Anyone want to comment on how to read the residual chart? What does it mean? How do you recognize bad data? What is considered too high? Is there a “good” range we should be looking for?
I've mostly been looking at the signal to noise ratio graph to look for loss of lock, low values, etc. ... and that seems to work fine for identifying bad SVs, and repairing vectors, but I'm
curious about the residual graph, too.
Thanks,
Greg
Greg,
Good question. From what I've been able to determine in four (nearly all nighters) heafty sessions, the residual chart looks like a really good clean up tool. I have turned off the SV's with the high residuals, reprecessed that individual vector, readjusted the data, and seen some improvements in final error estimates. But, like Brian suggested in the thread below, I think we'd better be real careful in turning off too many sv's. I believe the final data error estimates would clue one in on that scenario though, I'm just not sure. Another thing I have done, looking at the individual sv's in the vector residual chart, is isolate short 3 to 5 minute segments of an individual sv showing high residuals, and remove that segment of time from the processing solution. Another trick that seems to work. In the vector residual chart, I believe the only sv's shown are the sv's common to both units on that particular vector. By looking at the sv geometry for the time period of that particular vector, I think you can pick the best of the "best" to use in the processing. I would suppose that eliminating a particular sv just because of high residuals, could result in a weaker solution if it happens that particular sv is the one that holds the sky geometry together during that session. I've really been challenged this week (other than gravitationally, folically and otherwise).
J.D.
p.s. the small bits of time I trimmed were when residuals were peaking at or above 0.2. I think the sessions I ran Thursday may have been affected by a "mid range" solar disturbance. Used long sessions, during periods of 7 to 10 available sv's, and had to clean up quite a bit. I'm hoping it was the solar disturbance that led to the high vector residuals anyway.
Modified By J.D. Billings on 8/19/2001 at 1:57 AM