Localizing on a random point

I doubt you'll get flamed here, but I'll try and simplify from my also limited expertise about localizing a project (calibrate in Trimblese).

First off, once you decide to calibrate you are now on an assumed datum, which you want to be ground values and not grid. I always get off of SPC and use 10K, 10K and a close elevation to where I am.

Horizontal. You need 2 points for a rotation. Pick any 2. I like the longest line of my project where the distance fits ok. Then when you go back, bingo - you're on the same page, and/or your survey matches record data. Pick 4 points if you want adjustment warm and fuzzies, but you're still kidding yourself.

Vertical. Pick 1, and check into more. Or run OPUS. Now they have OPUS Projects and that is very cool, so check into that after you take the seminar.

Many here will never localize a job. They work on grid and know what in the heck is going on. Or their local community has a reliable RTN set up. I'm too old for new tricks and none of my projects are very big, so I think in terms of ground (we're around 3,000 ft plus a few more K here, so the scale factor can be a big deal)

Good luck. When in doubt, get the total station out as a check....;-)

First, I would never localize, but that's just me. If I did do it, I would only do it after locating the existing control using static sessions, then look at the existing control and decide if it's really possible. DO it once for a project and then, never, ever, ever do it again.

I would also only do such a thing if I'm trying to tie into old conventional control, if it's GPS control then there is no reason to localize, and no there is no way to do it to a random point. If the random point is tied into the existing control and the localizing was done correctly and the random point was set correctly then it should "fit" into the existing localization.

If I understand the term, Trimble calls it calibration and I haven't done that in years and years.

Much better to project to a known defined system with a good Geoid Model and then check the control to that. Any GPS control should be on a defined system already, of one kind or another, you just need to be sure you know what it is.

See if there are any elevations issues and take care of them by adjusting your numbers to hold what you are after.

Part of the issue is that Localization really performs two operations.

One, it creates a projection and additional rotation, scale, and planar tilt (if more than one vertical constraint is applied).

The second operation is determining a translation offset when using a known coordinate system.

If an operator is only concerned with One, then he could shoot a point on his project which would have a Lat/Long (based on an autonomous position most likely, although not necessarily) and then assign that point 5k,5k or 10k,10k. With only one point, the rotation would be zero at that point (geodetic North), scale would be unity and not planar tilt (possibly apply a Geoid and even a vertical offset if a better elevation were available).

So in answer to the question, yes, a localization could be done on a random point for a purpose. I'm in the camp of strongly disliking localizations, just as I strongly dislike modified state plane systems. But I recognize these solutions exist and can be implemented with good results within certain limitations.

I always localize to at least 4 horizontal points, not two, and usually use a geoid model and one point for the vertical. I also check a couple of other known benchmarks for the vertical.

In theory you can do a one point localization. If you use a geoid model, it would probably be alright. The key word is probably. There is no way to check it. An easy way for me to understand it would be like collecting topo with a total station setting up on one point with no back site. If your zero holds and the gun stays level, you could topo all day long and it would be good, but would any surveyor do such a thing.

Its been a couple of years since I've even localized. As I establish more control points in areas I work I usually work in state plane. It seems like the only time I need to localize is on a construction staking job where they have done some strange conversion from state plane to ground and I need to hold their control.

In my opinion, the first necessary step before any kind of localization can be attempted, is you must first establish the relationship of that point to the rest of the job.

You are right. If it doesn't have a coordinate assigned to it how can you localize to it?

This is a timely thread for me. I don't have much GPS experience but I do have some and I've been reading and studying over the last several months trying to get up to speed.

I am planning to do some construction layout on an L shaped 300 acre tract with the control points being the property corners. Control and topo and have been done by others. It was surveyed with GPS then scaled to ground.

Some of you expressed your dislike of localizations. In this case I don't think there is a reasonable alternative is there?

What is the optimal number of points to localize to? In the past I always thought you needed to bracket the work area. Not true? Sounds like I can use 2 points or I can use all 7. What is the best practice?

What about vertical localization? Seems to me my best option is to do a one point vertical localization at the official project benchmark and check in to the other points. But then how is the curvature of the earth dealt with? The project spans a mile so I need to consider it. What is the purpose of the tilted plane option? Is it a way of dealing with curvature?

To scale or not to scale? I think it's a no-brainer and I should scale to fit the control and if it's close enough to actual ground distance I should accept it. No other good option.

Any advice will be appreciated greatly.

Williwaw, I hope I didn't hijack your thread. This seemed better than starting a new one on the same subject.

I’m not sure what you are asking specifically but localization is a great tool and has been used by all my previous employers and I use it regularly for my projects.
First as you know many projects are hard to access so you want to get on-site and accomplish as much as possible before having to travel back to civilization. I am assuming the project requires recoveries so build an AutoCAD file of plat/ plats you are recovering with your favorite coordinate basis. Go to field and recover one corner. Localize yourself to the AutoCAD file (coords to that calced) and try and recover another corner now that you have navigation. When you get to that point localize again for rotation. Hold scaling to 0 unless you want to be surveying on some odd unit of distance measurement.
Tie all the remaining corners you need and before you leave field static a few points (if you require state plane or wgs84 for project) to reduce when you get back to office. After you have coords from static translate job to that system.

Another reason is that in my area we have many vertical datums to consider. I localize using a min of four points for a good vertical on a job. Points should be established with level loop so you know you are correct.
One good piece of advice is only use one localization file on a job or you could really screw things up.

Maybe it's just me, but on a project of that size I'd run my control with GPS for economic reasons to establish horizontal, then go check the elevations with my robot. Who cares about a couple decimal point on the ground, but don't put a puddle anywhere.

All other work would be done with the robot. I don't really use GPS for construction staking work, but do set control and check the grades with Mr Robot. He is my bestest of friends, and hopefully some will agree. He tends to be smarter than me.

I don't think if you have random, planet zeiod or even planet neon datum it matters. You need to always have 100% of the time be back on the same page you started with. Surveying 101.

Kinda like sticking with the girl you took to the dance.... smile and move on.

I am planning to do some construction layout on an L shaped 300 acre tract with the control points being the property corners. Control and topo and have been done by others. It was surveyed with GPS then scaled to ground.

Some of you expressed your dislike of localizations. In this case I don't think there is a reasonable alternative is there?

I would hope whoever ran the control would have metadata for the control, lat, long, ht, x y z or how it was put together, take that and imput into your computer file, load it into your data collectors and go set on the control and check them. If it was done correctly there should be no calibration needed

:good: +1

I'm not disagreeing that localization is a great tool, if used properly. I inherited hundreds of surveys that were done in local coordinate systems, that I return to years later to get started on new projects. I'll typically recover some control from the original job and start my base on that point. At this stage I'm relying on geodetic bearings, which may not coincide with the bearings that particular job is oriented to, but close enough to use to find other control. After recovering enough original control, I can perform a localization or calibration or whatever and I'm off and running with computed search coordinates for the particulars of the current job. I can then static in my own control and get everything into State Plane and avoid the future need for a localization. What I'm talking about is not setting up on one of those original control points associated with the datum I'm trying to mirror but setting a new point, completely random and assigning it some bogus coordinate that is in no way related to my target datum. Associating it with a WGS 84 lat/long and then going out and collecting control and attempting to localize. Last year I sent a fellow out to do a job while I was busy. Very straightforward. I had good coordinates on all the control throughout the entire project area. All he had to do was set up on one of those known points and do a localization. He didn't do that. He set a random point, made no attempt to get it oriented and proceeded to try and localize on the project control. The end result, eight hours into a two hour job, he ran the batteries down on everything trying to make it work and finally, his head just exploded. Talking total meltdown. It wasn't pretty. I'm trying to spare this guy I was debating a similar fate. I send him out on his own, he needs to understand what he is doing.

I really appreciate everyone's input on this. It's a subject I myself would like to have a better handle on.

Y'all have yourselves a great weekend. Willy's going some place there no phones or computers for a few days. 🙂

I see what you mean, setting up on a point not part of the coordinate file then doing the localization. That should work; at least I know it will work in Trimble, where the base is will be irrelevant to the localization since you're choosing which points control. If you have metadata to give you lat longs either from a file or by calculation from the xyz coordinates, then no don't do it. But if you have no lat longs, then it really doesn't matter Where you set the base, of course don't include it in a localization, let the localization determine its xyz values.