I ran one of my sessions through DPOS and am puzzling over the report.?ÿ They mostly copy the OPUS format for the first half-page, but one thing doesn't come close to matching. ?ÿ A very quick search didn't find an explanation of their report.
I don't think the number following the coordinates (for instance 0.004m) has the same meaning in the two reports.?ÿ For OPUS it is the peak-to-peak difference, the max -min of the three values obtained by solving with respect to three CORS.?ÿ But if I look at the later part of the DPOS report, I find a range of lat, lon, or El Ht in the cm range versus the 4 mm shown in the summary.?ÿ So is that Javad's estimate of a standard error, or are the individual CORS solution values not what I think they are?
I don't know. But I should. Waiting for answer to this too.
Most jobs, DPOS base position repeats within 0.06 or so.
I don't have one in front of me, but I imagine the numbers do mean something slightly different than the OPUS.?ÿ From notes I keep in front of me:
DPOS fixes 15-20 best ambiguities first, then recalcs all other float ambiguities to get final integer solution; so does not give a percent used like OPUS but instead will say Yes or No to fixed solution.
RMS < 3cm is good quality indicator same as OPUS.
Geometry factor < 2 is good quality indicator.
Uses 5 stations because knows less about CORS data quality than OPUS.
Peak to Peak values are one of the best quality indicators and are present in the DPOS report.
I ran one of my sessions through DPOS and am puzzling over the report.?ÿ They mostly copy the OPUS format for the first half-page, but one thing doesn't come close to matching. ?ÿ A very quick search didn't find an explanation of their report.
I don't think the number following the coordinates (for instance 0.004m) has the same meaning in the two reports.?ÿ For OPUS it is the peak-to-peak difference, the max -min of the three values obtained by solving with respect to three CORS.?ÿ But if I look at the later part of the DPOS report, I find a range of lat, lon, or El Ht in the cm range versus the 4 mm shown in the summary.?ÿ So is that Javad's estimate of a standard error, or are the individual CORS solution values not what I think they are?
My understanding from the developer is that the report gives "a value that equals : sqrt(sqr(SigmaX) + sqr(SigmaY) + sqr(SigmaZ)), where SigmaX, SigmaY, SigmaZ are diagonal elements of covariance matrix."
Personally, my favorite part of the report is the residuals of position. This shows the difference of the individual vector from the five stations used from the adjusted vector for the final solution. I typically ignore the statistics and go straight to this portion of the report. (The values are in meters).
I think you are right that the source is the "covariance matrix", but there should be a couple differences from your explanation.?ÿ A covariance matrix contains sigma^2 values, but it looks more like Javad gives you a mislabeled Correlation matrix.?ÿ You are then right to square the values to get covariances.?ÿ Add the values CovX1+CovX2+ ... CovX5.?ÿ?ÿ Repeat for each axis rather than combining axes.?ÿ But shouldn't you divide by the number of values (5) before taking the square root?
That is a measure of the variability of the position during the measurement, but ignores any error in the CORS positions relative to their record values.?ÿ That fact makes the residuals, which are affected by the CORS accuracy, an important measure.
