Kent McMillan wrote:

While WGS84 and ITRFxx are essentially identical, I believe that ITRFxx is actually the reference frame that the international fiducial stations are positioned in.

Trimble Man wrote:

I agree...The SV's have been refrenced to the ITRF for quite a while...It's not a major difference from WGS, but it's a difference none the less...

Brian Ewing wrote:

You owe me lunch. You are correct that the fiducials are published in ITRF, but I'm correct that the broadcast orbits (ephemerides) are referenced to WGS84.

I'm on the ragged edge of comprehension here (perhaps a bit over it, in fact), but I stumbled upon the following that would seem to add yet another twist -- however minor -- to the matter:

Since December 2, 2001, the National Geodetic Survey and the other Analysis Centers of the International GPS Service (IGS) have been providing GPS satellite orbits (ephemerides) that are referred to a new terrestrial reference frame, called IGS 2000 (or IGS00 for short). This new frame was designed to be consistent 'on average' with the International Terrestrial Reference Frame of 2000 (ITRF 2000 or ITRF00). That is, the best fitting Helmert transformation between IGS00 and ITRF00 for a set of 54 well-established, international GPS satellite tracking sites is the identity function, even though the IGS00 position and velocity for any particular site may differ slightly from its corresponding ITRF00 position and velocity. Despite the different designations, users can treat IGS00 and ITRF00 as equivalent. In particular, users can freely mix the use of IGS00 orbits with the use of ITRF00 coordinates for control points.

The foregoing may be found here, a page that also provides transformation parameters between various reference frames.

come to Connecticut.

How's that sound?

Heck, we've even got lots of that white stuff on the ground.

Thanks, Mike, but I've been to Connecticut. There ain't a decent relleno in the whole state.

Jim:

This reference frame business is getting out of hand. Sure enough, since last December 2 to be technically pure (and lord knows we all want that by the bushel, don't we?) one would have to say that the IGS orbital products are expressed in IGS00.

I haven't looked at the residuals between the IGS00 positions and ITRF00 positions of the common stations from which the Helmert transformation was derived, but I'll bet that they are on the order of a few millimeters. So the differences between IGS00 and ITRF00 may be less than the thickness of a tortilla.

Just to be on the safe side, though, why don't we all pay for lunch and let the waitress decide who gets a refund? Those gals at the Paradise Cafe in Kyle, Texas (where Brian will miss carne guisada today if he is out in California training people how to set up tripods) will do the right thing, I'm sure.

Best regards,
Kent McMillan, RPLS Austin TX

Come to "Stumpwater" and run your own equipment on the "POST", and process with anything you want, however you want. Just prove to me that my NAD83 coordinates are wrong. I'll buy lunch just for your attempt.

J.D.

All this discussion of coordinate systems has lead me to come up with my take on the subject. I will get it posted some time this weekend and let let everyone through darts at it.

In the mean time quite talking about good Tex-Mex food. It makes me homesick. Nowhere up north here can I get good chili with no beans and grease floating on top before you stir it.

Mr. G., I look forward to reading your post on this topic.

No more food talk. I'll just go home and cook up a pot of chile verde.

How about making it beer instead of lunch? The millimeters involved in all of these different frames would certainly seem less important after a few DOS XX. Then again, if Brian is invited (which I think he should be) perhaps we had better include some Henessy.

Shawn

two months notice and I'll brew whatever....everyone's invited......

but now we're talking liters rather than meters...or is that mililiters vs. milimeters.....

Shawn,

Someone's been badmouthing me. Before I switched to beer, my preference was single-malt Scotch whisky (especially Knockando).

J.D.,

Beer doesn't come in milliliters. ;>}

J.D.:

Just to correct the record, I've never written that I doubted your coordinates particularly. I've never written that, that is, unless Jim Frame finds that I have.

My entire difference is one of emphasis. I'm interested in:

(a) improving the efficiency of positioning and processing using CORS data and L1 receivers, and

(b) being able to fairly reliably estimate what will be possible under different conditions.

To me, efficiency means fewest number of CORS records downloaded and processed per point, shortest occupations, and most reliable uncertainty estimates generated in processing.

I haven't used precise orbits much in my normal land surveying work, but it is quite plain to me now that using the IGS Rapid Orbits significantly improves the chances of getting a fixed integer solution on long L1 vectors. I'd have posted more about the CORS-to-CORS experiments here, but it is strictly yellowbox type equipment and software and I'm not sure how welcomed it would have been by the sponsor of this message board. If there are no objections, I'd rather post here on the long L1 vector threads.

I didn't specifically mention this fact on the main board, but a couple of the 170+km vectors from the data of 1/09/2002 were solved (and with remarkably good looking statistics) from only 40 minutes of data (!). The noise in the residuals had an average RMS of about 8mm (!!!).

Here is an interesting question, "Does a mid latitude K index value of 0 always mean that 170km baselines will be a snap or do other factors such as tropospheric noise come into play?"

Sign me up for a Stumpwater Ale,
Kent McMillan, RPLS Austin TX
Modified By Kent McMillan on 1/12/2002 at 2:22 AM

Since I missed lunch with Kent at the Paradise Cafe on Friday, I would like to sign up for the Stumpwater Lunch Party. I would be happy to drink Stumpwater Pale Ale, but if not enough is brewed up, make mine a Negro Modelo.

Sitting here in Limbo...waiting for the wind to blow......Tom Bryant PLS North of Texas, MO

JD's "little 'speriment" (at one time) has blossomed and flowered much more than anyone could possible have hoped for.

Let's sum up so far;

L1 can definately produce consistent and repeatable results over long lines (given the constraint's of iono considerations)

We haven't yet determined (except through higher masks) a better way to defeat tropo (which seems to me is our largest error at this point..

We are working through the whole relationship of NAD-83/ISO/ITRF, ect...and the effects on actual point accuracy...

We've determined that precise ephemerides are a very great improvement in the statistics over long baselines.and likely station positioning...

Time is a factor, but how much is enough is yet to be proved...Not enough sets yet.

If I missed anything, then tag it on...

What I wonder....Hasn't the NGS done all this testing long before to determine the validity of OPUS? Maybe they never considered the L1 option? I wonder sometimes..I sometimes resort to "what the manufactures tell me" and wonder if the NGS has also done this at times, instead of robust testing..Not dissing NGS , but just wonder if we are breaking new ground, or just proving what they already know....

Maybe they(NGS) could toss out some ideas to pursue in this endeavor? Maybe they can use this data we are producing to actually produce OPIE...

That said, the metadata has been great and we've been able to repeat with different software/hardware basically the same results...

So much for stellar results..Maybe it's time to pursue what a normal survey would be with L1 (6 hour during the normal working day)...and see if this all blows up or that we can work through 'less than optimum' data sets....

TM

PS- The Stumpwater Brew shipment was accosted at the border and I am having to rely on oil cans (Fosters) at this time....

On the subject of long (say, >150km) L1 GPS vectors observed during ordinary working hours, I'm going to bet that the only days when they are feasible are dull overcast, cold winter days with K=O and exceptionally low GEOS electron fluxes.

If you can get an acceptable solution on a CORS-to-CORS vector >150km using only L1 observables in daylight, I'll be amazed. The nighttime is the right time, or so I'm willing to wager.

I've been using the Oklahoma and Kansas CORS points, but it might be interesting to compare lines of similar lengths and orientations at different elevations to try to see how the wet fraction of the atmosphere figures into the tropospheric effects in practice.

There is a West Texas CORS point, MDO1, that is up about 7,000 ft. or so in the high Chihuahuan Desert. Baselines from it to a station in New Mexico might be interesting to try.

Best regards,
Kent McMillan, RPLS Austin TX

Kent,

Just for the record, I didn't mean to sound like I was "throwing down the gauntlet". ....Well, not entirely anyway. I'm still just trying to make the point that I don't believe you would be able to see any appreciable difference in the results of my field work by using ITRF or any other system. I do still see the necessity of controlling our dynamic positions.

As for your latest challenge, I think you're right. Day time stuff, L1 only, would be so unpredictable it would not be worth the trouble. Just my opinion from personal limited test sessions. Just take any of your own control point sessions where you've left a unit on a control point while roving all day. Six hours, eight hours, whatever. Pull down the CORS data for the day. Not impressive for sure.

As for MDO1, I'd love to go put my hand on that one. That's the typical yankee vision of what Texas landscape is all about.

Also, I want to make sure you guys still want to run units for a group session of O.R.G.I. when everyone has a secure site for an "all nighter". I'm not sure what we're trying to prove with this one, but it's certain to prove interesting.

J.D.


p.s. Kent, I hope you do get the ITRF stuff figured out. Y'all have some software capabilities beyond some of us here. Remember, we (L1 blue guys)are restricted to the broadcast ephemeris, no precise, unless Jimbo can figure a way around that. And he may be able to.

J.D.:

Processing just the L1 observables in the CORS records for 1/09/2002 for:

MDO1 (MCDONALD VLBI STA CORS)
PIE1 (PIETOWN CORS)

Using Precise Ephemeris (IGS Rapid Orbits)

18deg Elevation Mask

SVs 3 14 15 17 18 & 21

An hour of 30 second data between
6:00 - 7:00 UTC gave a fixed integer solution of acceptable quality on the 557km vector between the two stations.

The "miss" on PIE1's ITRF97(1997.0) position was :
N -0.022m
E +0.002
U -0.065

I realize that the relative ITRF97(1997.0) positions are slightly in error due to crustal deformation, but was simply amazed to see a 2cm check on a 557km L1 baseline.

It means that I can darn near do positioning in Austin using the McDonald Observatory CORS records (if I do a real careful job of session planning, get lucky on the space weather, and don't get shot by poachers as I'm setting up the GPS receiver in the pasture at midnight).

Best regards,
Kent McMillan, RPLS Austin TX
Modified By Kent McMillan on 1/13/2002 at 10:57 AM

557KM...Outstanding...

The wet refraction can be somewhat negated by a higher elevation mask...., but I think that an algorythm can be made to compensate for this regional basis...Using each SV declination (for each site)and azimuth, then an matrix can be built to 'model' the respective tropo index and this can be applied on an SV by SV basis (which would be applied to each position)...

As to setting up in a pasture at night...Watch out for those 'cow patties'...They tend to stick to your boots and make the drive back somewhat less than desirable...(and please don't eat them...they leave you with terrible breath)

TM

PS- I tried your 'very simple' ITFR demonstration that you posted, but didn't see any appreciable differences...Maybe I'm not looking in the right place?

TM:

I'm going to try again later this week to provide a numerical example that shows the problem inherent in using IGS00 vectors (we are now technically pure) with NAD83 control over long baselines. Basically, what it does is make the uncertainty difficult to estimate in the adjusted results. It overestimates it since the residuals are artificially inflated.

The problem is easily solved just by rotating the raw vectors into NAD83 before adjustment. I've been meaning to write a quickie utility to do that, but haven't had the time. So in the mean time, I've been doing the adjustments in ITRF97.

Right now I'm crunching away toward a deadline and am playing hookie just to post this.

Would like fries with that?

Best regards,
Kent McMillan, RPLS Austin TX

Chances are that you're getting much more significant errors from multipath and iono (even with a low K index) than you are from tropo. We do use a tropo model, and my experience has indicated that it has little effect except where large differences in barometric pressure exist between stations, and in such cases, the model actually appears to introduce error instead of reducing it.