Opus-RS high residuals

It seems pretty weird that ALL the x and y residuals have the same signs.?ÿ How does it get a final position with all the individual ones on the same side of it?

I agree Bill...

I don't know much (if anything) about how OPUS_RS works, other than it doesn't work very well here in the Great Basin (not many CORS around). BUT, something isn't right in that data set. When you look at the E,N,U values, they are nearly identical (within expected ranges), and I can't explain that.

The Short Term Plots of the 8 CORS are also indicate that they are reasonably well behaved CORS. I dunno, above my pay grade I guess.

Loyal

Hopefully someone from the OPUS team will weigh in here, but it does look very strange that they don't average to close to zero (i.e. all the same sign). Because each CORS vector is weighted differently they won't average to exactly zero, but should be close. Sounds like a bug to me. I rarely use OPUS-RS, but have not seen problems like that.?ÿ

I just submitted a 16 minute file for my pedestal...

I have gotten that message then sent it in a day later or even an hour later and it comes back fine.

No that should be: "I KNOW NOTHINK..."

I would pull data from the nearest CORS site to your occupation site that exactly overlaps your occupation.

So, if your occupation starts at 9:05 am and ends at 10:10 am, trim the CORS data down to the same time frame.

Submit the CORS data to OPUS-RS (you won't need to exclude the site where you got the data from, that happens automatically.)

How does that process??ÿ

I have a theory that you should be able to submit a 2-hour file (less in your case where you are fighting OPUS-RS) from any CORS station and get a decent solution. There are (shall remain nameless) places in the USA where this is not even close to possible.

Also, have you hotwired your OPUS-RS occupation and submitted it as OPUS-STATIC yet. It is always interesting to see how that goes.

Finally, you don't have the elevation mask set really high on your receiver do you?

Mark

"I have a theory that you should be able to submit a 2-hour file (less in your case where you are fighting OPUS-RS) from any CORS station and get a decent solution. There are (shall remain nameless) places in the USA where this is not even close to possible."

That is an excellent suggestion, as I know for a fact that there are many CORS out here that are far from where their datasheet says that they are, especially in the vertical. I am going to do some testing of your theory for highly suspect CORS and see what happens. Thanks!!!

Exclude NYSM and resubmit, then exclude the next worse and see if things are improving.

I too am concerned with all the similar signs.

How long was your observation and what percentage of observations was used in the solution?

What exact date and time period was your observation because "I would like to see the sats used and a sky plot of same.

You are in the same area that I used when beta testing several early versions of RSGPS, the software OPUS-RS uses. I had about sixteen CORS available to download observations from, 14 in NY and at least two in PA. I cannot recall which one I used as my rover but I would purposely skew a majority of CORS either NS or EW for test purposes. I would whittle files down to small increments, adjust atmospheric and other parameters and see what did not work. I could make a solution explode if inclined. I did manage to convince NGS not to go with the extremely short times that thought possible because even with 31 satellites active there is still the possibility of two or three 15 minute periods in a 24 hour day when satellite PDOP is just too high to risk a quality solution. By choosing a data set carefully I could get 100% of rover ambiguities resolved in 3 seconds of data.

I suspect you fell into one of those PDOP voids. Don't believe that 7 or 8 satellites is golden, because I can show you times when 6 or 7 satellites are lined up within a 10?ø band almost perfectly over the rover site and there are zero satellites within two of the four quadrants. If you must use short observations you must make a point of still using planning, it makes it an easy choice to take a long lunch. The other choice is to work through the void and bury it in the middle of an observation. Since I am almost always looking for good vertical my absolute minimum is 1/2 hour and typically longer. Truth be told most of my OPUS-RS solutions these days are breakdowns of OPUS length observation.

Another possibility is that you had some serious weather changes during your observation. OPUS-RS uses a single adjustment factor throughout an observation, if it changes much, it can affect your solution.

Lastly I see a significant loss in active NY and PA DOT CORS stations and your problem could be lack of sufficient maintenance or indifference.

Paul in PA

Posted by: Paul in PA

Some excellent suggestions for figuring out bad results, but none of those explain a final position outside the region bounded by the individual vectors.?ÿ That's just plain wrong.

Not a CORS issue as far as I can see.?ÿ Downloaded, trimmed 2 CORS rinex files to exactly match mine, uploaded to OPUS-RS and they both came back with residuals that I would expect to see.?ÿ Edited my rinex file to end 2 hrs after beginning to spoof OPUS;residuals are better, but doesn't agree for poopoo with other stations and terrestrial measurements.

Not sure what my problem is, multipath @ site was no worse than usual.

Any chance the receiver moved?

Or a bird came and sat on it, or a truck drove next to it and parked for a while?

Not likely a truck, could have been a bird - I was out of sight most of the time.  Setup was still plumb & on the point when I ended the observation

Splitting up the file may reveal the problem. It may have been messed with.

I do know I've identified the time during the observation the instrument was pushed out of level by some cattle that way, of course it was tilted when I got back to the base so I knew it had happened. At least I knew which points to go back and fix.?ÿ

The Canadian PPP service gives you a plot of position versus time, which would show when and if it moved.?ÿ I forget the length of session needed.

Give us an idea of your setup. BTW, was there a chain link fence nearby? A chain link fence can act like a giant dipole antenna, the wire pattern very closely matches the typical patch antenna used within those plastic spheroids you buy. Even with an antenna above the fence top, the signal scattering is tremendous. Most antennas really don't have a decent ground plane. In a test I have gotten usable GPS observations with an antenna mounted upside down.

Paul in PA

OK, thanks for the input!?ÿ Nothing out of the ordinary (for upstate NY) about the setup.?ÿ No Ch-Lk fence, etc..?ÿ Chalking it up to WTF and moving on.?ÿ Too busy, and need to spend free-time with family, to spend unbillable hours on this.

take care,

SS

Hotwired your.... and submitted as an opus-static... what’s this and is there a link to more info?