Ok JD, Shawn and Deral, for the matter of discussion, let's propose a scenario and cover the differences.

Scenario: I've only got one L1 receiver and the L1 only version of Ashtech Solutions. I want a "repeatable" 5-7cm position. How am I going to acheive this and how long is it going to take, given that my nearest CORS is 125 miles away from my "unknown" position?

A) Does the geometry of the CORS used matter?

B)How many CORS sites should I use?

C) Should I include the CORS to CORS vectors in the adjustment in solving for my "unknown" position?

ok, GO!

Dangerous Dave
Modified By Dave Huff on 1/1/2005 at 10:29 PM

1. begging your buddy to borrow his receivers. Lots of them

2. prayer. Lots of it

Do you mind my $0.02?

Given your resources:

A) I am old fashioned in my approach to surveying and GPS work. My answer is a qualified YES. The qualification is that you have adequate CORS coverage to make a box and work inside it. Alternatively use only one CORS and repeat vectors to gain confidence.

B) At least one. Is ten too many? OPUS uses three. That is a good number because the control points are regularly verified.

C) NO. control point to control point vectors add nothing to the solution. That has been done for you. The real question is, can I compute coordinates for those other CORS by going through my GPS receiver. If you can make that work within 0.05 to 0.07 meters you have met your stated objective and then some.

I can do a GPS project hundreds of miles long with two ProMark2 receivers. All I have to do is break it into 12 mile long pieces and I am right in the box where I am supposed to be. Triangles are better, IMHO, than traverse lines but I will not debate good results regardless of your procedures.

If you get a chance read about the transcontinental traverse. That same concept of point pairs applied to a GPS traverse would be a very good method. Instead of triangles or a traverse it would be like a ladder. Sharon and I did ladders several times on corridor surveys.

One of my own single frequency projects was 139 miles long and surveyed in chunks to set pairs of points along a corridor.

I will let J.D. and Deral outline the ORGI procedures that had them processing vectors from Indian Territory to The Big Thicket. Some of the tips I gave them to start with included:

Work at night.

Think hours of data and not minutes.

A fast recording interval does not help.

Do not use a ten degree elevation mask. Something higher but I don't know how much higher. Use what works.

End the session with plenty of SV's

Work straight through any PDOP spikes.

Pick good spots. Obstructions will take the whole thing apart.

Mind the space weather but remember that it is probably less accurate than your local weather forcast. It is more likely to tell you why your vectors failed. I think J.D. and Deral know more about Space Weather than I ever will at this point. They can count electrons just by sniffing the air.

Use the data cleaning tools to get the best results. My bet is that the ORGI crew has worn the section of the manual called Post Processing Data Analysis smooth and shiny.

Be prepared to do it all again when it does fail. Don't whine about how much money you are spending. If you want to spend less time getting even better results using OPUS buy a ZMax.

Mike's idea about getting your buddy to help is not a bad one. Randy and I put eight of our GPS receivers in the field on one project by getting surveyors we knew to bring a truck with appropriate accessories and their cell phone. One of them told me later that he did not know it was possible to sit on your behind and make a survey.

A). Yes, it did seem to matter when squishing the error to the middle. At least during our experimentation.

B).Minimum of three seemed to be a key point, depending on geometry.

C). I see no reason to do this. As Phil points out, the CORS are already doing this for us.

The main reason for night was to allow the iono to settle down over a larger region. I think with the new models, then this will be solved for us.

Deral -(only semi-dangerous), unless your blonde and female.

Only one L1 receiver, now that is very hypothetical.

The positions of CORS is fairly inconsequential to L1/L2 solution single unit or not as the solution factors out much of the ionospheric variations.

Not so with L1 only. Being surrounded helps to get an average value for the ionospheric condition so that the ionospheric variations mean out.

3 is not necessary if you have 2, 1 in each direction that are in line with prevailing weather patterns.

Longer observations mean out a lot of variations.

And last but very important, work at night, if possible.

As per Phil, hold 1 and check against others. If 2 checks are not more than twice your positional error budget you are good. If they exceed it, then hold 3 or more and squish to the middle.

Squish to the middle can stay within your positional error budget, but can throw your vertical error budget into debt.

I am assuming you are holding local control for vertical anyway.

As I understand it, having only 1 L1 receiver can be readily overcome with sufficient beer and chili.

How long is it going to take? Budget enough time to cook up a good pot of hot chili, consume same with friends and then put out the fire.

8 hours if working alone. 4 each on 2 different days or overnight.

Paul in PA

never mind

it was a very successful experiment, now a long time ago (in the technological sense)

it is very reproducable, both horizontal and vertical

come to Stumpwater and I will demonstrate the complete procedure, show you archived data sets of both L1 only "mega-vector solutions" backed up by L1/L2 OPUS or HARN ties

Like Phil mentioned, it is completely doable, IF within your time budget and level of patience

chili, gumbo and beer were never part of the experiment either. just consumed in mass quantities when re-living bygone days

Time is what we all have.

How good were the verticals? No numbers neccessary.

When I get to Stumpwater, data is the last thing I want to see.

Nobody had given D.D. a hard time.

Paul in PA

I'll send coordinates for Stumpwater.

And for the record.... the Dangerous One is quite capable of giving himself a hard time. It's not so much that he talks to himself, or even the fact that he argues with himself....it's that he has a tendency to want to spike a whopper antenna or something when he claims to have won one of those arguements

jd

Of course I left it hanging for those of you who might tend to pick on your brothers.

Paul in PA

Well, this thread went nowhere. However, if you check the statistics you can read all about it as one of the "most popular threads" entitiled "Phil, about Cors".

And for the record--- I'm sorry I brought the issue to light.

(Just) Dave
Modified By Dave Huff on 1/4/2005 at 2:03 AM

Dave

I think we are outnumbered by folks who don't really care. You're right, there was several months of detailed discussions, including a whole lot of folks with a whole lot of smarts (not placing myself into that category)involved. A ton of the current questions of "how many, how far, for how long, etc." were answered by practical experimentation. I think we safely decided it did not matter what color of unit was used for the L1 only project, but for some reason Ashtech Solutions seemed to be the very best at handling the L1 mega-vector projects.

For the record I still think for the long vector L1/CORS projects that the station geometry does matter. I thought this in the beginning as part of the reduction of the ionospheric delay problem. To maintain proper geometry I always used a minimum of 4, preferably 5 well placed CORS. Leaving a L1 unit on site over night was never a problem for the projects we needed to ORGI. And, as for the question of whether leaving the CORS to CORS vectors in the adjustment: I may be waffling a bit. Originally it did seem to matter, and I thought the software may be using the constraint coordinates and observed vectors in the LS adjustment to help weight the computations for the unknown station. In every case I've tried with/without there's only been a marginal difference. So, it probably does not matter. Masking was done for each vector to obtain the best results possible. We found that one particular CORS had to be masked at 23deg in nearly every case; others 15-18deg.

Our common experience with the ORGI experiments is what brought a bunch of together and created some strong relationships. This forum, being one that normally encouraged some "out of the box thinking" has taught me a tremendous amout over the years. It seems we've just hit a spell here where it's more trendy to just dream up questions to talk about, rather than to take the time to attempt to answer the questions with a basic hands on scientific approach.

Let's see what NGS has to offer, when and if they are able to impliment their L1 OPUS.

JD

ps for the record, I think this forum is moving more towards a marketing tool for strictly the newest hardware (as it was probably originally intended anyway). we need to find a gps forum just to discuss techniques

I like the issue myself, (just) Dave. So I'll throw in some answers as I saw them.

A) GEOMETRY - I do think our results show that geometry makes a big difference. I think that this is becuase the processing and adjustment actually (accidentally) model the delays from the point to be positioned and the CORS used. If you aren't surrounded by the CORS, then your point will be outside of your home made model.

B) NUMBER OF SITES - We did ok with three (this seemed to be the critical number necessary for success) and saw slight improvements with adding up to 5.

C) CORS TO CORS VECTORS - This is a real brain puzzler. I will have to disagree with the others who said it would not matter. In theory I agree with them (and at the beginning of our tests thought it would be the case that the CORS - CORS vectors would be unimportant). But in looking at our solutions, we always saw impovement in our positions when we added the CORS - CORS vectors. I suspect that in the least squares adjustment, these vectors are used for weighting out the CORS to point vectors. I really don't know, I can only vouch for the results we've seen.

A few more things. Again night time is the right time as the ionosphere is cooled and becomes more homogenous (sometime watch the TEC on the NGS website). Time on site should be at least 6 hours with 8 being much better. Ashtech Solutions will show vector error estimates in the 1-5 foot range on a lot of vectors (ie process only), but the adjustment will show more realistic errors for your point (ie after adjustment). Process the vectors somewhere between 18-22°. This uses the satellites that are above the "noisy" horizon. Process each vector changing the mask in one degree increments while watching for the best (lowest) vector error estimate.

So what results can you get with this?

Our average position of station POST using three separate sessions of Locus data was:

SPC TXNC 83 (US Survey Foot)
N 6835851.421
E 3100685.746
Ortho (Geoid99) 402.795'
El Ht 316.62'


Our average OPUS derived position from 10 sessions was:

SPC TXNC 83 (US Survey Foot)
N 6835851.447
E 3100685.772
Ortho (Geoid03) 403.008'
El Ht 316.44'

As you can see horizontal and vertical were both very tight. The difference (if memory serves) in the sessions comprising the L1 "ORGI" average did not vary more than a tenth of a foot horizontally and less than two tenths vertically. I would expect that whe NGS unleashes OPIE that it will have a 2-3 day delay (so that the IONO modelling can be done) and a minimum of 4 hours will be required. With that I believe a user will see results that rival OPUS results now. This is speculation on my part, but from what I've seen and read, it all seems very possible.

By the way, the vertical values posted for the "ORGI" average were recomputed using our home made phase center offset for the Locus antenna of 0.142m instead of the published 0.125m.

It is not neccessary to compute L1 vectors from CORS to CORS since the CORS do not move. In general it will not solve well anyway. I believe the reason it does not solve is the significance of ionospheric differences. Do an L1/L2 solution and solving becomes easy, since ionospheric differences are resolved.

Having said that, I would solve L1 CORS to CORS anyway, because it gives me a scale of the differences that are being meaned out in a squish to the middle solution.

I have seen differences in elevations in squish to the middle solutions using CORS that are much closer to the site than ORGI. Maybe a greater distance to the CORS sites has less impact on elevation.

J.D. Geometry matters as I stated above, because it is a significant way to mean out our ionospheric/atmospheric differences.

Paul in PA

:-)