Starnet help

If memory serves:Starnet help

If memory serves:

TB 1
T 2 218-__-__
Etc
Etc
T 7 75-__-__
TE 8

I don't use the Traverse feature much. I just do a network, but I'm sure I can figure out the traverse format. An advantage of a network is that you can easily add redundant measurements and cross-ties to the basic info, and I'm not sure yet how much of that you can do with a traverse format.

I was initially confused by the renumbering from the picture I saw from you, about August 3, and am updating my own program data to match. Star*Net should be able to take any point numbering you like, regardless of where you start or the numerical sequence.

You are providing azimuths? The manual says your input should be angles turned at each point UNLESS you pick an option for azimuths. But since your actual measurements were angles I'd stick with entering angles because I think the statistics will be more meaningful.
T At_point angle distance

I'll work on it as I have time, and try to post something that works, but thought you might want those comments now.

I'm taking baby steps. I thought by definition I was doing a traverse, not a network. That's where I started in the tutorial. I can change the calls back to angles and find the tutorial for networks. My guess is you'll get to the finish line before I do, lol.
Thanks very much for the assistance.

Oh, and BTW, My traverse is a bit different than before: I threw out my bad stations, did some new good ones with triple angles, and re-measured my first angle (POB-2-3) very carefully. Sorry for the confusion.

.MAP ON

C 2 1000.00 1000.00 ! !

TB
T 2 218-__-__ 329.69
ETC
ETC
T 9 75-__-__ 376.702
TE 2

TB
T 2 123-36-47 130
T 1 123-36-47 108.81
TE POB

You did a traverse, but can enter the data from a traverse in network format, which is what I usually do, or enter it as in traverse format.

I haven't fully digested the traverse options, but did get something to run and give reasonable answers. There is another suggestion posted that may be better than mine in dealing with point 1.

I used mostly your older angle and distance values, so you will probably want to update them.

# 2D Traverse by rfc
# Bill's version  2014 Aug 13

# Be sure to set estimated standard errors
#  for centering, distance, and angle
#  in the project parameters
#  so the statistics are meaningful.

C   1  1000 1000 ! !
B  1-2  N56-23-13W   # 303-36-47
D  1-2 130

TB 1
T  2  94-57-47  329.69    # angle 1-2-3, distance 2-3
T  3 177-08-20  170.195
T  4 142-58-33  170.195
T  5 282-10-17  271.084
T  6 264-22-53   96.09
T  7 116-34-41   43.16
T  8 274-56-30  384.65
T  9 218-55-50  376.702    # angle 8-9-2, distance 9-2
TE 2

Thanks, Samurai, and Bill93 for another option. Here's what Samurai's version says. I've no clue yet whether it's good or bad news. I hope those ellipses don't indicate the possible real physical locations of where my stations really are, lol. I'm in trouble if they do.
Got lots of reading to do.

I'm uncertain what to use for the instrument settings, or what effect they have on the output. Here they are:

Got lot's of studying to do.

Under Plot Options / Sizing there is an Ellipse Exaggeration Factor. Perhaps a good value would be about 10 for this problem. It's arbitrary, just used to make the ellipses show up nicely in the plot, since they would be almost invisible on a tight survey at the same scale as the traverse.

Put in the instrument specs for distance and angle error. A modest total station might be +/-5 mm = 0.016 ft and 5 ppm; many are better. I always forget for sure, but I think the standard requires you to do an average of two direct and two reverse angles to get the instrument spec, so if you did single angles loosen that up.

For sure add typical centering error that is now showing as zero. For instance, if the reflector pole wandered an (rms) average of +/-0.05 ft, then put that in for the target centering and if your instrument was set up within 0.02 over the points use that for instrument centering.

When you get done, and have fixed all blunders, you should find the report saying your "Error Factor" for angle and for distance are in the neighborhood of 1. If they are large, you aren't admitting to the program all of the variability of measurements. If you fail the Chi-squared test, it means your error estimates are not realistic or you still have a blunder.

All of those estimated error tolerances are used to weight the measurement to come up with the best estimate. For instance, when the backsight or foresight is a short leg of the traverse the angle can't help as much because no matter how accurately you measure it, you may be measuring to a slightly wrong point due to centering error. The program takes that into account and puts less dependence on that angle measurement than on one with long sights.

Finding Blunders First

You have to consider your observations first. If you observed angles right, that is what you work with.

I entered your data? into a .dwg and noticed a considerable closing error, 0.308', from 2 to 10. I decided to let that error tell me where to start looking. I constructed a perpendicular to the closure error and it pointed almost directly to the observed angle at pnt 7. That makes sense as 7 is the shortest BS/FS combination so any error gets magnified. However the closure error still seemed too great for modern equipment so I looked at the angular error right at pnt 7 as in angle 2-7-10.

The 0.308' distance error is an angular error of 0°01'31" which says blunder more than error, but let's look at the picture a little more. What else does the data tell us?

From 2 to 7 the distance is 701.254', from 7 to 10 is 701.257'. Totally reasonable numbers which indicates no probable distance errors.

Let us assume all the angular error is in 6-7-8. That 0°01'31" error would place pnt 8 0.019' from where it is currently calculated. Instead of an angular error that 0.019' could be setup and/or observation error.

It is necessary to reobserve angles at 6, 7 and 8 with careful BS and FS setups and I suggest 4D4R because of the disparity in distances. 7-8 = 43', 7 to 10 is 701', meaning 16 times magnification of error.

An observation adjustment cannot really adjust anything if there are insufficient observations. If angle (POB/1)-2-3 was observed then 9-10-(POB/1) needs to be observed and the data included. Alternatively 9-10-3 could be observed and included.

My observation is that you are trying to run before you learn to walk. My grandson did that and fell flat on his face quite often. Lose your balance while walking and you can gently sit down and re-evaluate the situation.

You cannot adjust out blunders, you need to correct them.

Paul in PA

Finding Blunders First

> You have to consider your observations first. If you observed angles right, that is what you work with.
>
>
> My observation is that you are trying to run before you learn to walk. My grandson did that and fell flat on his face quite often. Lose your balance while walking and you can gently sit down and re-evaluate the situation.
>
> You cannot adjust out blunders, you need to correct them.
>
> Paul in PA

Well, between you and Bill93, this is like having my own personal Surveying Professors at my service! Thank you (especially for the part where you suggest I have to go out and do it again, which I will).

I believe I can see 8 directly from 6, so my plan is to try to shoot 6 to 8 and 8 to 6. I'm unclear though whether I use 7 at all for these angles. Is it better to use 9 as a backsight to shoot 6 (and 5 as a backsight to shoot 8), or just do the triangle 6-7-8 extremely carefully?

I'll also set the prism in a tribrach to rule out the well documented "Unstable Wife Phenomenon".:-D

I will grant you one thing though: I'm a "like to run before learning to walk" kind of guy. I'm trying to learn the math surveyors use to adjust measurements at the same time as learning sound practices to make sure those measurements are as good as they can be in the first place.

Can't I just blame it on the instrument? It is, after all a 20 year old 10" device.

Finding Blunders First

Is there not a Key in error for your distance from |6-7|,
I see you entered 96 feet where it should be 96.09 feet.

Chr.

Walk First

Since this is a learning exercise, collect data at corners 6, 7 & 8. You can also bypass 7 and collect data for 5-6-8 and 6-8-9, but set it aside for now. Get the least squares solution that you originally intended to work.

Then use your alternate observations for a new traverse 1-6,8-10.
,
Finally, combine all your observations, yes very redundant in the weak corner and then compare the results.

Having done all three you will gain very good knowledge to be used in planning future traverses and the advantages of redundant observations.

Also, check that your least squares program is granting extra slop to those short line angles.

Paul in PA

Finding Blunders First

> Is there not a Key in error for your distance from |6-7|,
> I see you entered 96 feet where it should be 96.09 feet.
>
> Chr.

Yes! Fixed that; Thanks.

Walk First

Interesting. First, once I put what I believe to be more realistic numbers in for the instrument (10"), and centering errors (I used .02'), the ellipses turned to circles. See plot.

Also, once I fixed the 96.09 error, the StdErr has a big number from 7-8...135.53. Do you think that is telling me the same thing you are about the potential for 6-7-8 to be the culprit here?

Walk First

If you had observed angle 9-10-3 you could have nailed 7.

Small steps in the right direction.

In reobserving 6-7-8 you want to be very careful in aligning your prisms to the instrument to avoid any extra angular error. I would not doubt that you could get an 0.02' error if the prism is off 30° and you were sighing the imagined center of the glass.

Paul in PA

Confused

> If you had observed angle 9-10-3 you could have nailed 7.
>
> Small steps in the right direction.
>
> In reobserving 6-7-8 you want to be very careful in aligning your prisms to the instrument to avoid any extra angular error. I would not doubt that you could get an 0.02' error if the prism is off 30° and you were sighing the imagined center of the glass.
>
> Paul in PA

I DID measure 9-10-3....it's 37 07' 04", from 4x readings Left and Right. But that's the same number as calculated from the difference in azimuths between 9-10 and 10-3. I thought that when you're trying to close a traverse, you have to leave one point "open" to let the closure get reduced.

Can I just add that angle as a measured angle (which it is) into Starnet, and would that change things? Not sure how to do it. Is that the disadvantage of using azimuths rather than just turned angles in Starnet? Should I start again, using angles, not azimuths?

Observed Angle Sould Have Been Close To 37°05'33"

Please post all observed field angles.

Redundant measurements are what least squares needs to give results closer to the truth.

Paul in PA

Observed Angle Sould Have Been Close To 37°05'33"

You're probably right. But I'm horribly messed up now, because, going through my field notes (they still are terrible), I found yet more discrepancies. The program I'm using to plot may be introducing misleading errors. The table above is (to the best of my knowledge, your Honor), the measurements I took.

I'm going to try putting them back into Starnet as angles rather than azimuths, and see what I get.

Starnet syntax error

[msg][/msg]

Argh! Still not right

Starnet syntax error

You forgot the dashes in the angles. Star*Net insists on that to help with the parsing of optional entries on the line.

There are many approaches to figuring out the discrepancies. One that I would consider is getting a distance from point 4 to something on the other side. Angles would be a bonus. Cutting the problem in two should go a long way toward isolating any blunder.

You are going to run into the number of points limitation on Star*Net's demo mode, so you might have to do a separate 2-leg traverse file from 4 to x to 8 or whatever and then transfer resulting distance 4-8 into your main traverse file.

I'm a little confused about the discussion of point 10. I took that as the computed end of the open traverse, replacing point 2 for sighting from point 9 only. When you close the traverse, everything would refer to point 2.

And to save on point count, you don't need both 1 and POB.