What do you take from an OPUS-S extended report to use as your standard error input to Star*Net or other least squares program?
There is quite a mix of types on that report, and I can't completely relate them. Have I missed the definitive guide to interpreting the report?
A few lines from the top there is an "Overall rms" number. RMS of what? I can't relate it to any of the other tolerance data in the report.
I think I read that the numbers that follow lat/lon/height values are pk-pk differences between solution vectors, right?
I think the covariance values are for the average of the three vectors, which is the reported position. I notice that if I use the covariance matrix diagonal values to compute horizontal or vertical rms, the reported "network accuracy" is about 1.95 times what I compute. Since 1.95 sigma is 95% confidence for a 1-dimensional gaussian normal distribution, I suppose it applies to each horizontal axis with the two axes muddled together, and 2.44/1.95 times that if you want to quote a radial positional error at 95% confidence? They could have made it a little more obvious.
My LS program doesn't accept the covariance matrix as an input, so I'd just use the component sigmas (sqrt of the covariance diagonal values) as standard error on each axis. Am I on the right track? Does your LS use the covariance?
I mostly use OPUS Projects now, but the extended report does contain the G-Files, which Star*Net can import directly. The G-File format specification (FGCS Blue Book Annex N) details that information.
Dr. Chris Pearson, former NGS Advisor to Illinois, created the following:
http://www.oregon.gov/ODOT/HWY/GEOMETRONICS/docs/presentations/opus_results.pdf
The PDF covers both the OPUS and OPUS-RS tools. N.B. the OPUS-RS tool does NOT use peak-to-peak errors.
Discussion below refers to OPUS not OPUS-RS.
Some adjustment packages allow the import of NGS G-Files which are included in the extended report. You have to extract them (cut and paste) and edit identifying numbers/designations. The G-File format contains vector information including DX, DY, DZ components, their standard deviations, and correlations. For complete details of the file format see the ADJUST documentation. There is a link on the home page in the "What's new" at center of the page.
BTW, reading the documentation in the adjustment package will give you insight into the way NGS performs adjustments and how the analysis is performed. There are even sample data sets.
Unfortunately, work to develop a Windoze version of the software was never finalized. I like ADJUST for its "rich numeric output." Unlike the commercial packages it does not have the nice colored lines and other frills.
BTW, there is a new edition (4th) of Alfred Leick's GPS Satellite Surveying. There are now two co-authors. Looking at the table of contents on the Amazon site I see a lot of new info. It has extended coverage of adjustments and related topics.
Unfortunately it is about twice as long (now almost 800 pages) and pretty expensive.
Link to the page on Amazon is here:
Bill,
I extract the "G-file" data from the Extended Output Report (OPUS_Static), as follows (example)
G-FILES
Axx2015 728 15 728
B2015 7281654 15 7282358 1 page5 v1209.04IGS 126 1 2 27NGS 2015 729IFDDPX
IIGS08_1842 IGS 20150426
C00090004 884381846 7 1137170466 15 1594454458 14 X2095ABASEX2095AP030
D 1 2 8462200 1 3 -7652390 2 3 -8898704
And convert it to a COLUMBUS input record (converting the sd & correlations to a 3x3 covariance matrix);
$GPS_COMPACT; base ; p030; 88438.1846; 113717.0466; 159445.4458; 0.0000004900; 0.0000008885; 0.0000022500; -0.0000007499; -0.0000018687; 0.0000019600
The 'translation" of the COLUMBUS record is as follows;
GPS vector
base to p030;
dx 88438.1846
dy 113717.0466
dz 159445.4458
x y z
x 0.0000004900
y 0.0000008885 0.0000022500
z -0.0000007499 -0.0000018687 0.0000019600
I also constrain the CORS as partial fixities, using 5mm (0.005m) as the X,Y,Z variances.
I do my adjustments in igs08 @ the mean epoch of the observation, and also pull the CORS coordinate estimates out of the OPUS report:
BASELINE NAME: p030 base
XYZ -1.1473 -0.2583 0.7337 + XYZ ADJUSTMENTS
XYZ -1758928.9033 -4578603.6455 4066900.0486 NEW L1 PHS CEN @ 2015.5722
XYZ -1758928.8839 -4578603.5919 4066900.0001 NEW ARP @ 2015.5722
XYZ -1758928.2899 -4578602.0490 4066898.6209 NEW MON @ 2015.5722
LLH 39 51 12.02466 248 59 6.30650 2188.1382 NEW L1 PHS CEN @ 2015.5722
LLH 39 51 12.02464 248 59 6.30645 2188.0633 NEW ARP @ 2015.5722
LLH 39 51 12.02466 248 59 6.30650 2185.9103 NEW MON @ 2015.5722
USE the "XYZ NEW MON" data;
XYZ -1758928.2899 -4578602.0490 4066898.6209 NEW MON @ 2015.5722
After the adjustment, I THEN convert the igs08 de jour to NAD83(2011) Epoch 2010.000
NOT saying that is the best way, or even the right way, but it works for me, and I can combine multiple OPUS_Static submissions (same or multiple occupations) together.
Loyal
