I recently got into it with a coworker over standard procedures for monitoring, and was curious if any of the monitoring gurus here have heard of this method.
There is a single station point and two reference points on site.
The crew sets up the total station over the station point (same one they use for each monitoring session) and plug in their station & backsight heights. They then:
-observe both reference points
-adjust the instrument position on the tripod (sliding it around)
-modify the instrument height in the data collector
until they get checks within tolerance. They then complete their station setup using one of the two reference points, observe the other one, and start sighting the monitoring points.
The data look unbelievably good, of course - backsight checks and secondary point checks of two thousands or so H/V every time. But they are messing with the location of the instrument and the measure-up to get there. They were instructed to do so by the PM - the one with whom I am having this disagreement.
While I'm not a monitoring expert, I've run my fair share in both the field and office, and I've never employed this procedure before. Number one, because a zero-height resection will yield better results, and number two, because faking in (maybe I should say forcing) perfect observations and instrument heights makes it impossible to properly analyze the data. It goes against every text/manual that I have.
I have problems with other parts of the session observations (no redundancy, F1 observations only, no independent checks, turning record angles to monitoring points without actually sighting them, etc.) but the setup issue is just...bothering me.
Now I'm not in the cool kids' club of our geomatics department. My coworker is, and also very senior to me (my 17 years experience versus his 40), so I predictably got steamrolled when I brought it up. The coworker told me that I didn't know what I was talking about, that "this is bucking in and it is totally normal".
I know what bucking in is, and this is not bucking in; it's forcing a single-backsight setup to look good.
I was brought in to help with data processing, but it's not my project, so I raised my concerns and then butted out, telling them that I wouldn't process the data and didn't want my name associated with the work. I don't even know how I could be confident in a dataset gathered using that setup method, never mind the lack of redundancy or checks.
Am I crazy? Am I wrong? Is this something that others are doing?
My only serious monitoring experience was leveling only.?ÿ So, I would be well outside of the cool kids club.
It sounds like a strange procedure to me.?ÿ They are forcing the instrument set-up to match between their two reference points to within some defined "tolerance", then they proceed with measurements.?ÿ To me, that sounds like it could be a big time drain and has the potential to introduce more human error than just sighting their two reference stations and recording the data as observed.?ÿ That time would be better spent with at least a doubled angle to the monitoring points.
Shouldn't the occupied station, BS to one reference station and turn to the other reference station (preferably more than face left only) be enough to assure they are set-up well??ÿ Then the difference between the two references could be more closely looked at in the data reduction by holding both references as fixed.
Forcing the turned angle to match some record sounds completely wrong to me.?ÿ It seems like they should be sighting on the target no matter what the previous angles are.?ÿ Isn't that the purpose of a monitoring project - see if the targets are moving.
You don't say what is the important parameter being monitored-horizontal, vertical, or both. That could affect the answers.
I'm not an expert either?ÿ but I would think that being as accurately as possible over the control point would be critical for horizontal measurements and set regardless of the secondary points.
If the secondary points are permanent targets on stable structures I can see using them for vertical rather than a measure up. Less so for the horizontal.
Redundancy is your friend.
I guess that procedure makes some kind of sense if if the assumption is that the monument at the instrument position is unstable. The procedure you describe makes sense if the instrument position is the one being monitored for movement.
My procedure would be to resect the instrument position each time from 3 or more points that are well offsite. If I was to set up?ÿ the instrument over a particular monument each time, and run this resection, I'd get a slightly different set of numbers for the instrument each time even though it was over the same point.?ÿ But it should be within the expected tolerance.?ÿ
In any case, I don't think that you are crazy. I hope that there is simply something going on here that you haven't quite been made aware of.?ÿ ?ÿ
I'm assuming the station point and reference points are all control you established and the points being monitored are elsewhere. I think it is fundamental in monitoring to prove your control has not moved before evaluating the points to be monitored. Their procedure is allowing the reference points to move and still control the location of the station point. That in itself would cause me to reject their procedure. If I'm doing the monitoring, I'm using 5 to 6 control points to prove to myself that I'm setup where I think I am then I can have confidence in the positions of the monitored points.
They can have movement in those 2 reference points and still force a position on the station point that works but is not the same location that was used previously thus showing different locations on monitored points that have never moved. I consider that a flawed system for monitoring.?ÿ
You had mentioned the reference points are on site, but I'm going to assume they are outside the zone of influence as I wouldn't trust either method mentioned if the reference points were also monitor points.
I will have to disagree with any mention of re-sectioning for monitors since you can't always set up a good mathematical re-section reference points when monitoring (ie, downtown areas where your only off-site reference points are down an alley in opposite directions), so it's hard to introduce re-sectioning as a standard for monitoring to your field crews.?ÿ
That being said I understand bucking in to 2 reference points (outside the zone of influence) as this reduces your setup error, which would potentially be the largest amount of error introduced prior to the readings. More importantly it also would put you at the position of your original station point, not at the position your station may have moved to during the course of the monitor.?ÿ
This would essentially mean the station point is not looked at as a control point or monitoring point, rather just a position that is able to see all the monitor points from a single setup. I would never report the positional data of this station point for the monitoring reports of course as you are never truly monitoring it.
This is why the method your colleague has suggested in my eyes is superior to setting up over the station, backsighting a reference point, and checking the 2nd reference point. You would be able to see if your station point has moved, but then what? Let's say your station point has moved 0.1', and your reference points don't check. You are now monitoring from a point that you know has moved, but you aren't sure where it currently is at, and collecting data for the monitoring points on site from that.
That's my 2 cents at least.
?ÿ
I will have to disagree with any mention of re-sectioning for monitors since you can't always set up a good mathematical re-section reference points when monitoring (ie, downtown areas where your only off-site reference points are down an alley in opposite directions), so it's hard to introduce re-sectioning as a standard for monitoring to your field crews.?ÿ
I agree that there are circumstances where good resectioning isn't possible. That doesn't seem like a reason to reject resectioning as an option in all cases.
I've usually found that the monitoring program is terminated when the excavation work is complete. In other words, while the monitoring is going on the sight lines are clear enough. Later, when the building rises above ground level and blocks lines of sight, the monitoring is no longer needed.?ÿ?ÿ
That being said I understand bucking in to 2 reference points (outside the zone of influence) as this reduces your setup error, which would potentially be the largest amount of error introduced prior to the readings. More importantly it also would put you at the position of your original station point, not at the position your station may have moved to during the course of the monitor.?ÿ
What is the difference between this and a 2 point resection?
?ÿ?ÿ
I will have to disagree with any mention of re-sectioning for monitors since you can't always set up a good mathematical re-section reference points when monitoring (ie, downtown areas where your only off-site reference points are down an alley in opposite directions), so it's hard to introduce re-sectioning as a standard for monitoring to your field crews.?ÿ
I agree that there are circumstances where good resectioning isn't possible. That doesn't seem like a reason to reject resectioning as an option in all cases.
I've usually found that the monitoring program is terminated when the excavation work is complete. In other words, while the monitoring is going on the sight lines are clear enough. Later, when the building rises above ground level and blocks lines of sight, the monitoring is no longer needed.?ÿ?ÿ
That being said I understand bucking in to 2 reference points (outside the zone of influence) as this reduces your setup error, which would potentially be the largest amount of error introduced prior to the readings. More importantly it also would put you at the position of your original station point, not at the position your station may have moved to during the course of the monitor.?ÿ
What is the difference between this and a 2 point resection?
?ÿ?ÿ
I'm not saying that the reasoning to reject re-sectioning is due to those cases. I'm stating that when your company is training field crews it is easier to come up with a repeatable procedure rather than utilizing their understanding of mathematical error ellipses to make their choices on how to set up the monitor. They should have a certain level of general understanding of this though if you are having them set up a monitor in the first place, but it's hard to expect that they all understand it.?ÿ
The difference between a 2 point resection and bucking in would be the ability to spot setup error. Re-sections produce residuals to decipher, while bucking in is much more simplified. Backsight 1 point, stake to the other, don't like it, reposition and try again, can't get it to fit within tolerance, might need to check your equipment. So you either rely on your crews understanding of residuals shown during a resection versus the guess and check method until they get the simplified results needed. I'd choose the latter, much more dummy proof.
My only monitoring experience is with dams. For those we setup a control quad off of the dam preferably with pedestal monuments. So the instrument and prisms are locked onto the monuments and do not move. We occupy each and turn every possible angle amd shoot every distance forward and reverse 8 times. These data are then analized and run through a LSA. This is just the contol quad and does not include the monitoring point work or the levels. It also does not include the calibration baseline work and tribrach adjustment work that occurs prior to travel to the site.?ÿ
I wrote all of that to say, reading the procedure you described that included moving the instrument around on the tripod until the desired solution is reached sort of blows my mind. But I??m locked into my procedure and mayne that is keeping me from being able to understand the reasoning.?ÿ
Isn't the point of true monitoring to aquire A LOT of data/measurments on every point (including the instrument), so you can run mathematical analysis on each point, and also the trends in their relationships to one another via those measurements?
If the site isn't permanently set up for a monitoring schedule all day, every day, different times, different temps, different pressures, etc. for some duration, you're not really "monitoring". Visiting a site once a week (pick your period) and setting everything up (or just the gun) simply doesn't acquire enough data to be meaningful to assess the quality of the positions in the context of "monitoring" a site at the 0.001' - 0.005' level of accuracy.
Watch some of the Trimble 4D Control/Montoring videos on their website or youtube, if you're not familiar.
Lots of people with 40 years worth of experience are wrong. Especially surveyors and geomatics cool kid duffers.
"Bucking In" would be the antithesis of lots of data and mathematical analysis of said data.
A method that may have been adequate 40 years ago because that was the best "technology" to be had is simply not going to cut it by today's standards with the technology we are using on a daily basis. If a professional is not current with new tools, new tech, new methods, and rigorous analysis, admit your shortcomings and learn. The future is here, and pining for the past should be retired, including surveyors and geomatics people with "40 years of experience" trying to throw their weight around because they're too proud to admit they might not know something.
Sure, you might be able to throw a chain better than I, but most production oriented surveyors aren't using chains because we have better, more consistent, and efficient tools.
Every job is different of course, but I'd imagine the type of monitor the OP is doing is keeping an eye out for is subsidence during excavation for a building, which is what most monitoring jobs are for.
Mi-Other-Left, I can't tell if you are being serious or facetious. What kind of monitoring is done at the scale you mention of 0.001'/0.005'? I've had to align the main turbine at a hydroelectric dam to that level of precision, but monitoring? Any Surveyor asking for help through this website for that kind of monitoring advice is in the wrong forum.
@tim-libs I agree. I keep the table in the attached image on my bulletin board for reference. Its the USACE requirements from their Deformation Survey Manual. These kinds of numbers take a lot of work, preparation, and good instrumentation to acquire.
Thanks all for the replies, this is why I love this forum.
To answer a few of the questions posed and issues raised, this is for monitoring in both the horizontal and vertical, with a minimum movement detection levels of a half inch and quarter inch respectively.
A few of you hit the nail on the head with respect to resection versus the "bucking in" procedure.
I have always chosen resections because they not only give residuals and allow for refinement of position in the field, but cuts out measure-up and centering error as well as allows for statistical analysis of the observations. (To address the geometry issue that @tim-libs raised - it's almost never a concern with angle + distance observations, and in any case I will place points spaced out in the vertical if I can't space them out in the horizontal. I always run preanalysis to make sure I can hit my desired precisions anyways.)
And as @bushaxe observed, a fundamental part of monitoring is running least squares analysis, first to check that your reference points (which are ostensibly out of the zone of influence) have not moved (so better have more than just one for a check), then identify and remove any blunders or outliers, before finally constraining to reference points and obtaining error ellipses on the monitoring point observations.
The manuals and texts make it clear that last point is critical to monitoring campaigns, because we have no idea whether a monitoring point moved if we have no way to correctly compute error ellipses. Only if the absolute positional change exceeds the combined error of the current day and previous day's computed ellipses can we call the movement statistically significant, and thus likely to be actual movement. No error ellipses? No way to tell.
Because the bucking-in procedure does not incorporate true observations of the backsights (they've been "forced" to look good) I cannot be confident in any statistical analysis. It's skewed data and will yield skewed results.
Beyond that, performing a single observation on each reference point, and a single observation on each monitoring point, gives me maybe 3 or 4 degrees of freedom in a network of 10 or more unknowns, where the recommended number per the USACE deformation manual is at minimum the number of unknowns and should be twice that.
Again, such low redundancy is not good for statistical analysis. Couple that with the screwy setup method and I get some very, very unusual results. Every time. I've never seen anything like it in any of the networks I've adjusted before.
@michigan-left : I've worked monitoring jobs that require continuous monitoring (I'm familiar with T4D but never set it up personally) and jobs that don't.
In my experience it depends on the type of structure, its condition, and whether work is being actively done. Soldier pile wall? We'll check it throughout the day as it is set, do it again the next day, then once a day because the work is being done away from the wall. Large hydro dam with seepage problems that is currently being shored up? Yeah, continuous monitoring 24/7 is the way to go.
This particular project is for boring a utility line (24" casing) under a major roadway. Work is not continuous, and the location required some pretty strict structural design requirements, so this is not the same level as a failing dam or a large mine with continual excavation. I'm not a structural engineer, so I will defer to their expertise on whether things are expected to move and how often they need to be checked.
There are different tolerances for the launch/landing pit and the targets set in the lanes/medians. The vertical is tighter than the horizontal, but based on our gear and the projects I have done before we should be able to hit target specs and detect movement.
But I can't reliably work with not-enough data incorporating skewed observations that look far better than they should be.
Beyond that, I think @lurker made the most critical point and basically sums up what I told the PM:
?ÿ
Their procedure is allowing the reference points to move and still control the location of the station point. That in itself would cause me to reject their procedure. If I'm doing the monitoring, I'm using 5 to 6 control points to prove to myself that I'm setup where I think I am then I can have confidence in the positions of the monitored points.
They can have movement in those 2 reference points and still force a position on the station point that works but is not the same location that was used previously thus showing different locations on monitored points that have never moved. I consider that a flawed system for monitoring.
?ÿ
Just as disturbing is the fact that they are merely turning record angle (staking out) to the monitoring points, then measuring without actually locking on to or sighting the monitoring points. Of course things are going to check perfectly.
A proper monitoring project allows for setup and observation without "cooking the books" on the raw data. Beyond being a massive liability for the firm, it's just poor procedure, and there are at least 4-5 junior staff that are being taught that this is acceptable.
Aggravating, but not surprising. Going to drop it and go do some records of survey.
Again, thanks to all for the input!
where the recommended number per the USACE deformation manual is at minimum the number of unknowns and should be twice that.
Just a bit of disclosure. I was part of a team that wrote the last update to that manual. I'm not patting myself on the back or anything. I just wanted to let it be known that my perspective is biased toward those procedures. It took a lot of work to review the existing manual and determine what if any procedures and nominal accuracy requirements should be updated. I'm glad to know it is being used as a reference outside of USACE projects.
My apologies as I stand corrected. Under those circumstances I believe bucking in is inappropriate in the way you've said it's being done. They should always store their backsight shot and check shot on their 2 reference points to begin with even when bucking in. The station setup should also be stored for reference as well. And even better would be checking into a 3rd/4th reference point with good geometry.
Are you saying the monitor points were then staked out and measured for distance only, without sighting for vertical or angles? Seems completely wrong if that's the case. I have no problem with re-sections as long as the person utilizing them understands the limitations and the initial monitor is set up correctly. It just makes it harder to spot human/equipment error on those running the instrument at the time of monitoring.
?ÿ
@bushaxe?ÿ
Well in that case I owe you a great big thank-you, because when I switched from just being the guy on the ground executing the observation plan to actually planning the project myself, I relied most heavily on that manual.
I also have Precision Surveying by Ogundare, Geodetic Network Analysis by Kuang, and of course Ghilani's Adjustment Computations for reference, but the USACE manual was far and away the best resource.
On the first project that I ran myself, using the techniques outlined in there helped us detect a tiny amount of sloughing in a roadway behind a soldier pile wall shortly before the edge collapsed. It turned out to have minimal safety implications, but we were continually asked whether our data was correct, and because we did our due diligence, it was.
On that one, the client also contested our fees for monitoring (well over six figures of T&M in addition to our scoped construction ) but since we had backed up our procedures with a detailed plan (referencing the manual) that clearly stated what it would take to achieve their goals, as well as how much time it would take, we took the wind out of their sails almost immediately and got paid in full.
Not only does shoddy work incur a great deal of liability, but it also leaves money on the table and incorrectly convinces clients that monitoring is a trivial task...
I don't have a problem with bucking in for certain applications, and it can be a great tool like you mentioned upthread. It just didn't seem right for monitoring, though, and it was really throwing me when my network adjustments returned incredibly strange results in the statistics. It's a good tool to have in the toolbox but I just can't work with the data in the way that the monitoring analysis requires.
Yeah, they have additional problems with how they are sighting (or rather, not sighting) the monitoring points themselves. I can't get a straight answer from anyone about exactly what they are doing, but an office tech mentioned that the are just staking out but not sighting the points (turning record and hitting measure) plus the raw data and their raw checks indicate way, way better checks than I have ever seen before. Something's fishy, but again I'm backing off. I've made my concerns known in emails saved and backed up on my personal PC...
Would love to see that "bucking in" stuff input to least squares. They are monitoring "something" but it sounds like a bunch of one-third-giraffe scale monkey business.
?ÿ
@rover83 We had a really good team that worked well together.
I've yet to do a monitoring project but it looks to me like there would be zero difference between sliding the instrument around to fit the reference points and not having any reference points at all.