What is "localization", exactly?

In GPS terms, "Localization" (also called Site Calibration, Transformation, or other terms depending on the software) is the mathematical process used to relate local grid coordinates to the GPS earth centered, earth fixed ellipsoidal datum. The actual process involves collecting points with coordinates in each system and then allowing the software to calculate (usually 7) transformation parameters between the two systems.

> In GPS terms, "Localization" (also called Site Calibration, Transformation, or other terms depending on the software) is the mathematical process used to relate local grid coordinates to the GPS earth centered, earth fixed ellipsoidal datum. The actual process involves collecting points with coordinates in each system and then allowing the software to calculate (usually 7) transformation parameters between the two systems.

Lee,

I doubt most localization utilize a 7 parameter transformations but rather simple translation, rotation and scaling. That is why you should always 'localize' about points on the outside of your local project. Rigorous 7 parameter transformations are usually reserved for transforming from one well defined geodetic system to another i.e. WGS84 to NAD83.

We often use the terms "localization" and "calibration" in the same manner.

We have assigned coordinates to every section corner in NAD83 on a local ground projection. Say we wanted to tie in the center of a particular section. We would "localize" or "calibrate" to the four corners of that section by using the known values. This then gets us on the same local coordinate system. When the center of section is tied in by GPS, the value we get is then in direct relationship to the other corners of that same section.

My understanding of localization or site calibration is this:

Say you have a list of monuments with local, assumed coordinates (hopefully good ones) on monuments or control points. You don't know how these relate to the geodetic reference system (e.g. latitudes, longitudes & ellipsoid heights/ECEF coordinates). You go to each point with your GNSS rover and using good methods obtain good geodetic coordinates. In Trimble controllers you pair each point's geodetic point with its grid point. The software reverse engineers a map projection complete with scale factor and mapping angle. Then it computes a theoretically perfect grid coordinate for each point and compares it with the grid coordinate you entered, the difference is the residual.

If you local, assumed coordinates are really state plane grid then your scale factor and mapping angle should look very close to the state plane scale factor and mapping angle at your project.

Originally the idea was you could start your base on a here position (not very accurate), go tie your known control in your local system and go survey with your RTK and generate positions in your local system. This can be dangerous particularly if a new "here" is done each day but the same localization is used.

7 point localization is a bit higher order work then I generally need to do, but otherwise you are dead on.

John -

By 7 parameter I meant translation in x,y,and z; rotation about x, y, and z; and scale. But you're right; this is from the GS manual:

If you do not know the map projection and datum transformation parameters when creating the job
and defining the local coordinate system, specify No projection / no datum.
Then specify whether grid or ground coordinates are required after a site calibration. When ground
coordinates are required, you must specify the project height. In this case, the General Survey
software performs a calibration that calculates a Transverse Mercator projection and a Molodensky
three-parameter datum transformation using the supplied control points. The project height is used to
compute a ground scale factor for the projection so that ground coordinates are computed at that
height.

From "Tom's practical glossary of Surveying terms"-
Localization is a form of mathematical violence wherein the error of an existing control network is compounded by the error of a second set of observations, usually performed with inferior methods and a complete disregard for optimal network design. Miraculously the end result is usually published a minimum of one order higher than the underlying control...

I'd listen to the above guys. They are all right, and there are lots of guru's here who stand by the "stay on grid" motto. Ok, but it does get complicated.

The very second you decide to localize, or in my world calibrate, you are no longer in the GPS round world. You are creating a flat plane on an assumed datum. I do it all the time for breaking down a section, or a lot survey. Not so sure if I'd want to go there on a 20 mile route survey that is requiring SPC's though, or even a 6 mile square township. But it is a viable option for some projects. Ask yourself if your level or total station cares if the world is flat? Or round? Or some other shape?

Here in sunny AZ I'm typically working in the 3,000 to 5,000 ft elevation range. The difference from grid to ground can be significant. Live by a coast and I don't think it's as severe.

The critical point is to recognize the significance of that "tool" in your truck and it's limitations, and the liability you take on by using it (and relying on it).

Good luck grasshopper..;-)

:good: :good:

It's a black box way to set up an LDP with unknown parameters used by Charlie Trimble to sell enough GPS equipment to clueless users and become a billionaire.

Localization can be used to determine a best fit projection to coordinates in an unknown coordinate system tied to an underlying geodetic coordinate system. Localization can also be used to determine the reference station coordinates based on observations to known coordinates in a known coordinate system, developing a dx,dy,dz for the base position. This is useful for applications where you may start the base with an autonomous position and correct that position with ties into multiple known points.

Seven parameter transformation is generally correct, however one can generally select between over determined and rigid transformations. Over determined includes the observed scalar difference between the observations and best fit adjustment. Rigid body determines translate and rotate only. Rigid body results in higher residuals between observation and adjustment, but preserves the definition of the linear unit (a foot is still a foot, rather than basing the definition of a foot on observations to the control coordinates).

The newly created projection from a localization can be set a scale factor close to one at a project elevation (a quick way to developing a low distortion projection). It can also include a rotation from North to match a record bearing or mimic the convergence angle of an existing projection. The projection type will vary with the software (oblique Mercator, stereographic, etc.)

There are a lot of moving parts in a localization, which make comments like Bionic's above justifiable, because localizations done haphazardly can result in a real FUBAR scenario. It's very much like a resection with a total station. If you know what you are doing and pay attention to what the software is telling you and you have good observations and good control coordinates to work from then a localization (or resection) can be very useful. How often this is the case, I can only speculate.

It is GEOMAGIC!

:good:

Well, more simpley a one point localization translates you to a calculated coordinate system. A two point translation translates and rotates you into the coordinate system's basis of bearing. A three point translation will scale you into the coordinate system. A four point translation will rubber sheet you into the coordinate system and as someone else stated all hell can break loose here if you don't have good residuals.
Some claim a two point translation also scales you into the coordinate system, but I haven't seen that a scale factor is being applied. Trimble claims they use an adjustment routine, but its unclear to me how they can do a viable determination of scale with only two points.

Martin,

Surveyors have embraced the term Localization over Positioning. These things happen. It makes a certain amount of sense, really. Richard "Sinc" Sincovec used to refer to points converted from Grid to Ground, but before "truncating", as Localized Coordinates. After truncating (translating, actually), he called them Project Coordinates.

Dave

:gammon:

The dangerous part is good residuals may mask a defect in the assumed grid coordinates.

My posting is somewhat belated and others have many good things to say about "localization." My comments are somewhat abstract and more theoretical but, I believe, can be very useful. See link on my web site.

Yes the joys of surveying in an area under the elevation of 100'.:-D