Kent McMillan, post: 415981, member: 3 wrote: Another case where Star*Net is a truly brilliant solution is in downtown districts where the obliteration of evidence of the surveys made in the streets by the City Engineer has been nearly complete. In Austin, the City Engineer's work with transit and tape was excellent during the period from about 1913 through 1950 or so. There are, as a result very good records of networks that were surveyed across some districts that tied centerline monuments together that have since disappeared. If one can find and position a sufficient number of marks that remain, the adjustment of the transit and tape network via Star*Net is probably the best reconstruction possible.
In the first example I'm not sure if you're saying you didn't find anything or didn't look. If it's the former then the restoration is controlled by the statutory or customary method. Never solved thst method in Texas so i'm not certain what it would be. In most States i work in it would be a grant boundary restoration which is not a least squares adjustment. The software could be used but it would be less efficient than using basic CADD functions.
In the second case, the nearest reliable evidence should control. Rejecting long-standing right of way monuments, lot corners, party walls and other possession for a math solution is one of the biggest problems we face in our profession.
As Gary Kent says, stop slapping the math on the ground and passing people on to the attorneys. Pure math solutions (that includes least squares) are technical tools used in arriving at a professional judgment.
thebionicman, post: 415998, member: 8136 wrote: In the first example I'm not sure if you're saying you didn't find anything or didn't look. If it's the former then the restoration is controlled by the statutory or customary method. Never solved thst method in Texas so i'm not certain what it would be. In most States i work in it would be a grant boundary restoration which is not a least squares adjustment. The software could be used but it would be less efficient than using basic CADD functions.
In the second case, the nearest reliable evidence should control. Rejecting long-standing right of way monuments, lot corners, party walls and other possession for a math solution is one of the biggest problems we face in our profession.
As Gary Kent says, stop slapping the math on the ground and passing people on to the attorneys. Pure math solutions (that includes least squares) are technical tools used in arriving at a professional judgment.
In the first example, the corners in question had been originally established along the bank of the Rio Grande. The river in that area has historically shifted by hundreds of feet and it was easy enough to determine that it was impossible for some of the evidence of various corners to have survived the processes at work in the river channel. Texas does not have the cookbook familiar to PLSSia, so the land surveyor is left to weigh and reconcile evidence by the most reasonable method instead of falling back on a mechanical set of instructions.
As for the surveys by the City Engineer, the object of the exercise was to construct the positions of the centerline monuments that have been accepted as defining the rights-of-way of the streets, in many cases following an action of the City Council adopting them as the correct locations or a decree of District Court. The centerline monuments are the right-of-way monuments in that the policy of the City was and continues to be to recognize the right-of-way lines as being at exact distances from the established centerlines.
It is one thing to say that "the nearest reliable evidence should control", but quite another to reconcile that evidence comprehensively as can be done in a least squares adjustment. As a general rule, any construction that uses a barely sufficient amount of evidence to reconstruct an important location is a recipe for an unpleasant surprise later.
Kent McMillan, post: 416009, member: 3 wrote: In the first example, the corners in question had been originally established along the bank of the Rio Grande. The river in that area has historically shifted by hundreds of feet and it was easy enough to determine that it was impossible for some of the evidence of various corners to have survived the processes at work in the river channel. Texas does not have the cookbook familiar to PLSSia, so the land surveyor is left to weigh and reconcile evidence by the most reasonable method instead of falling back on a mechanical set of instructions.
As for the surveys by the City Engineer, the object of the exercise was to construct the positions of the centerline monuments that have been accepted as defining the rights-of-way of the streets, in many cases following an action of the City Council adopting them as the correct locations or a decree of District Court. The centerline monuments are the right-of-way monuments in that the policy of the City was and continues to be to recognize the right-of-way lines as being at exact distances from the established centerlines.
It is one thing to say that "the nearest reliable evidence should control", but quite another to reconcile that evidence comprehensively as can be done in a least squares adjustment. As a general rule, any construction that uses a barely sufficient amount of evidence to reconstruct an important location is a recipe for an unpleasant surprise later.
It appears you are saying no, but not actually committing to no. I won't assume either way, but I will make an observation. You are so bent on finding a way to insult the PLSS (and it's Surveyors) that you again miss the point. In the narrow circumstance you describe there will be a math solution for sections of the line. The method of restoration is driven by the long established order of calls, not the BLM Manual. We do use judgment which much of the time says leave that book on the shelf.
In your city example, we have dozens of blocks in the same boat. The ties to nearby corners were observed at the time of the original survey and have been perpetuated to varying degrees. If a map appeared in the record ignoring these nearby corners in favor of some mathemagical solution i would do my best to educate the other Surveyor. If they refused to understand i would have the board explain it to them.
I am a long time user and advocate of StarNet software. Finding unique ways to use it is great, right up until the time it becomes your evidence.
Don't see what the PLSS has to do with anything here. For LS, I can see the use of analyzing and adjusting ones own raw data or as an analysis tool to compare results to record or to adjust record between found monuments, like a Compass Rule Adjustment or Grant Boundary Adjustment, these all seem legit. I question these other uses like a best fit line or some other weighted resolution where the program is actually calculating your final boundary positions. This seems to be a radical departure from traditional methods of honoring an element of record, say record bearing or record distance. Is this sound/legal practice to be using the program to actually compute a final boundary position?
thebionicman, post: 416015, member: 8136 wrote: It appears you are saying no, but not actually committing to no.
Sorry, I thought it was clear from the context that I didn't waste any time searching for the rock mounds built to mark corners whose positions are now hundreds of feet in Mexico and that were washed into the river as the bank eroded. There were some locations where the corners fell in flats where erosion hadn't altered the channel that drastically, but those showed evidence of farming and clearing that explained why nothing remained. The record of the 1910 resurvey was good enough that it was possible to determine where he had searched without success. Considering that the flats along the river were being farmed then and the great thickets of tamarix had yet to appear it was reasonable to conclude that if the 1910 surveyor couldn't find a cottonwood stump, there was nothing to be found presently.
In your city example, we have dozens of blocks in the same boat. The ties to nearby corners were observed at the time of the original survey and have been perpetuated to varying degrees. If a map appeared in the record ignoring these nearby corners in favor of some mathemagical solution i would do my best to educate the other Surveyor. If they refused to understand i would have the board explain it to them.
I am a long time user and advocate of StarNet software. Finding unique ways to use it is great, right up until the time it becomes your evidence.
That's fine, but suppose you had no maps of record giving ties to the vanished centerline monuments and suppose that all that remains in some districts are some sparse remaining centerline monuments that had been tied together by transit and tape surveys of good quality. At some point a surveyor has to concede that the evidence of the record of a prior survey can be the best evidence that remains. To call a retracement made on the best evidence merely mathematical unwarrantedly supposes that there is any other basis available.
My concern in using least squares to resurvey a boundary is that it ignores the original method of survey. Since you don't have the raw data you have to assign weight to lines that were never run to connect your found monuments.
A simple example:
Say I find original monuments at the NW cor. and NE cor. of a subdivision and one in the interior. Most subdivisions were surveyed by traversing the exterior and then setting interior points from the exterior points. Why would I use the error in position of an interior corner to relocate a lot corner on the North boundary. In this case a simple proportion would be a better solution.
There certainly are situations when least squares might make sence. I think in resurveying an RTK survey it may be appropriate.
aliquot, post: 416056, member: 2486 wrote: My concern in using least squares to resurvey a boundary is that it ignores the original method of survey. Since you don't have the raw data you have to assign weight to lines that were never run to connect your found monuments.
A simple example:
Say I find original monuments at the NW cor. and NE cor. of a subdivision and one in the interior. Most subdivisions were surveyed by traversing the exterior and then setting interior points from the exterior points. Why would I use the error in position of an interior corner to relocate a lot corner on the North boundary. In this case a simple proportion would be a better solution.
There certainly are situations when least squares might make sence. I think in resurveying an RTK survey it may be appropriate.
A least squares adjustment of record data requires adequate constraint. In the situation you mention, that sounds unlikely to be the case. Unless the orientation of the subdivision (bearing basis) is well known and can be reproduced with minimal uncertainty.
I really appreciate Helmert transformations for evaluating complex geometries quickly. Most of the time, if I use it to calculate and set missing corners, it's going to be in an interior lot in a relatively modern subdivision. There are times that there is no single precise, long baseline to which to rotate. Often I question the validity of rotating to a single pair of points when there may exist many potential pairs. How does one justify arbitrarily selecting a single pair in such instances? In a matter of a couple of minutes, literally, I can see the goodness of fit for dozens of points without ever using the inverse. I can immediately spot outliers (such as most points agree within 0.15' but one is off by 3 feet).
Having said that, there are times to keep that tool in the box, or at least to minimize its use. Old surveys that were surveyed line by line instead of as a network would be one example. Another would be comparative dignity of calls and junior/senior rights. If a corner was supposed to be on a straight line between two senior monuments, then the theoretical transformation point must yield to the senior line. Most of those old surveys, particularly compass and chain, should be surveyed line by line and not as a network in most cases (based on my experience anyway).
Helmert transformations determine the best fit rotation (in three axes), translation (in three axes) and scale. That's seven parameters, thus the other name for Helmert transformation is the 7 parameter transformation. For boundary purposes, we typically only use one rotation (around the z axis), three translations, and one scale. In most cases, I believe it's best to ignore scale and only use rotation and the three translations, unless you have reason to believe that there is some scalar difference between "your" foot and the previous guy's foot.
roger_LS, post: 415848, member: 11550 wrote: I am curious how least squares is being used amongst surveyors out there. I've always thought of it as a tool to adjust systematic errors out of ones survey, but am wondering to what extent it may be being used to determine final boundary placement. From time to time I've picked up maps and been perplexed about how a specific corner was reestablished; record bearings weren't held, neither were record distances, proration not used. There seems to be no rationale behind anything. So I'm wondering, are there folks out there using least squares to actually create final boundary positions?
The tool I use in subdivision retracements where I am relatively confident that all or most of the monuments are original is Transform. It is invaluable for understanding how monuments are positioned with respect to the record. It uses Least Squares to compare Record and Measured coordinates - nothing like Star*Net, which deals with measurements between monuments. The help file alone is worth more than the cost of the software.
Shawn Billings, post: 416109, member: 6521 wrote: I really appreciate Helmert transformations for evaluating complex geometries quickly. Most of the time, if I use it to calculate and set missing corners, it's going to be in an interior lot in a relatively modern subdivision. There are times that there is no single precise, long baseline to which to rotate. Often I question the validity of rotating to a single pair of points when there may exist many potential pairs. How does one justify arbitrarily selecting a single pair in such instances? .
By taking cross tie inverses between found monuments and record and comparing the bearings, this pattern would demonstrate how much angular variation there is in the survey and show how strong a particular bearing base is. Are you using this transformation just to find a bearing base or to also resolve for missing corners? If you were resolving for missing corners, I assume that you'd have the program hold all the found points (excluding obvious outliers) then adjust the record in between, so that you'd end up with final calcs slightly off record in both bearing and distance. Is this how the process would work?
[USER=11550]@roger_LS[/USER] there are three ways I use this in the real world. Before I go into them, let me say this: In one sense a transformation can bring an "old" coordinate system to a new coordinate system. We're looking to bring the boundary calls from that survey in 1995 to our shiny new coordinate system as accurately as possible. The other way is to get our new coordinate system fitted to an "old" coordinate system. We're looking to do an as-built on a lot with numerous short lines and multiple curves along the frontage. The math is the same, the perspective is not. Helmert transformations can go both directions.
I use the old>new approach when I want to quickly verify coordinates on monuments I've found that are described in a modern survey (EDM era) and develop tight search coordinates for remaining points. I likely won't use this for stake-out because this approach only considers the modern survey and not the legal aspects of rights of adjoining property owners.
I use the new>old approach for boundary analysis and staking. If there are enough monuments that fit within my personal (or statutory, if applicable) precision, then I may elect to report the calls verbatim from the subdivision plat. In Texas, we no longer have a statutory limit, but depending on the lot, I would probably allow up to a tenth of a foot in each monument before I would break away from the plat calls and report my own measurements. It's an arbitrary tolerance, but it's where I've landed. So after doing the best fit transformation, and comparing the resulting residuals (coordinate inverse of matching pairs), I would stake any missing monuments using the calculated transformation. I don't think I could do any better with any other method and the fact that most modern surveys are radial, I don't think working exclusively from the two adjacent monuments to calculate a missing coordinate is always the best solution, particularly if there are two or more consecutive monuments missing. This places me on the plat quickly and gives me a good idea of the accuracy comparison between my work and the previous work. Also imagine a lot where with 8 monuments (points of curvature and tangency and corners). Four of them are missing, making it impossible to pick two points on a line of any substantial length to which to rotate. If the geometry of the four found monuments is evenly distributed (not clustered in one quadrant of the lot) then Helmert can really be an asset.
The third way is to use it like a resection. If I return to a survey I performed in the past and some of my control is missing, I may traverse through, tying in as many old points as possible, then use Helmert to get my new traverse onto my existing control. Resections are great, but sometimes you can't see all of the control from a single setup, which is a prerequisite for resections. I can traverse through and use Helmert to get back onto control and verify that the control I'm using is within tolerance.
Roger - I strongly urge you to check out Transform by Primacode Software. I think it is exactly what you are looking for. You can download a copy that will work with a limited number of points and access the Help file - its worth reviewing!
Jim in AZ, post: 416250, member: 249 wrote: Roger - I strongly urge you to check out Transform by Primacode Software. I think it is exactly what you are looking for. You can download a copy that will work with a limited number of points and access the Help file - its worth reviewing!
Jim's Transform probably has some really nice features, but you may be surprised to find that your current software may have a best fit transformation routine in it already. I'm not trying to talk you out of Jim's suggestion, just be aware it's in a lot of office and field software.
[USER=6521]@Shawn Billings[/USER] Thanks for the response, sounds like some powerful stuff, I'll have to check it out. From what I'm hearing, you are not really using this for boundary resolution, per se, just fixing yourself onto the coordinate system of a modern map and letting it distribute this tenth floating around here and there, where it may. You're not in downtown New York City, that tenth or two is insignificant and is not even being reported. After learning about this best fit line routine in the other thread, where significant discrepancies were resolved by computer programs, my real curiosity is how folks might be using these programs to resolve actual significant differences between field and record. Does this method have merit? Has it ever been tested in court?
I would not use this for reconstructing boundaries with significant differences in call coordinates/measurements and as measured coordinates/measurements. I'd use conventional reconstruction techniques such as intersection and holding called direction and distance.
Well you will find it is different state by state as to what people use, and there is no one solution to do anything. Here in Washington, per the Department of Natural Resources if you are doing GPS, "A least squares adjustment or other multiple baseline data analysis must be performed to produce weighted mean average point coordinates and point uncertainties in order to verify that the required level of positional accuracy has been achieved". I don't use least squares all the time but if the site is such that you can have multiple base line data and multiple observations to points from other parts of the network then I will.
The problem I have is that all these programs will show you how well their program works, and how easy it is......but that is using a good clean set of data. You have to have all of your field crews do their traverses in a certain way. For us we have 4 brands of instruments and of course all are proprietary don't talk together, so if you figure out a good process/method for using least squares on one, you more than likely cant do it for another, unless you use something like Starnet which accepts multiple formats. I have Starnet, Traverse PC, Civil3D, Trimble Office, Spectra Office and Topcon Magnet Office Tools, and I agree it's difficult to wrap your head around trying to make something work for all.
To me, it is clear that most of these programs are not written by surveyors, or at least surveyors that have heavily worked in the private sector. They all have their strengths and weaknesses and It seems like they are all trying to add features that aren't necessary for most. Some features/options make me scratch my head as to why they would they ever be needed, and just cause confusion and greatly increase possibility for someone to miss a simple setting and really make things go bad fast. Just need to keep things simple. Also, some of them are more geared to calculating the completed data only and not easy to maintain, keep on top of data on a continuing project from initial survey, to staking and then asbuilts without spending a lot of time.
Sorry, I got a little off topic.
In regards to your single missing corner question, I would not use least squares and more than likely I would also hold bearing bearing intersection. As for multiple missing corners, you wont often see anyone around my area balancing out multiple deed/record courses to make a monument work, you just hope that it is written with enough metes/bounds calls to figure out what they were trying to do. If they determine the corner/monument should be held, then usually I see people calculate holding the angles and letting the distances float. However every situation is different, like a puzzle, which is why I like being a Surveyor.
Brian Gillooly, post: 416807, member: 12268 wrote: produce weighted mean average point coordinates and point uncertainties
Brian Gillooly, post: 416807, member: 12268 wrote: The problem I have is that all these programs
Another problem is that there is a basic assumption with two possibilities the programs might use (with some possible minor variations) regarding the estimation of the uncertainties, so you don't always know what someone is reporting.
Star*Net reports the standard error (sigma) of the results assuming that you told it the correct standard errors for your measurements. Some other programs assume you don't know the correct standard errors of your measurements, only the relative proportions between them, so it should estimate the overall scaling for you. This results in a more pessimistic report. This difference between programs is not well-known.
Some old threads:
https://surveyorconnect.com/community/threads/breaking-news-carlson-survnet-14-error-ellipses-now-smaller.329213/
https://surveyorconnect.com/community/threads/relative-positional-precision-why-did-i-fail.281864/
A long thread where I eventually was taught this fact
https://surveyorconnect.com/community/threads/least-squares-ellipse-confidence-calculations.242017/page-3