My StarTDS converter doesn't mean the sets (we do 4) although I have the option checked. I think that artificially increases the redundancy. On most projects it would be a big effort to mean all those sets. I never thought about it reducing the effect of centering errors.
Thanks Mark and Bill, I read the manual and my understanding of the standard error of the measurements or instruments used but not the formula of the standard deviations.
I have input my instrument settings in already for my adjustments and those stated above was the comments on the report I submitted. Now to find an automated way to do this. I have more than a thousand lines to process for these!
Initially I ran adjustments for all my raw observations.
Is microsurvey or starnet able to re write it as M line plus my standard errors as listed by Mark?
I collect all of my conventional data in SDR format and import it into Star*Net using StarSDR.
In the scenario you describe, the answer is fairly simple. You just measure as many sets as you want to do without entering a new backsight record. In otherwords, two sets would be:
BS F1
BS F2
FS F1
FS F2
BS F1
BS F2
FS F1
FS F2
On import the BS and FS directions (and slope distances and Z angles) should be averaged and will appear as the mean angle, distance and Z angle in the input file.
The mean of two sets will naturally have a lower standard error than one set would. Specifically, it will be expected to have a standard error equal to the standard error of one set divided by the square root of the number of sets. So the standard error of the mean of two sets is expected to be 0.707 x s.e. one set.
The way to take that into account in the adjustment is to create an instrument in the "Instrument Library" called "SokkiaX2" or whatever that is to entered as an inline option in front of all measurements taken as the mean of two sets in the input file.
If you group them together in the input file, that means that the inline command
.INSTRUMENT SokkiaX2
will lead the group and will be followed by another ".INSTRUMENT" inline command to restore the project default.
The advantage to this approach is that all of the repeat sets from the same instrument setup will be treated as having the same random errors of instrument and target centering as opposed to considering them to be independent setups with different random centering errors.
In the project option screen, instrument tab, (see Jim Frame's posting early in this thread) you set your standard errors for the instrument you are using. These are the default standard errors that are applied to each measurement. You do not usually enter errors for each individual measurement unless they are different from the default for some reason.
Kent McMillan, post: 338384, member: 3 wrote: I collect all of my conventional data in SDR format and import it into Star*Net using StarSDR.
In the scenario you describe, the answer is fairly simple. You just measure as many sets as you want to do without entering a new backsight record. In otherwords, two sets would be:
BS F1
BS F2
FS F1
FS F2
BS F1
BS F2
FS F1
FS F2On import the BS and FS directions (and slope distances and Z angles) should be averaged and will appear as the mean angle, distance and Z angle in the input file.
The mean of two sets will naturally have a lower standard error than one set would. Specifically, it will be expected to have a standard error equal to the standard error of one set divided by the square root of the number of sets. So the standard error of the mean of two sets is expected to be 0.707 x s.e. one set.
The way to take that into account in the adjustment is to create an instrument in the "Instrument Library" called "SokkiaX2" or whatever that is to entered as an inline option in front of all measurements taken as the mean of two sets in the input file.
If you group them together in the input file, that means that the inline command
.INSTRUMENT SokkiaX2
will lead the group and will be followed by another ".INSTRUMENT" inline command to restore the project default.
The advantage to this approach is that all of the repeat sets from the same instrument setup will be treated as having the same random errors of instrument and target centering as opposed to considering them to be independent setups with different random centering errors.
Hi Kent,
I am curious what kind of SDR is that? Below is what I download from my total station.
13NMBacksight HR:0.000
09F1 100 1016.56900000 87.88805556 0.00000000
09F1 100 1028.56400000 91.25361111 318.69555556
09F1 100 1038.56400000 91.25361111 318.69527778
09F2 100 1038.56400000 268.74888889 138.69305556
09F1 100 1038.56400000 91.25236111 318.69541667
09F2 100 1028.56400000 268.74888889 138.69305556
09F1 100 1028.56400000 91.25236111 318.69555556
09F2 100 1016.56900000 272.11444444 179.99750000
02TP 100100.00000000 100.00000000 0.00000000 0.00000000
03NM0.00000000
07TP 100 1010.00000000 0.00000000
13NMBacksight HR:0.000
09F1 100 1012.02500000 95.29861111 0.00000000
13NMBacksight HR:0.000
09F1 100 1012.02300000 95.29916667 0.00000000
13NMBacksight HR:0.000
13NMBS check 100 - 101:ZE95.29917,SD2.023,HD err= 0.002,VD err= 0.000
13NMBS Circle Check : horizontal angular err= 0.00000, vertical angular err= 0.00056
09F1 100 1042.02400000 95.29833333 0.00000000
09F1 100 1052.02400000 95.29750000 0.00027778
09F1 100 1062.02500000 95.29861111 0.00000000
02TP 100100.00000000 100.00000000 0.00000000 0.00000000
03NM0.00000000
07TP 100 1010.00000000 135.00000000
13NMBacksight HR:0.000
09F1 100 1018.35500000 92.37750000 0.00000000
13NMBacksight HR:0.000
09F1 100 1018.35500000 92.37750000 0.00000000
09F1 100 1078.35500000 92.37694444 0.00055556
09F1 100 1086.60500000 89.29583333 40.31361111
09F1 100 1095.38000000 88.05000000 97.62611111
09F2 100 1095.38000000 271.94944444 277.62333333
09F2 100 1086.60500000 270.70444444 220.30972222
09F2 100 1078.35500000 267.62250000 179.99694444
09F2 100 1018.35500000 267.62250000 179.99694444
09F2 100 1018.35500000 267.62250000 225.00000000
09F2 100 1078.35600000 267.62277778 225.00027778
09F2 100 1086.60500000 270.70500000 265.31388889
09F2 100 1095.38000000 271.94916667 322.62722222
09F1 100 1095.38000000 88.04944444 142.62722222
09F1 100 1086.60500000 89.29583333 85.31750000
09F1 100 1078.35600000 92.37694444 45.00305556
09F1 100 1018.35600000 92.37694444 45.00305556
09F1 100 1018.35600000 92.37694444 90.00000000
09F1 100 1078.35500000 92.37666667 90.00027778
09F1 100 1086.60500000 89.29611111 130.31500000
09F1 100 1095.38000000 88.05083333 187.63000000
09F2 100 1095.38000000 271.94861111 7.62388889
09F2 100 1086.60500000 270.70416667 310.31055556
09F2 100 1078.35500000 267.62277778 269.99805556
09F2 100 1018.35500000 267.62277778 269.99805556
09F2 100 1018.35500000 267.62277778 315.00000000
09F2 100 1078.35600000 267.62277778 314.99888889
09F2 100 1086.60500000 270.70472222 355.31222222
09F2 100 1095.38000000 271.94916667 52.62472222
09F1 100 1095.38000000 88.04972222 232.62777778
Will try the instrument error as what you describe later. Thanks1
That looks generally like SDR format, I assume that the column formatting is what prevents my version of StarSDR from converting that file.
I use an SDR33, but as far as I know SDR format is SDR format. It shouldn't depend upon the device logging it.
Kent McMillan, post: 338391, member: 3 wrote: That looks generally like SDR format, I assume that the column formatting is what prevents my version of StarSDR from converting that file.
I use an SDR33, but as far as I know SDR format is SDR format. It shouldn't depend upon the device logging it.
Hi Kent,
Have sent you a pm. Thanks
I don't think that I've ever gotten an email through surveyorconnect.com. I don't know why that is. My email is
kentmcm AT swbell DOT net.
I have been bringing all of my Total Station data in as individual measurements and now I am concerned that setups with multiple rounds could be unfairly weighted over ones with only 1 round.
I think the solution is to use one of the converters to mean the rounds. I have avoided these before because I like to see the F1/F2 split etc;
Would there be an option in Starnet to treat average and treat multiple observations in the DAT file as one mean observation? I use DB, DM, DE records in the DAT file. So each setup is quite clearly defined.
I have been bringing all of my Total Station data in as individual measurements and now I am concerned that setups with multiple rounds could be unfairly weighted over ones with only 1 round.
It really depends mostly upon the sight lengths and the contribution of target and instrument centering errors. For example, if the sight lengths are more than 100m and the target centering has a standard error of +/-1mm, then that has an angular uncertainty component of about +/-0å¡00'02" in just the direction to the target.
However, those components add by root sum of squares. So if the standard error of the basic horizontal angle measurement (apart from centering errors) is, say +/-0å¡00'04", the net effect of target centering errors is:
SQRT [ 4^2 + 2^2 + 2^2 ] = 0å¡00'04.9".
Treating repeat angle measurements at the same setup to the same targets at sight lengths of more than 100m is going to relax the standard error, but not by much. When sight lengths are much shorter than 100m then the relaxation of weights may become problematic.
If you're traversing by forced centering with target and instrument centering errors of effectively 0.0mm, it shouldn't matter whether you enter observations as independent measurements or as means of a series of repeats.
Make that: " For example, if the sight lengths are more than 100m and the target centering has a standard error of +/-1mm, then that has an angular uncertainty component of less than about +/-0å¡00'02" in just the direction to the target."
Kent McMillan, post: 338487, member: 3 wrote: Make that: " For example, if the sight lengths are more than 100m and the target centering has a standard error of +/-1mm, then that has an angular uncertainty component of less than about +/-0å¡00'02" in just the direction to the target."
I am thinking particularly of a new (redundant) leg I added to a traverse today. One of the stations on this leg only has a single round. The rest generally have 3 rounds. I am thinking that any adjustment will not give this observation as much weight. That is the angle observation and also the distances observed at this station are not as numerous so do not have as many entires in the DAT file.
In reality, the leg is a more direct route through the main area where I need accuracy (new bridge site) and I would like that angle measurement to have equal weighting. The leg used new setups on existing stations so there will be a centring error as well as any angle and distance errors.
Thanks for your advice it is greatly appreciated.
I have each set of rounds as a separate line of data. So when I measure three rounds each round is an M line of data. When I do 2 rounds there are 2 M lines of data. And when I do one there is only one line of M data. Each line is a product of the data from that particular round. So each line has its small variations. It's not 3 copies of the average of 3. For example:
[pre]
DV 2-1 309.9535 90-47-03.66 5.210/4.740 'CP BP
SS 2-1-1000 359-59-58.70 309.9508 90-47-01.39 5.210/4.740 'CK 1
M 2-1-100 208-46-35.11 8.8551 91-45-41.93 5.210/5.000 'CP BM4491
M 2-1-101 70-45-50.99 7.7417 91-05-08.26 5.210/5.000 'FD BS LEAD
M 2-1-101 70-45-50.02 7.7390 91-05-06.01 5.210/5.000 'FD BS LEAD
M 2-1-102 341-38-03.59 164.8529 90-42-17.45 5.210/5.000 'FD HOLE IN LEAD
M 2-1-103 315-57-54.42 71.4834 89-37-12.57 5.210/5.000 'FD BS WASH
M 2-1-104 287-44-48.91 161.3961 89-04-46.87 5.210/5.000 'FD HOLE IN LEAD
M 2-1-105 269-42-46.76 152.6424 88-53-53.69 5.210/5.000 'FD BS WASH
M 2-1-3 269-41-18.96 309.7725 89-03-30.15 5.210/4.920 'CP BP
M 2-1-3 269-41-18.96 309.7708 89-03-30.31 5.210/4.920 'CP BP
SS 2-1-1001 0-00-04.54 309.9508 90-47-03.01 5.210/4.740 'CK 1
[/pre]
Each single line is weighted the same, and thus if I measure 2 or 3 rounds that measurement gets double of triple weighting in the adjustment.
I am thinking particularly of a new (redundant) leg I added to a traverse today. One of the stations on this leg only has a single round. The rest generally have 3 rounds. I am thinking that any adjustment will not give this observation as much weight. That is the angle observation and also the distances observed at this station are not as numerous so do not have as many entires in the DAT file.
In reality, the leg is a more direct route through the main area where I need accuracy (new bridge site) and I would like that angle measurement to have equal weighting. The leg used new setups on existing stations so there will be a centring error as well as any angle and distance errors.
I'd say that if you want angles measured once at one control point to have the same weight as angles measured several times at another control point, the thing to do is to remeasure the angle that was only taken once several times more. If this is a critical control survey, you'd want to do that anyway, I'd think.
The idea is that all measurements contain errors and the task of the surveyor is to attempt to insure that those errors are random and of relatively well known distributions. In other words, the standard error is a fairly realistic estimator of the size of error that about 2/3 of the measurements made with the same process under the same conditions won't exceed.
But an angle taken as the mean of three repeated measurements is expected to have a standard error significantly less than that of any single measurement with the same equipment under the same conditions. So it's realistic to reduce a measurement by adding more repeats, but not realistic to arbitrarily increasin the standard error of the mean of a series of repeats just to do it.
Norman Oklahoma, post: 338510, member: 9981 wrote: I have each set of rounds as a separate line of data.
I do the same thing, partly or if habit, and partly because SurvCE doesn't know how to record multiple rounds correctly.
I make an RW5 in Access then convert that with StarTDS.
I wind up with an M line for each pointing so 4 rounds results in 8 DV lines and 8 M lines. I'm not particularly fond of that but haven't found an efficient way to fix it.
Kent McMillan, post: 338391, member: 3 wrote: That looks generally like SDR format, I assume that the column formatting is what prevents my version of StarSDR from converting that file.
I use an SDR33, but as far as I know SDR format is SDR format. It shouldn't depend upon the device logging it.
Hey Kent,
Have managed to get hold of microsurvey and they figured the problem. It is in the missing instrument record. Magnet Field did not include that in. Thats what causing the error to convert.
Cheers,
Jan