I went to the Javad Party in Hot Springs, last week. The Nate Report

Out of Shawn's presentation of the 7 perimeter localization, I came up with an idea. Do you remember the old "Cobbler's Bench" Children's toy? (I used to have one as a kid) IF I took 2 of those, made them LARGER, and used the points on the nails, as Coordinates, it could be a useful tool, to explain to a NON initiate, just what is happening. Shawn did something quite useful. He used an airplane illustration. Yaw, Roll, and Pitch, to explain the vertical relationships, from SPC to some local system. Of course, there is Scale, Easting, Northing, and Elevation.
Shawn also said "The earth does not yield itself easily to a mathematical model".
Anyway, I am trying to come up with ways and means, to bring the older crowd, into the wonderful world of newer technology.

There is a 2 fold process, in all of this. First, to know what is going on, then how each manufacturer addresses things. I feel that for MANY, they are weak in both, and some have never heard of it!
Technology is fine, but it TENDS to hide the mechanics of the process, that we sign off on. Shawn is helping this for many.

HP 42 and a total station, and my old compass.... 🙂

N

Never have quite figured out what a localization is, but I sure wouldn't want to Yaw, Roll and Pitch SPC into a local system. 😮

Well Mike we're only talking seconds we're not talking minutes with yaw, pitch and roll.
But... Some localizes... Dont know...

Nate The Surveyor, post: 432324, member: 291 wrote: Well Mike we're only talking seconds we're not talking minutes with yaw, pitch and roll.
But... Some localizes... Dont know...

Generally, a calibration or localization is warping an existing control network into a projection that will work with measured latitudes and longitudes, I can't see where SPC comes into play, it seems like a layer of data that would just make the stretch and fit of a calibration more complicated than necessary.

Of course if the idea is to warp the lats and longs into existing SPC then.........but if so why is there a second set of coordinates.
Think I need a head scratching emoji;)

MightyMoe, post: 432343, member: 700 wrote: I can't see where SPC comes into play,

Well, that is what is fascinating with the LS.
It uses "Pages".
Think ACAD LAYERS, sort of.
Page 0 can be set to show LAT LON (The underlying coords inside, are LAT LON)
Page 1 can be set to SPC, AR South,
Page 2 can me set to SPC AR North.
Page 3 can be set to SPC north Texas.
Page 4 can be set to to a local system, true north, at some point, and tilted a little, due to changes in gravity. Approximated "Ground".
Page 5 can be set to Old survey,
And, while all this is going on, THE CORE FUNDAMENTAL Lat Lon are unchanged. Those Pages are projections, based on unchanging lat lon.
This allows you to set up something local, and delete it, when not needed.
It's like "Multiple Expressions, of the same core coords."

I think I understand your questioning.... I used TDS, 200C for a long time, and it does not allow "Multiple, and simultaneous projections". It's a one trick pony. (Mine still has the TDS sticker).

I often set ARKANSAS south, to be my "Base Reference Frame", This is one layer up, from Lat Lon.
Then, I set a minor scale factor, for a local, LOW distortion projection. This way, Coords, in, and Coords out, are all close approximations of "Ground".
Typical is 108 ppm, or 87 ppm. Or such like, in my area.

Usually, that's an average of the area.

If you want to study the area, select ANY coord in the LS, and it will give you Grid to Ground, and both scale factors, ie EF and SF, and combined. Also, it will give you convergency, all relative to the underlying SPC that you have selected. You can select the whole job, and it will give you the AVERAGE scale factors, and average convergency. (Think Theta).

Just this little utility to look at scale factors is HANDY.

IF I can tie another surveyor's base, and I know he USED Geodetic at base, and all scaled to ground, @ base, I can then fully replicate his projection, and convert either his to mine, or mine to his, WITHOUT changing my underlying lat lon's. The pages are just "Windows to see the data through".
So, I can do a single job, and view it through a number of projections.
I can tie into a BLM job, that used "True North for the reference frame, for every line". And, I can ENTER these brngs as geodetic bearings.
THEN change pages (windows) and see these coords re-projected, onto SPC, of my choice, or a local rectangular coord system, that approximates ground. I can traverse in either system, by making that page current.

Owning an LS, and going to one of these little educational parties, makes me feel like an 8 yr old, getting schooled at stuff I SHOULD HAVE KNOWN, for years. Actually, alot of it I did know... but the doing of it, is no longer a theory... it is becoming a part my my daily practice.
Then, when my job is all done, and I want to RE-express the job, in any format, I can do just that. Output CSV (That's Comma Separated Values) ascii file, in pure LAT LON, or SPC, of my choice. This way, I can put a few SPC on my plat, with all the metadata.
Several other things... I can use up to 256 characters in note....

It's not for everybody. But, it's been good for me, to PRACTICE the THEORY, and have tools to do it, in any format I wish...

There is a certain fulfillment, in owning, and operating a tool, made for me, that allows me to live every day, in the practice of these things. It's not just theory. I use it every day.

Carry on...

N

I guess I did not answer your question...

Localize is a projection, on top of a projection.

Your SPC is a projection.
Your local coords are a projection, On top of the local SPC reference frame.

I feel like I am preaching to the choir or here... Mike, I feel that you are MORE knowledgeable than I am, on this subject, but, I have a tool that allows, more versatility than my old TDS way.
Gotta go....

N

Anybody take a bath?

Nate -

ALL GPS systems use ECEF Cartesian coordinates for positioning. As you said, any grid system is just a display mask; grid coordinates are calculated on the fly. In the case of SPCS, the ECEF coordinates are converted to LLh which is then projected onto the SPCS grid using the projection for the zone you're in, normally Lambert for states running predominantly east - west and Transverse Mercator for states running predominantly north - south.

"Localized" coordinates - in Trimble, anyhow - bear no relationship to State Plane, except in specific instances where a site calibration is being used to fit, for instance, VRS positions into an older SPCS control network. But when using truly "local" coordinates, like a plant grid, the software projects the LL to a local TM with its origin at the site, then performs the transformation (I think a seven parameter Helmert, but I need to look that up to confirm) between that local TM and the local grid coordinates. For the vertical it calculates a best fit plane, and you have the option of first applying a geoid model, which will reduce the distortion in areas of significant geoidal undulation.

The ability to page through different coordinate systems in real time is a neat feature, but it's not difficult to change the coordinate system in Access either.

What ALL GPS users need to realize and understand is that we put way too much emphasis on coordinates - what is actually important is the measurement. GPS vectors are stored as delta xyz ECEF Cartesian coordinates (not azimuths and distances, etc.). Anything other than that is a calculated value based on parameters that you have entered into your software.

Regarding TDS - the reason I never liked TDS for GPS was that they took a piece of software that was written around using grid coordinates and tried to make it geodetic, and that's the wrong way to go about things. Truly geodetic software like Access (or any Leica GPS I ever ran) does not require any grid definition unless you need to perform COGO.

Just to clarify... Trimble uses a seven parameter Helmert to transform between geodetic datums. For a site calibration it uses a four parameter transformation for the horizontal component - translations in Northing and Easting, and a single rotation angle and scale factor. The vertical component of the site calibration is treated separately.

Lee D, post: 432447, member: 7971 wrote: Just to clarify... Trimble uses a seven parameter Helmert to transform between geodetic datums. For a site calibration it uses a four parameter transformation for the horizontal component - translations in Northing and Easting, and a single rotation angle and scale factor. The vertical component of the site calibration is treated separately.

As I understood it back in the day when calibration was the only way to get GPS close to elevations, Trimble used a ft/ft ratio to simulate the geoid in a small area which in essence creates a plane. In the early days for one of the firmware upgrades meters got mixed into the ft/ft calculation, that was a difficult one to track down..... 😎

Nate The Surveyor, post: 432426, member: 291 wrote: Well, that is what is fascinating with the LS.
It uses "Pages".
Think ACAD LAYERS, sort of.
Page 0 can be set to show LAT LON (The underlying coords inside, are LAT LON)
Page 1 can be set to SPC, AR South,
Page 2 can me set to SPC AR North.
Page 3 can be set to SPC north Texas.
Page 4 can be set to to a local system, true north, at some point, and tilted a little, due to changes in gravity. Approximated "Ground".
Page 5 can be set to Old survey,
And, while all this is going on, THE CORE FUNDAMENTAL Lat Lon are unchanged. Those Pages are projections, based on unchanging lat lon.
This allows you to set up something local, and delete it, when not needed.
It's like "Multiple Expressions, of the same core coords."

I think I understand your questioning.... I used TDS, 200C for a long time, and it does not allow "Multiple, and simultaneous projections". It's a one trick pony. (Mine still has the TDS sticker).

I often set ARKANSAS south, to be my "Base Reference Frame", This is one layer up, from Lat Lon.
Then, I set a minor scale factor, for a local, LOW distortion projection. This way, Coords, in, and Coords out, are all close approximations of "Ground".
Typical is 108 ppm, or 87 ppm. Or such like, in my area.

Usually, that's an average of the area.

If you want to study the area, select ANY coord in the LS, and it will give you Grid to Ground, and both scale factors, ie EF and SF, and combined. Also, it will give you convergency, all relative to the underlying SPC that you have selected. You can select the whole job, and it will give you the AVERAGE scale factors, and average convergency. (Think Theta).

Just this little utility to look at scale factors is HANDY.

IF I can tie another surveyor's base, and I know he USED Geodetic at base, and all scaled to ground, @ base, I can then fully replicate his projection, and convert either his to mine, or mine to his, WITHOUT changing my underlying lat lon's. The pages are just "Windows to see the data through".
So, I can do a single job, and view it through a number of projections.
I can tie into a BLM job, that used "True North for the reference frame, for every line". And, I can ENTER these brngs as geodetic bearings.
THEN change pages (windows) and see these coords re-projected, onto SPC, of my choice, or a local rectangular coord system, that approximates ground. I can traverse in either system, by making that page current.

Owning an LS, and going to one of these little educational parties, makes me feel like an 8 yr old, getting schooled at stuff I SHOULD HAVE KNOWN, for years. Actually, alot of it I did know... but the doing of it, is no longer a theory... it is becoming a part my my daily practice.
Then, when my job is all done, and I want to RE-express the job, in any format, I can do just that. Output CSV (That's Comma Separated Values) ascii file, in pure LAT LON, or SPC, of my choice. This way, I can put a few SPC on my plat, with all the metadata.
Several other things... I can use up to 256 characters in note....

It's not for everybody. But, it's been good for me, to PRACTICE the THEORY, and have tools to do it, in any format I wish...

There is a certain fulfillment, in owning, and operating a tool, made for me, that allows me to live every day, in the practice of these things. It's not just theory. I use it every day.

Carry on...

N

Nate, that all is interesting, what I was curious about was why SPC would be used to do any kind of calibration, and it looks like that isn't the case, you just have different projections based on one set of lat, longs which makes perfect sense to me. This is how most of us have worked along for quite some time, hold the measured lat, long and use different projection parameters to shift between coordinate realizations.

Just to clarify what Nate was saying.

A localization (or Trimble calibration) is a comparison of two grid systems (2D or 3D Cartesian coordinates). The math is based on a Helmert 7 parameter transformation, which is a least squares approach to determining the best translation (NEU), rotation (around NEU axes) and scale (7 different parameters) between two sets of grid coordinates that represent the same points. The Helmert transformation doesn't require any geodesy. A Helmert transformation can be made for any 2 grid systems. For geodetic purposes, one grid system needs to be a projection, the other (local) system, however, is typically unrelated to the Earth (or is treated so by the localization process). The use of a localization is to determine the relationship of the local system (unrelated to the Earth) to the Earth by means of common points expressed in a known projection. The result of a localization isn't necessarily a new projection but a transformation (or set of instructions) to get from the local system to the projection system.

As Lee mentions, it's possible to not use a common projection, such as State Plane, and set a controller to geographic (LLH) coordinates. But (again as Lee points out) a bit of slight of hand is taking place behind the scenes. The controller is creating a projection at the project site and projecting the LLH coordinates to this intermediate projection, then the software is comparing the local coordinates to the grid coordinates from the intermediate projection to determine the transformation.

As Lee also mentions, TDS was a grid based software with GPS support added later, so the database was not geodetic. Carlson is very similar. The point database is grid and a user must select a grid system when creating a new project which prevents reprojection on the fly.

As to Mighty Moe's comments about warping. Technically a Helmert 7 parameter transformation doesn't "warp" anything. It's what is referred to as an affine transformation, meaning that the proportions of the source and destination coordinates remain the same, however they may be scaled up or down, rotated or translated. The real danger for localizations for most surveyors is how is the vertical being treated (tilted plane) and how is the scale being determined (hold Grid to Ground factor or based on Helmert determined scale). This is why it was always preached to work inside the box, but if a user is careful in evaluating and manipulating these values (if his software allows it) then the "inside the box" limitation need not always be a concern. A Helmert 7 parameter transformation determines rotation in 3D. We're familiar with rotation around the up axis (horizontal plane) as we do this all of the time. Helmert also solves rotation around N-axis and E-axis (pitch and roll). When and how to use these values (or not to use them) is important.

Mighty,
You are correct. This isn't "new". I don't know the exact year that Helmert developed his transformation process, but I suspect over 100 years ago.

It isn't very well understood by many surveyors who use localizations. I used RTK for a long time and avoided localizations because I was almost burned by one due to my own ignorance. Sine I knew I didn't know what I was doing with localizations I avoided them for years. When I started consulting for Javad, this was no longer sustainable and I went to work educating myself about what a localization was and the mechanics of them. This was why I included it in the training event. We discussed the general mechanics of a localization (find centroid, determine least squares best fit for scale and rotation of each point to the centroid by comparison of the two systems), what to expect from the residuals with any number of points, how to evaluate and manipulate the parameters, the impact of a tilted plane created from poor geometry, point selection and how to spot outliers.

Shawn Billings, post: 432462, member: 6521 wrote: Just to clarify what Nate was saying.

A localization (or Trimble calibration) is a comparison of two grid systems (2D or 3D Cartesian coordinates). The math is based on a Helmert 7 parameter transformation, which is a least squares approach to determining the best translation (NEU), rotation (around NEU axes) and scale (7 different parameters) between two sets of grid coordinates that represent the same points. The Helmert transformation doesn't require any geodesy. A Helmert transformation can be made for any 2 grid systems. For geodetic purposes, one grid system needs to be a projection, the other (local) system, however, is typically unrelated to the Earth (or is treated so by the localization process). The use of a localization is to determine the relationship of the local system (unrelated to the Earth) to the Earth by means of common points expressed in a known projection. The result of a localization isn't necessarily a new projection but a transformation (or set of instructions) to get from the local system to the projection system.

As Lee mentions, it's possible to not use a common projection, such as State Plane, and set a controller to geographic (LLH) coordinates. But (again as Lee points out) a bit of slight of hand is taking place behind the scenes. The controller is creating a projection at the project site and projecting the LLH coordinates to this intermediate projection, then the software is comparing the local coordinates to the grid coordinates from the intermediate projection to determine the transformation.

As Lee also mentions, TDS was a grid based software with GPS support added later, so the database was not geodetic. Carlson is very similar. The point database is grid and a user must select a grid system when creating a new project which prevents reprojection on the fly.

As to Mighty Moe's comments about warping. Technically a Helmert 7 parameter transformation doesn't "warp" anything. It's what is referred to as an affine transformation, meaning that the proportions of the source and destination coordinates remain the same, however they may be scaled up or down, rotated or translated. The real danger for localizations for most surveyors is how is the vertical being treated (tilted plane) and how is the scale being determined (hold Grid to Ground factor or based on Helmert determined scale). This is why it was always preached to work inside the box, but if a user is careful in evaluating and manipulating these values (if his software allows it) then the "inside the box" limitation need not always be a concern. A Helmert 7 parameter transformation determines rotation in 3D. We're familiar with rotation around the up axis (horizontal plane) as we do this all of the time. Helmert also solves rotation around N-axis and E-axis (pitch and roll). When and how to use these values (or not to use them) is important.

mmmmm, let me see, scaled, rotated, translated, pitched and rolled all at the same time,,,,,,I would say that warping pretty well describes that process,,,,,,,

One of my problems with doing these is that there is only one place to find the new projection parameters and that's in the DC that does them, I've never seen anyone publish the projection parameters for a calibration, so unless they can send you a copy of the DC file, the best that can be done is have the calibrator send the lat, long, height, N, E, Elev for the points and do an office calibration and there are issues doing that. Of course for some control it's the only choice, but for most control it's very overdone, often cause the calibrator only learned how to do it and wasn't taught all the other options.

Anyway I'm in agreement with you Shawn, but I still won't stop despising calibrations when there are better and more proper ways to use existing control.

Shawn Billings, post: 432462, member: 6521 wrote: The math is based on a Helmert 7 parameter transformation, which is a least squares approach to determining the best translation

That's slightly confusing because the transformation itself doesn't involve least squares.

They probably use least squares to find the best transformation, but that's not strictly required, and might not be done if a minimum number of points were used to define the transformation.
------
I wouldn't use the term warping for the Helmert transformation because as Shawn said, everything is proportional. Warping to me implies some nonlinear process, like an SPC projection where the scale factors vary all over the map.

For a localization (or calibration) it's actually even deeper than a projection definition. That's only half of it. There is also the transformation that sits on top of it. So you have the projection (datum, projection type, defining constants such as latitude and longitude of origin, grid origin, etc.) and then you have the transformation (rotation, translation, scale and centroids). I've never seen these values supplied on a plat or report either.

Now for our (Javad's) pdf reports, this data is included and perhaps other controllers can provide an exportable report of this information as well, but I've never seen them presented by a surveyor as part of their product.

My typical use for a localization is to get a good estimation of where the old system fits on planet Earth. But there may be times that I need to go from a proper projection to the old system (Lee's plant coordinate system is a good example). In either case, it's a good idea to make sure that the underlying coordinates are good geodetically (CORS based rather than autonomous, or Here, based), and to have a report that can be reviewed later.

Bill93, post: 432471, member: 87 wrote: That's slightly confusing because the transformation itself doesn't involve least squares.

They probably use least squares to find the best transformation, but that's not strictly required, and might not be done if a minimum number of points were used to define the transformation.

I suppose that's true. Translation does not require least squares. It's based on the difference of the two centroids regardless of the number of points. With more than 2 horizontal pairs of points, least squares is used to define the rotation around the U-axis and scale (residuals are zero with only two points). With more than three vertical pairs of points, least squares is used to determine rotation around N-axis and E-axis (U residuals are zero with only three points).

I've had too many issues with calibrations to want to ever use them again, there are usually better options than using them.
One is simply to take the existing control and shift it into a good system.
Rotate and translate the old control to a new GPS set up system.

If for some reason you can't do that, you can almost always incorporate the old control into a known projection. This is normally pretty simple, not as easy as setting on 3 or 4 existing points and pushing a button but it's way better because you have parameters that anyone else can use.

Maybe the biggest reason not to use a calibration is the vertical control. Modern Geoid models are so much more accurate than any calibration can hope to ever be, this wasn't always the case but Geoid 12B is not the same thing that Geoid 90 was.