiest As a practical matter I think most of the time a user will let the software pick a reference satellite and let it go at that. Unless there is some obvious reason to do otherwise ( eg there is a problem fixing integers), it simply takes too much time to too little gain to always be testing every possible permutation.