I ran a 7 hour session with the 4000sse on a point that has low to moderate multipath. The antenna did not have a ground plane, but was above the height of the chain link fences and 40 ft or so from houses, and some leafless trees are to the east and NW.
I ran it through OPUS-S and then broke it into 7 sessions of 1 hour each for OPUS-RS. I plotted the points and approximate 95% confidence limits using the covariance matrix values and "Vertical Accuracy" number from the extended report. I didn't work out how to plot ellipses, so the limits are a crude approximation to the shape.
The OPUS-RS 1-hour results were pretty good horizontally, all within about a half-centimeter of the average and a cm of the long session, although their average is off a bit from the long session. Vertically all are within 4 cm of the long session. But the 95% confidence limits obviously don't take into account some of the effects because too many of the limits, either horizontal or vertical, don't include what appears to be the best answers.
The VDOP does not seem to correlate with the outliers. The reported VDOP from OPUS-RS ranged from 1.88 to 3.03, and the two height outliers had values right in the middle.
I did some comparisons of runs last year using the older 4000sst receiver, but it won't do OPUS-RS, and the OPUS-S results weren't this good. The "prior estimate" horizontal point value was the average of several long sessions on this point using the older receiver.
I did 3-hour run on a HARN station (I know, should have been 4 hrs). The rapid orbit results look encouraging, but we'll have to wait another week for the precise orbit answers to compare with the data sheet. Among other things, that will serve as a check on my measure-up to the ARP.
How does this data stack up against your experience?
Graphical is nice, but generally the numbers are enough to understand what you have.
Take a look at the OPUS-RS extended data CORS stations variances, quite often there is a station or two to exclude and refile. Also check that the same stations were used for the whole series, most often not.
You should have up to nine stations commonly used for the OPUS-RS series, so include them in some additional OPUS solutions. Compare them on a CORS map and use 3 each, 3 times that form triangles around your site. Recently I have had some projects where my first OPUS solutions came back tight because I was surrounded by the three nearest CORS.
I have often been able to select 9 more CORS and get additional OPUS-RS solutions.
More often than not one or two useable OPUS CORS stations would not be used for OPUS-RS, since OPUS-RS is very fussy about clean data. The additional observables used in OPUS-RS being not up to spec.
The precise orbit may move you a few mms , but will never improve the CORS station or you observations data. However given time it is possible that some filtering was done following a checkup by those in the know with some CORS station data. I sometimes see burps in the automatic data stream to the CORS that is later resent to NGS.
A measure up error will never be found by GPS observations, however a measure up blunder will.
Paul in PA
I think I prefer OPUS-RS.