After much consultation, deliberation and consternation........

Considering that OPIE implies reference to ITRF........

The StumpWater R&D Office and the SW La. Crawfish Kitchen, also having been referred to as
"The Joint Texas-Louisana Not Quite OPIE but Damn Close Research and Development Consortium.", have prepared this joint announcement for the continued development of O.R.G.I. - Optimistic Real Geodetic Intersections

We felt further development should not be restricted to a limited geographical area and we invite all those L1 users out there to perform experimentation.

Please begin preparing your equipment for the Nationwide O.R.G.I. Experiment.

Date TBA in the near future.



Thank You,

The former Joint Texas-Louisana Not Quite OPIE but Damn Close Research and Development Consortium.
Modified By James Webb on 12/10/2001 at 10:45 PM

Further instruction to all who wish to participate in this joint venture includes erection of a stable base station in the location of your choosing, preferably a secure location. We would hope that the erection could be completed prior to the month of February so that the initial testing of O.R.G.I. may begin at that time.

Any L1 users (preferably Ashtech, but not limited to) who may wish to participate in this venture make yourselves known.

J.D. Billings, Stumpwater R&D Northeast Texas Division O.R.G.I.

p.s. sure would be nice of Ashtech to volunteer to process and adjust this session when completed. What a deal.

Preparing the stealth receiver at this moment (if I can match the blue Ashtech paint color and if it adequately will cover this bright yellow stuff)...

With RINEX, then we should all be able to contribute data to this most noble cause..

Changed my mind about using the HARN point and will set a point in the back yard, then a couple of OPUS sessions and ties into the HARN network...People sure look at you funny sitting in the lawn chair at 2am with little flashing lights the only sign of activity..

TM

Also curious to find out if the NGS has already done some testing on this facet...They seem only to rely on dual's and totally ignore our L1 only situation.

BTW- I think that the mask raising (instead of the coordinate seeding was causing my bias).. I was trying to figure out some basic difference and all things pointed to the mask we were using. Makes sense when you think of it...With short baselines then the tropo doesn't contribute to much to each line, but using these lengths, then it becomes substantial (using my lower mask)..I think the tropo can be one our our larger error contributors on these increased lengths....

The more I know, the less I know, but at least it makes life interesting..

TM -Chief Scientist at the SW Oklahoma Research Center for ET (Extreme geodetic Ties)

TM,

I worked with each vector, one at a time, and played the mask angle until I saw the lowest possible vector error. Then started with noisy sv elimination and trimming small bits of time. I don't know what you saw, but I can tell you that every vector I processed in conjunction with the Winfield LA CORS had to have a 22-23° mask. This was the case in all three of my once weekly sessions. Every vector I processed required a bit different mask. I am under the impression that the lower mask angle allows some bad data (speaking in terms of L1 only) due to ionospheric conditions. Higher mask to clean up the data reduces the number and time continuity of the available sv's, thus the reason for the lengthy sessions.

Glad you want to play. We really need 8-10 units in the network, with preferably no more than 200 mile spacing if possible. I'll take, or make the time to put together the details of proposal for the project after the negotiations between the Stumpwater R&D and the Crawfish Kitchen are complete.

J.D. Billings, Rookie CEO of Stumpwater R&D, O.R.G.I. Northeast Texas Division

JD,

I just did something really simple....

Just used the 5 vectors from CORS to POST in the adjustment with a 20° mask angle for processing. Excluded all other vectors.

Time set from 0:00:00 08:27:00

POST jd base
East. 3100685.684 0.111 Adjusted
Nrth. 6835851.399 0.100
Elev. 403.384 0.124

And with a bit of tweaking.......

POST
East. 3100685.682 0.048 Adjusted
Nrth. 6835851.421 0.043
Elev. 403.367 0.054

Is there a way to print/report the individual vector adjustments ?, i.e. the tweaks ??

Is there a way to toggle off that notification box about not finding L2 so only L1 will be processed ?

Jimbo
Cajun Kitchen Bug Eater
O.R.G.I.
SW LA.

PS: In 135± miles what is the claimed accuracy of L1/L2 ?? Looking to me like there is about .2' HZ Float in said distance....or about 1 in 3.5 million, depending on who is tweaking.


Modified By James Webb on 12/12/2001 at 10:32 PM

when you get a chance.

Jimbo

seems to be the very best I've seen. First process all at the predetermined 10° mask, just to get it over with. Then start with vector 1 and process at 15°. Watch the vector length error estimate column, as well as solution type, and heep changing the mask for that single vector till you see the best results. You may even see a change from a "partial solution" to a "fixed solution" (nothing in the QA window) but higher value for the vector length error estimate. In this case, I've been going back to the partial solution with the lower error estimate. Every vector needs a different mask to get the best results at O.R.G.I. vector lengths. For example, seems that WNFL (Winfield, LA) needs a 22-23° mask for every vector I've seen, in all three L1/CORS tests. I just figured the antenna was in a semi-multipath location. All other vectors tweaked at anywhere between 15° and 20°. Considering the unwanted iono disturbance, I figure the 15°-20° mask is sufficient. After individual masking of vectors, I used the vector residual chart and isolated noisy sv's for elimination and small segments of high residual time, per the graph. BUT, often you'll cut a really crappy 15 minutes out of the middle of a single sv and cause the solution to sour, or just blow up completely. Gotta have at least a 4 sv continuity over the "long haul" to make it work. Or so it would seem.

Try my session #2 by examination of each individual vector. I'll send it a.s.a.p.

As for the "no L2 notification box", and turning it off, I don't have a clue. Heck, they don't expect us the be using L2 stuff without the lock anyway. Not a problem, just a nuisance huh.

I don't seem to see a 0.2' float in the horizontal between the three sessions. Only thing I saw near that was the spread on the vertical in the 3 independent tests as I processed them. Please explain the ".2 HZ Float".

I wish there was a way to print out the "vector tweaks". I have stacks of scribble notes scattered all over the place for comparison purposes.

J.D.

O.R.G.I. in Stumpwater

Seems to be difference in the various solutions I've run thru doing different things to different vectors.

" (nothing in the QA window) "

You mean no change or it passes ?

I've been seeing mostly FAIL with usually Partial or Float solutions. I think I've seen about 4 vectors go fixed but I haven't done all 15. Tried the alternative (prior post) before I'm going back to vector time tweaking.

2 more things on my AS wish list.

Jimbo
O.R.G.I.
SW LA.

May I suggest opening a notepad file, use cut and paste for some of the stuff from the window at the bottom of the workbook plus your own notes and comments as you go along. It's sort of like a diary of the project.

Even if popping from window to window can get a little confusing it beats rumaging through bits of paper.

what's one more window !!! Reasonable alternative.

I was hoping that something in the report functions might have a way to print out the tweaks in a nice pretty format like the screen windows.

Jimbo

Jimbo,

Every one of these vectors (all 15 in each test) can be cleaned to "partial". What I was alluding to is the fact that you can "overwash" the data to get a fixed solution and actually decrease the validity of the vector results. By looking at the vector length error estimate (right hand column) you can watch it happening. I've had some error estimates double after tweaking for a fixed solution from a partial. It's my understanding the AS2.x just gives us 2 catagories of a "fixed" solution. Partial in some cases being of lower quality, but fixed none the less.

I haven't used a failed vector in any of the 3 test sessions. Had 2 or 3 of the vectors reach what appeared to optimum results with "fixed" solutions. Doesn't seem to matter on the O.R.G.I. vector lengths, and could obviously be that many cases "partial" is better. Shawn just mentioned that Mr. G described some details concerning partial solutions back a few weeks ago. Will need to go back in the "way back machine" and reread that.

A "partial" solution is one in which the s/w did not fix ALL of the biases (even one not fixed constitutes a "partial" solution). There's nothing wrong with this, it is not inherently less accurate than a "fixed" solution (especially on long lines, where even a "float" solution may be the best).
Modified By Brian Ewing, PLS on 12/13/2001 at 10:41 AM

Some Comments:

(1) When you are dealing with long observation periods, say more than 4 hours, whether or not integers are fixed makes only marginal difference in the answers. In fact you are usually better off not to try to fix integers for satellites that have short observation periods because fixing to an incorrect integer will usually cause much greater error than having an unfixed integer.

(2) You mentioned cutting out a segment of observations in the middle of a satellite pass. Unless the software introduces new integers after the segment that was removed you run into the possibility that there is a cycle slip that cannot be seen that exists between the two segments. That is what can cause a solution to blow up.

(3) Something I have done to see where errors were when you have the situation where you actually know the position of both stations is to hold both stations fixed in a vector solution and look at the residuals. I don't know if your software allows that. But if it does it would be interesting to look for problems in the CORS data this way by computing vectors between CORS stations while holding both stations fixed.

When I noted the problems with time trim on "some" mid session segments I think the problem had more to do with the fact that only 4 sv's were within that segment. On these long lines, especially the 125-250 mile lines, there were very lengthy segments of only 4 simultaneous satellites. Admittedly, 4 sv's does not permit redundancy. But, my thinking is that the necessary redundancy is in the "time geometry" created by the 8 hour session. I've tried (on test 1 only) to run the data as a "first 4 hour session" with what I consider to be very poor results. Tried the last 4 hours, with even different results. I suppose I could go back and spend the time (if I had more time) and attempt to tweak the 4 hour sessions. I am beginning to think the "time geometry" with even as few as 4 common sv's is working. According to the software, the only data I see to validate the findings of my final adjusted position is the adjusted error estimates.

Also, I still cling to the theory that using all vectors, including CORS to CORS, in the solution is necessary. My thinking is that with the CORS coordinates being constrained to published values, the software is weighting the data accordingly. The C to C data just gives the least squares more data to adjust by.

I appreciate your input, and remind you I'm still just a novice and shade tree scientist. Damn good thing I didn't get a chemistry set (other than the Stumpwater Brewery)instead of gps equipment.

J.D.

My statements above were not intended to be hard and fast findings. I was/am presenting to you, for your assessment, my line of thinking. Consider every statement above to actually be a question, in particular the idea that using all vectors in the fully constrained adjustment for determination of the one unknown point could strengthen the least squares adjustment (for the unknown) as a whole.

There I go again, putting a question in the form of a statement. (?)

Thanks

J.D.


o.k. I'll use the "dependent vector" phrase. I've been dancing around it so as not to get off on the "independent vector only" debate again. My thought regarding this long base line experiment is that the so called independent vectors help strengthen the least squares solution of the single unknown.

Modified By J.D. Billings on 12/15/2001 at 12:23 PM