Activity Feed › Discussion Forums › Strictly Surveying › To GPS or not to GPS, that is the question
-
To GPS or not to GPS, that is the question
MightyMoe replied 7 years, 8 months ago 25 Members · 63 Replies
-
billvhill, post: 407289, member: 8398 wrote: Ok, so maybe we can attribute it to the instrument man not relaying the message every second or following the rod man fast enough.
I’m so used to working robotic that I forgot about “manual” total stations. With a robot staking is no different than RTK, you go where the display tells you to go. I agree that after a certain distance the total station errors start to catch up to the RTK errors. But again, RTK errors are somewhat dependent upon distance to base.
-
Shawn Billings, post: 407184, member: 6521 wrote: In the case of a so-called one point localization in which the underlying coordinate system is geographic, the software develops a projection at that point, a sort of on-the-fly LDP creator. No transformation involved.
I suspect we are talking past each other a little bit, and it’s how we understand the terminology. I suspect now it could be I’m not using localisation like many people have. I think you guys might mean localisation as a general term for most means of getting GPS measurements to match ground measurements. Single or multiple matching points, 3D or onestep. I have been using the term for the one point, onestep method. I think as a field tool, localisations of the one point, onestep kind are a hassle free and accurate way to perform much work. And at an accuracy within most GPS user’s error budgets. Within a few km, coordinates derived from them should match perfectly fine coordinates derived with the best of care.
On the other hand, I consider transformations based on multiple matching points to be an approximate tool, great for conducting initial searches, some boundary re-marks, and for approximate corner pegging for construction, but not for plans that I’m signing.
Are you referring to a geodetically determined scale factor or the least squares, best-fit, scale factor from observations?
Geodetically determined. That’s all I’ve ever been talking about when referring to my preferred method of localisation, namely, the one point, onestep kind. A determination of scale based on something like the formula (H+R)/R. This will give the user a scale factor perfect for the site, and one good enough to show up longer TPS measurements made without the requisite level of care.
Whenever I’ve seen the term ‘localisation’ used here I’ve taken it to mean the one point localisation with geodetic scale factor. I see it’s not how others exclusively use/take ‘localisation’ in GPS discussions and I’ll remember that for next time. No doubt there will be that next time as RTK GPS discussions generate great excitement and divided opinion here.
-
I use one point localization quite often as well. It’s a fast, accurate way to develop a custom projection with small, generally negligible distortions.
Thank you for struggling through the conversation with me to reach a better understanding. It’s helpful I think.
-
If I understand correctly the term localization is short for a Low Distortion Projection of some kind. When the data collector does the calculations for the localization is it a Transverse Mercator, is it a Lambert, do you get to choose?
Then it needs to apply a scale factor of some kind if you want the elevations to be somewhat close to ground, do you pick it or does the data collector do that? -
IF I am getting this right:
There are 2 fundamental ways to utilize localize.
1.) The manufacture of a LDP, based on known values. This is objective, realistic, and keeps the surveyor in control of what he is doing.
2.) The warping of good data, to fit bad data. This is unobjective, potentially risky, causing a loss of objective “Truth” points.
This is a good discussion.
Maybe we need 2 names, instead of calling them both “Localize”.
Maybe we can call one “Localize to LDP”
And two
“Localize with a crapshoot, and wonder what happened”
Actually, number two can be quite handy, IF you know what you are doing, and later place things objectively onto number one.
N
-
Nate The Surveyor, post: 407324, member: 291 wrote: IF I am getting this right:
There are 2 fundamental ways to utilize localize.
1.) The manufacture of a LDP, based on known values. This is objective, realistic, and keeps the surveyor in control of what he is doing.
2.) The warping of good data, to fit bad data. This is unobjective, potentially risky, causing a loss of objective “Truth” points.
This is a good discussion.
Maybe we need 2 names, instead of calling them both “Localize”.
Maybe we can call one “Localize to LDP”
And two
“Localize with a crapshoot, and wonder what happened”
Actually, number two can be quite handy, IF you know what you are doing, and later place things objectively onto number one.
N
I think the term is being used sometimes as a synonym for calibration, which is a more descriptive term for that process, other times as a truncated way of describing the creation of a LDP,,,,,,,,I guess, I’m not sure………..it seems to be one of the slipperiest words in surveying. 😮
-
Conrad and Shawn, great discussion about localizations. I understand what both of you are talking about. But Charlie c, probably doesn’t. It’s important to understand the RTK system. For Charlie a one point localization will generally remedy your GPS to total station measurements within your job site. Just remember you can’t just turn the thing on and start collecting coordinates without understanding a little about localization.
-
Localizations do one of two tasks:
1. Create a Projection (1-point localization)
2. Perform a Transformation between two planar coordinate systems (1 or more point localization)Creating a projection:
How this is done will depend upon the software. Some will only use a particular projection type. Some give the option. How the scale factor is determined will vary. I can speak with some authority on the Javad localization process. With it, users get to set the projection type, but the default is Oblique Stereographic. The user selects a point or coordinate in the underlying geographic coordinate system (NAD83 or WGS84) with a valid elevation. This is the geographic origin of the projection. The user then provides a point or coordinate of the grid system which become the grid origin for the projection. The scale factor for the projection is determined from the elevation of the geographic coordinate. If this elevation isn’t valid the user can calculate a scale factor and override the scale factor determined from the geodetic position. To my knowledge Carlson Software does not allow for this sort of localization and requires the user to select an existing projection, which brings us to single point localization using transformations.Performing a Transformation:
This can be a based on a single point or multiple points. There is no projection being created in this scenario. There is already an underlying grid system and a set of transformation parameters to transform coordinates in the underlying system to the new local system is determined. Nate commonly creates a localization over a State Plane projection. There is a set of transformation parameters that transform the State Plane Coordinate to his own 100k,100k system. If he chooses to leave the rotation alone, but uses a translation and scale factor, he can have “surface” coordinates with a State Plane Grid bearing relation.
For multiple points, the rotation, translation and scale factor are determined by least squares. This is commonly performed to make an RTK system match an existing total station survey. The scale factor is based on a best fit average of the compared distance from the selected points to the centroid of the points. So you have baselines of RTKpt1-centroid, RTKpt2-centroid, RTKpt3-centroid, etc. and TSpt1-centroid, TSpt2-centroid, TSpt3-centroid. The best fit average scale factor to bring RTKptX-centroid into agreement TSptX-centroid is determined. Whatever errors there are in the measured distance from RTKptX-centroid and TSptX-centroid will affect this calculation. The same process is used for rotation as well (in all three axes N, E and U). Extrapolation becomes a serious issue with this sort of localization. A scale factor determined from a compared baseline of 200 feet may be fine within the 200 feet, but once the operator extends his survey, the scale error compounds. Localizing to large total station surveys will include all of the error from the total station survey in the determined rotation and scale. It’s not exactly correct to call this a “rubber sheeting” even though it has the potential for various error sources. Because the scale factor is consistent, this is referred to as an “affine transformation”. The geometry of the RTK and TS survey are preserved, just scaled and rotated to reduce differences (residuals) in coordinates of common points.My Point summed up:
It’s important to know what the software is doing if you are using localizations. What is the bearing relation? What is the reference surface for vertical differences? How is the scale factor determined and how will it affect the inverse between two coordinates? Will the inverse represent a “ground” distance? If a user cannot answer these questions, he should stick with well known projections like State Plane and to Geoid models until he has a better grasp of the concept (in my humble opinion). As [USER=6642]@Conrad[/USER] said, and I agree, multi-point localizations are fine tools for determining estimated coordinates from surveys that are not geo-referenced but possess a reliable and accurate geometry (such as a modern total station survey). But I would not generally recommend such for a final coordinate system. -
TJE Yogi, post: 407334, member: 963 wrote: Conrad and Shawn, great discussion about localizations. I understand what both of you are talking about. But Charlie c, probably doesn’t. It’s important to understand the RTK system. For Charlie a one point localization will generally remedy your GPS to total station measurements within your job site. Just remember you can’t just turn the thing on and start collecting coordinates without understanding a little about localization.
No doubt this is more than Charlie was asking for. It’s possible to be very productive with RTK and not understand every topic related to projections and transformations. I certainly don’t understand them fully. I avoided any sort of localization for several years as I was almost burned by a localization I performed with Carlson. I then stuck with standard projections. It was not until I started working with Javad to develop his data collection software that I really delved into localizations, so that I could speak with some intelligence about it.
Now I speak with many users providing support and training. Most of them seem to see localizations as a way to avoid wading into the deep end of the geodesy pool. I think that they are inviting trouble in general though as there are so many moving parts to a localization. So when the topic comes up as a panacea to all your grid to ground woes… I think it’s a dangerous perspective to have, and it has probably been perpetuated in large measure by sales people who don’t understand the products they sell.
-
MightyMoe, post: 407321, member: 700 wrote: If I understand correctly the term localization is short for a Low Distortion Projection of some kind. When the data collector does the calculations for the localization is it a Transverse Mercator, is it a Lambert, do you get to choose?
Then it needs to apply a scale factor of some kind if you want the elevations to be somewhat close to ground, do you pick it or does the data collector do that?My understanding of the leica onestep, one point localisation method is that a transverse Mercator projection is created at your location (CM at your location). Scale to ground is calculated automatically and you are left to determine rotation by way of wgs84, grid convergence, user entered or by 2 points. The transverse Mercator is the one the documentation mentions so you can’t choose others. The distortion is very low around a local area and will fit a good TPS survey many miles outwards. Height is matched at your chosen point and you can choose your geoid model to reliably propagate heights around if your geoid model is a true geoid and not a quasi geoid.
In the case of a onestep with more points, the local transverse Mercator is generated before being matched horizontally by means of a 2d translation, rotation about z and a scale. See the following link for why these types of transformations can be more useful than the rigorous 3D transformations in some senses.
-
Nate The Surveyor, post: 407324, member: 291 wrote: IF I am getting this right:
There are 2 fundamental ways to utilize localize.
1.) The manufacture of a LDP, based on known values. This is objective, realistic, and keeps the surveyor in control of what he is doing.
I think you describe it ok. In the one point localisation on my gear, this the the generation of a local transverse Mercator projection to best match your ground point, which should presumably be at a site to best represent the typical elevation of your project. The TM projection offers low distortion over a fair area. And, as scale can be calculated (rigorously?) by the unit, you’d better have decent TPS obs or the GPS may show up you didn’t set your PPM or reduce your distances to a common plane if the project is large enough and changes elevation significantly.
2.) The warping of good data, to fit bad data. This is unobjective, potentially risky, causing a loss of objective “Truth” points.
This could be a description of the full on classic 3D 7 parameter transformation for mine. I don’t know if it’s what you’re referring to though. This is the one that I don’t imagine I’ll ever use. It could be dangerous if not based on pretty good ground control and will tend to hide outliers better as least squares has a tendency to do. Perhaps for a geoidesist doing larger scale work involving control on known projection it could be the method of choice?
Maybe we need 2 names, instead of calling them both “Localize”.
The general nomenclature has caused me confusion twice recently already, but I know what I’m dealing with now so I’m good. But I think it’s helpful sometimes not to generalise as within the localise family there are reliable tools suited to the less edumacated user, and tools not suited to them. We could be steering people in the direction of increased reliability and ease, rather then making them all sound dangerous. Hopefully I’m making sense to you!?
-
I did a similar experiment a year or so ago. It was not nearly so rigorous. I was looking for transformation parameters for our city control network that was based on HARN and Geoid96 and today’s latest adjustment, NAD83, 2011 Adjustment, Epoch 2010, and Geoid12A. I came to much the same conclusion, that a simple translation was all that was needed. Our localization parameters allow for all of the methods mentioned in the paper you linked, except we have no distinct names for each type. If you turn off the North Tilt and East Tilt the vertical translation will be based on a mean of the points selected for vertical control. If you turn of rotation (around the U axis) and scale then the result will be a translation. The user has full control by modifying the parameters.
-
Here is a link (on Javad’s website) to the paper I wrote about my experience:
-
Jim Frame, post: 407293, member: 10 wrote: But again, RTK errors are somewhat dependent upon distance to base.
Are you talking about short distances, surely not long ones.
-
This has been a fascinating thread. It should become a favorite “search” for as long as this forum exists.
-
Conrad, this one of the absolute beauties of the javad LS.
You can have up to 9 systems going at the same time.
And, your underlying spc remain unchanged.
Whereas, with TDS, you had to commit yourself with both feet.
This allows you to construct a different scale factor, for different parts of rhe survey, with differing elevations.
And, your underlying spc remain unchanged.
One job file, different expressions.
It’s kind of like the ucs command, in autocad.
Your core coords are un affected. They are just differing expressions, same work. -
billvhill, post: 407289, member: 8398 wrote: I’m not knocking total stations if that’s what one prefers to work with, but relaying my experience to disbute the comment that nothing beats a two man crew with a total station.
I assume you meant ÛÏdisputeÛ. 😉
You said:
“Ok, so maybe we can attribute it to the instrument man not relaying the message every second or following the rod man fast enough.
The RTK is faster regardless of the reasons.
Yes the total station is more accurate until about 500 feet then the tables begin to turn.”Loyal said:
“If you are talking about a å? acre Lot, then the Total Station should kick the RTK’s butt.ÛÏMy response to Loyal was about å? acre lots, of which we do 100Ûªs. In most layouts the furthest distance we measure is around 150Ûª, and three check in points at various distances. We use the TS for horizontal as well as vertical when staking houses. RTK is simply too slow for us.
Like it or not that is my opinion (like anybody gives a crap anyway), and remember we are all ÛÏSurveying BrothersÛ, and in this together!
Take care bro, and have a great weekend. 😎
-
hmmmm,,,,,localization seems to mean almost anything,,,,,,
I never heard the term until I saw it on this board, I think it needs some further definition. -
I think that Trimble’s “calibration” is synonymous, but would appreciate any definitive evidence to the contrary.
-
Shawn Billings, post: 407762, member: 6521 wrote: I think that Trimble’s “calibration” is synonymous, but would appreciate any definitive evidence to the contrary.
Trimble had it set up so you had some choices, calibration has been strictly a process to merge GPS data to existing control, you basically locate as many points as you want and it will calculate the “errors” it sees and gives you a report for each point located, then you get to accept or reject and apply the calibration, throw out points, keep some, it is a process that can be done in the field with RTK or in the office with RTK or Static. At one time it was even taught to do a calibration AFTER a Static control adjustment.
I don’t like it much, it never made any sense to do it unless you just didn’t have any choice, but if I do have to do it I want to sit in the office and use Static numbers.
You could also do a no datum, no projection, I never liked that choice either, I do know some that used it and there were/are issues with it, I don’t think for most users it’s doing what they think it’s doing.
At one time you could do a plane projection which was what I USED to do, that is basically what I THINK is a one form of localization in Leica speak.
They quit allowing the simple Plane projection about 2000 with the advent of TGO, since then you need to project on a Transverse Mercator system, allowing you to do your own local LDP which I think is another form of what is sometimes called a localization.
There are also options to do a datum shift, apply a subtraction or addition to coordinates and to elevations, you can also rotate, possibly to simulate state plane bearings. There are many ways to mess up a coordinate system by pushing buttons.
Log in to reply.