I just downloaded the last 2 of 5 CORS sessions to process with 8 hours of Locus L1 data from 01.333 0700-1500 UT.

With very minimal data cleaning the results look rather promising. The differences so far between this week's session and that of Test #1 last week are all within HUNDRETHS (not tenths) of a foot. And still, session #2 seems to agree with the coordinates of my POST as determined from local control and Harn stations in the area to within the same general tolerance of Text #1. Hundreths of a foot.

I will (given time from the decoratin' the house kinda stuff) spend time on the data cleaning this weekend and post the final numbers for those interested.

I would like to plan test #3 for next week, but due to recent developments on the sun (really nasty equitorial spot pointing at us) it may have to be rescheduled. Don't want to start the solar influenced experiments with OPIE.

J.D., Stumpwater R&D

think I'll have a beer now

still something here. As Kent pointed out below, OPUS takes into account the ITRF positions, velocities, rotations and other data to account for the small movements of the CORS stations over time. I would suppose that my POST is moving on the same plate with the five outer positioned CORS stations used for coordinate determination. My observations having been made one week apart would not show this movement, but as the North American plate is in constant motion, although slight, my POST is moving along with the CORS stations. Given this, even though I started with the lat, lon and ellipsoid height of the CORS stations published as NAD83 "transformed from ITRF (epoch 1997)" I will probably not be where I think I am, relative to ITRF. But.... wherever I am, I'm there twice in a row.

Having spent a bit more time looking at Test #1, trimming data with high residuals, I came up with very slight changes in coordinates from the first "announced position". Test #2 was processed and adjusted with the same care. The following are the coordinates from my original determination of POST by Locus from local points, results of Test 1 and results of Test 2.

Original
N 6835851.505 E 3100685.732 EL 402.979

Test #1
N 6835851.427 E 3100685.739 EL 402.873

Test #2
N 6835851.422 E 3100685.757 EL 402.956


Radial residuals in Test No 1 (adjusted vectors) ranged from 0.109' to 0.450' with the exception of DeQueen AR to Winfield LA - 0.712', and Palestine TX to Winfield LA - 0.543'.

Radial residuals in Test No 2 (adjusted vectors) ranged from 0.064' to 0.381' with the exception of the vector from Houston, TX to DeQueen AR which was 0.678'.

Both sessions were 8 hours to a single Locus unit, at night under very good solar conditions.

What have I learned? Looks repeatable to me. I'll find out again later this week if the sunspots cooperate.

J.D.

J.D.
I like the way you speak so we(I) can understand.
I can relate to being the same place twice. Can you briefly explain the high residual trimming process for new users? Do you think the .08 diff. in N is an ITRF thingy or more of the "pull" created by the control network?

sorry, double post
Modified By Norm Miller on 12/5/2001 at 1:52 PM

Norm,

By "trimming", I think J.D. is referring to eliminating short periods of data that returned high residuals in the first processing run.

Regards,
Brian

...what Brian said re. the trimming. I look at the residuals charts for suspect data and in the vector process routine, under the advanced heading, trim segments of observation time. Given my vector lengths being what they are (65 mi, 124 mi, 124 mi, 124 mi and 181 mi), I figured segments in the residuals above or below 0.6 could be trimmed. In most cases, not all. I also eliminated sv's (again based on the residual graph) that contributed only 10-15 minutes of data. In a few cases elimination of these brought the error in vector length estimates from 5 feet or so to under 3 feet. For quick and dirty analysis during the individual vector processing I watched the estimated (95%) vector error length for improvement each time I eliminated an sv or trimmed segments of time.

As for the 0.08' difference in Northing?? In L1 vectors of 65 to 181 miles?? Heck, I want to know why this has so far proved repeatable. But, keep in mind my original (and still held) theory is that these results can be obtained with the unknown point being situated within the geometric "interior" of the fixed CORS stations. As for ITRF, one week would not make a shift or rotational difference that could be detected.

I guess I'll have to wait for the dual frequency observation on my POST, both CORS determined and OPUS, to make any true assessment of what I've learned. Haven't checked solar weather this P.M., but if all is cool Test No 3 begins at 0600 UTC 01.340.

J.D. Billings

Blindly going where others have surely been.

JD..Did you send Kent the files and if so , has he had any luck with the processing...I too am wondering about the difference in our two answers..
Just off the top of my head, I'd bet mine would agree with my previous, but again not with yours...
I do have the choice of different iono models and ways to deal with it, but haven't experimented with changing any of these settings....

TM-From the labs at the Optimum Opie Precision Search (OOPS)facility..

If you're interested I'll send you Rinex files for Test 2, and Test 1 again if you need it. Shouldn't need 1 again should you?

If Kent's still interested I'll send him both sessions as well. One at a time.

I plan to install an Ashtech L1 Global Positioning System Device on the Antenna Reference Point of my Fixed Base Station Structure (said that for Brian...really means "screw a Locus on the POST") later tonight to start Test No. 3.

Let me know if you want to look at the files again.

J.D.


Deral,

I'm not sure about the iono modeling. Would the iono file from my Locus session have made a dramatic difference if you could plug it into the solution? Do the CORS station data files carry individual iono files to be used in processing? I don't know. As for the differences in our results, I'm guessing you may have tweaked the data too much. At this stage of the project, I'm still looking for just the right combination of quality analysis tools to use to achieve optimum results. With these vector lengths though, in regards to the amount of data trimming, it does appear that "less is better", at least at the Stumpwater Data Reduction Facility.

Modified By J.D. Billings on 12/5/2001 at 8:47 PM

Makes it easier for me to compare when they are all together !

Stumpwater Data Reduction Facility Results

Original
N 6835851.505
E 3100685.732
EL 402.979

Test #1
N 6835851.427
E 3100685.739
EL 402.873

Test #2
N 6835851.422
E 3100685.757
EL 402.956

Optimum Opie Precision Search Results

Fully constrained
N 6835851.541
E 3100684.934


Jimbo's Crawfish Kitchen
You Catch'm, We'll Cook'm !
Epochs, Bugs, Books, Our Specialties

INITIAL RUN -
No trimming on files holding all 5 CORS sites, H & V.
N 6835851.279 1.275
E 3100685.772 1.295

20° mask - only difference between two runs
N 6835851.133 0.301
E 3100686.352 0.337

****************************************
A different set of CORS !!!
NDS1 ENG1 JTNT MEM2 ARP3
No trimming here either but a 20° mask
N 6835851.088 6.033
E 3100685.569 6.297


Jimbo

PS: Thinking about that dual session !!
Gotta get me a post built - Thursday maybe.




Modified By James Webb on 12/5/2001 at 9:14 PM

Jimbo,

I processed each vector individually (after the initial run) and watched for improvements in the vector length error estimates. When I got the vector error estimates to their lowest possible (seems WNFL had to be around 22°) by masking, I then did the (obvious bad data) sv elimination (one at a time) and watched for changes. Next was cutting suspect segments from high vector residuals, again one at a time.

I haven't looked at the CORS site map to see where your other set of CORS stations are located. Is POST inside the geometry?

Hey, what about the vertical?????

J.D.

soon as you get a station built, I'm ready for the "dual session" spearment


Just looked at the CORS map. Dontcha think 300+ mile vectoring is a bit much????
Modified By J.D. Billings on 12/5/2001 at 10:22 PM

300 miles is kinda long but while we are spearmintin a bit....why not really push it.....especially considering if we run a dual session the logistics of CORS and Locii procesing will have considerably longer lengths. We are both inside the second set of CORS sites, but you are about dead center.

I'm gonna peel some SV's tonite. Maybe help out that Northing coordinate. Once I clean up the SV's Hz, I'll pay attention to the Vert.

I think its interesting that our e/w is just a few hundreths different and Trimbo's is a few tenths. Do you have results from a raw no tweaking run.
How do they match up with mine ??

Jimbo

Let's make sure we're on the same page all the way here concerning the constraint coordinates and system. I'm using the CORS data for "L1 Phase Center of the current GPS antenna" and the "NAD 83 POSITION (EPOCH 1997)"

example coordinate for comparison:

ARL5 Arlington, TX

Lat = 32 45 32.49931 N
Lon = 097 03 36.99084 W
ellipsoid height = 144.120 m

I'm setting up using geodetic/ellipsoid and meters. All constraint data is entered in above format. After processing I then change the settings to grid (my case Texas North Central Zone), ortho elevations/99 ellipsoid and U.S. feet.

Just want to make sure we're using the same procedures when the Texas/Louisana L1 Gumbo merger takes root.

J.D.

p.s. downloaded last night's 9 hours of Locus wisdom to AS2.5 this a.m. for L1/CORS Test No. 3 and am patiently awaiting the appearance of the friendly CORS data for processing. Test No. 4 will be to crunch this stuff all together and see what emerges. I've used a different Locus unit for each of the three test sessions and have left those files on the units. I plan to download them all straight from the units for the combined processing and adjustment.

HOUS WNFL ARL5 DON'T START TILL MIDNITE.

I musta clicked wrong or something. All sites go past your ending time.

I had left it in ellipsod el. mode just cuz. Doesn't really make a diff. in the processing...I don't think, anyways.

Now with 20° mask with time cut to midnite/8:27:00

N 6835851.413 0.148
E 3100685.731 0.167
Ortho EL 403.248 0.179

With default 10° mask

N 6835851.413 0.148
E 3100685.731 0.167
Ortho EL 403.248 0.179

Never realized that the first 1/2 hour was only running against 2 CORS sites...late nite attention.

Jimbo

PS: Even tho the mask angle was changed I kinda wonder if it really did....reprocessing f5 then f7 didn't change any results. Do note the 20° mask was run first.

HHHHHHHMMMMMMMMMMM


Gotta try this again with the 10° mask first.

EDIT 2: No difference. 10° and 20° mask give me the same answer with the time from 0 to 8:27
Modified By James Webb on 12/6/2001 at 11:18 PM