I have a question about using an RTK (base/rover) for setting control. I understand it's not the ideal way and I would MUCH rather traverse with TS. Unfortunately I do not have the time to run a conventional traverse.
The job is a water main transmission project. The plan has control points that do not check well H.15' V.15 error. I started with the base on a control point on one end of the 6 mile long job site. I localized on plan points. I then went about 7,000' with rover and set a point. My normal procedure take 5 shots over the course of the day usually total of 15 shots sometimes more and then average. I have been seeing very good results to plan points with the base on the new point. I then had to that process 3 more times each time my process was the same when setting new point. I am now seeing .8' error when checking plan points.
I am new to GPS, I have been reading a lot and asking a lot of questions. I am wondering if my procedure is wrong and the error is on my end, or is the error with plan points?
Thank you for any help.
The problem is that you don't know how the control points were set. By a skilled surveyor using the best procedures and equipment, or somehow by some doofus. Until you have that information you won't know what to hold and what to reject.
Could possibly be a grid vs. ground issue. Better make sure you're comparing apples to apples, or, what Bruce said. Localizing from the far end of the job could also have something to do with it. I try and avoid localizing until I have a feel for the overall quality of what I'm trying to get on to. I'm not a real big fan of warping my perfectly good survey to match someone else's boned up mess.
US Survey Ft vs. International?
Step No.1 when problems are encountered. Check and then double check all of your assumptions.
> I have a question about using an RTK (base/rover) for setting control. I understand it's not the ideal way and I would MUCH rather traverse with TS. Unfortunately I do not have the time to run a conventional traverse.
>
> The job is a water main transmission project. The plan has control points that do not check well H.15' V.15 error. I started with the base on a control point on one end of the 6 mile long job site. I localized on plan points. I then went about 7,000' with rover and set a point. My normal procedure take 5 shots over the course of the day usually total of 15 shots sometimes more and then average. I have been seeing very good results to plan points with the base on the new point. I then had to that process 3 more times each time my process was the same when setting new point. I am now seeing .8' error when checking plan points.
>
> I am new to GPS, I have been reading a lot and asking a lot of questions. I am wondering if my procedure is wrong and the error is on my end, or is the error with plan points?
>
> Thank you for any help.
A good rule of thumb to use when "localizing" or "calibrating" GPS to existing control is NEVER, NEVER, NEVER, localize/calibrate to existing control until you have fully verified the quality of said control. This can be done using conventional or GPS (either RTK, Static, etc.), assuming you use proper procedures for the quality of results you need.
Also, be sure your "new" control points are not outside of the "box" of the existing control.
+1
Also, on a six mile project RTK (when done properly) will smoke a total station. If the initial control was run in with total station I'd consider 0.8' near miraculous.
Stoner, this I feel is the WEAKEST link in junior practicioners of GPS. Localizations, Projections, basis of bearings.
Don't feel to bad, to keep digging at this, until you actually understand what is going on.
It is also one of MY weak areas.
I know enough, but not enough. Enough to get by, but not enough to understand all that is involved.
If you are an acad user, think in terms of align command.
A big one to watch is an "Inclined Plane". Whereby you are have a TABLE with some legs longer, and some shorter than others.
Rotate, Translate, and Scale, all rolled into one. With one more thrown in:
Vertical component. So that you can have a postage stamp, on a basketball, that is tilted, so that one end of the stamp is closer to the radius point of the ball, than the other.
And, you can really screw it up on this area. If it is inclined very much, your horizontal goes to heck.
Until you really know what you are doing, I suggest you only hold ONE point for vertical control. This will tend to keep you away from the "big stink".
N
As I have set more control in my area, I rarely localize anymore. I usually just work in state plane coordinates. I would ignore the coordinates of the existing control and set my own or use their points with my coordinates. Then see if you can identify a problem with their control.
Another question, for a water main job are you chasing an error that isn't important?
The first thing I would have done is contacted the original surveyor and asked him to describe his methods. Until you do that you are just guessing...
> Also, on a six mile project RTK (when done properly) will smoke a total station. If the initial control was run in with total station I'd consider 0.8' near miraculous.
I call hyperbole. 8 tenths in 6 miles would be about 1 in 40k. Pretty much routine expectation for a TS.
RTK control, properly done, can easily be good to about 0.05' horizontal and 0.10' vertical. But it is so rarely properly done. IMO, anybody who accepts money to establish project control should be able to produce metadata.
Tooting the StarNet (or any other LS package) horn yet again - if it was used to establish the control there probably wouldn't be a problem with it. If it wasn't it would be a very good tool for finding where the problems are.
A lot of good suggestions so far. Of course that won't stop me from weighing in...;-)
With Base / Rover I would place the base in the center of the job. That puts your Rover within three miles of the base. Be sure to gather Static data at the base.
Take the Rover along the route, observing all control in good environments '30 by 30'. That's simply a 30 second RTK, rotate the pole and take another 30 second reading. If the radio gives you any grief change gears and gather 15 to 20 minutes of static on the point. When you finish go back to the beginning and hit all RTK control 30 by 30 again. A second base gathering static is awesome but probably overkill.
For data treatment I would design a projection that is simple and repeatable or work in State Plane.
Once data is cleaned and adjusted you have a few choices to make. It isn't terribly difficult to compute parameters relating your positions to those published and quantify the distortion between the two. Having the metadata and coordinate derivation from the original surveyor might simplify the comparison process. I say 'might' as there are some strange things done by even very smart people...
Note that nowhere in the mix is the word 'localization'. They have their place but it is rare as far as I'm concerned. In most cases they simply bury the errors of both surveyors until it's too late to fix it...
My .02, Tom
As others said, get the metadata from the source before going too far. Be careful if the builders are constructing from both end and joining at mid-point.
If you have access to Leica Geo Office, the software provides a pretty useful planning / analysis application. You can load the geoid contours within your project in the map background. It is a good way to determine the difference in geoid within the project boundary. It gives you an idea of what to expect when you review the source control.
>I call hyperbole
I posted a similar thought, then reconsidered and deleted it.
In prolonging a nearly straight line for 6 miles with proper attention to T & P settings, yes, it should be routine to get the distance that close.
But are we including the effects of projecting an initial azimuth for 6 miles? A reasonably good celestial observation for azimuth might be off that much (5 seconds). And if this was an open (one-way) traverse for 6 miles with sight lengths limited by rolling terrain, then the accumulation of angular errors could easily be much greater than that.
I came on the job after work had started. The guy that did the training for the RTK told me I should localize. I told him I want to be in state plane, he said I would be because I am localizing on state plane. The way it was explained I thought once I localize I am good to go. I made the mistake of setting points outside the localization "BOX".
When I start a new job I always run my own conventional traverse and close as I have seen a lot of busted control over the years. On this job I just don't have the time to, as I have to be with the pipe laying crew to stake each pipe they lay. They have taken a 2 week break for the holidays and are coming back Monday. I was attempting to set new control ahead and checking the plan control along the way.
Is there any way I can run new control with the RTK? Starting on a point at the beginning and running to the other end? The job is in state plane, NAD83, NAVD88. If I am understanding this correctly, the way I should do it is setting up on one known point and run the control down in state plane?
The pipe is underground and crosses many roads that have lots of utilities above and under, some are very tight. The pipe also rests on existing culverts.
Thank you all for the help, hope you all have a safe and HAPPY NEW YEAR!!
> Is there any way I can run new control with the RTK? Starting on a point at the beginning and running to the other end? The job is in state plane, NAD83, NAVD88. If I am understanding this correctly, the way I should do it is setting up on one known point and run the control down in state plane?
Yes, state plane. Dispense with the localizing it the control is state plane. Salesmen are always in love with the localizing. Few salesmen are surveyors. Localizing they understand. Projections, not so much.
Assuming that you have good(enough) sky in the places you want to set control you certainly can. There are many ways you could proceed. Without knowing anything about the specifics that you haven't shared here let me suggest a course of action.
1.Set up your base station somewhere near the center of the project. But not necessarily right in the right of way. Select a really open place in a nearby vacant lot, city park, church grounds, school yard, or the like with really open sky. It is surprising how effective it is to get the base in wide open sky. Set base to collect raw data for OPUS while it transmits RTK corrections.
2. Set your control at intervals along the project corridor -that you will later set the TS on. The sky conditions are likely to be less than optimal. Do the best you can. in some cases you may be obliged to traverse through a gap in your string. So it goes.
3. Tie each point 30 by 30, as per TheBionicMan, above. Then repeat. Just make sure that your second pair of ties occur at a different time of day. Watch that you PDOP doesn't get much above 3. Your data collector can probably apply a weighting to "average" all these duplicate shots in some manner.
4. Tie a number of the existing control points in the same way.
5. At this point you have control positions relative to the autonomous position of your base station. The next step is the adjusting. This is where having StarNet would be useful. IF you don't have it you can probably do what most people do. That is, shift the collected coordinates in autocad onto the project control. When you do that, if the control is valid, the base station position will shift along with it and come close to agreeing with the OPUS position for your base.
thanks Bill. That's kind of where I'm coming from. We’ve surveyed a lot of miles of routes for sanitary sewer and water lines over the years. We’ve had some amazing results with conventional data when we were careful, turned 2 sets, measured foresights and backsights, had the ability to use balanced shots (not always possible) and adjusted it all in least squares. The least squares is more important for the geodetic reduction in my opinion that the actual adjustment. Although I do believe that averaging foresights and backsights (horizontal and vertical) helps a lot.
We’ve also had crappy results even when turning multiple sets. The problem is, partly that errors only seem to compound in a route survey. It’s not like surveying a square where errors tend to compensate and the effects of curvature and gravity may cancel each other out on the opposing sides of the figure. And if you try to rotate the beginning and the end to a geodetic azimuth (or astronomic) it seems that the middle always goes out of whack.
My point has more to do with the level of skill and effort required to traverse six miles and be within 0.8 of where you think you are. Perhaps I'm too pessimistic, but I don't see too many surveyors out there with the desire to do all of that. That's the miracle. (Not unlike the legitimate assertion that too few do RTK surveying right).
+1
Localizing is a cool tool, but not when you already know the geodetic relationship of a set of coordinates (ie state plane).
When you localized, what residuals did you see? What rotation and scale factor did the localization apply (in other words how was the localization attempting to redefine the foot and distort the central meridian of the projection)? Did you try a minimally constrained localization and then add point pairs individually? What defines a minimally constrained localization in this case? Before you press accept on a localization, you need to know the answer to these types of questions. Generally a simple translation is all that needs to occur if the original work was good and your work is good.
RTK absolutely rocks and can provide as good as static results in seconds. Take that time savings and apply it to understanding what you are doing.
Your trainer was wrong. Setup the projection and work on state plane. By you trying to localize, you are creating the error as you mentioned .15 one day then .80 the next day.
I think I read one of your other post and you may be using Topcon gear with Pocket? If so give me a call, and I can help.
No need for you to contact the surveyor who did original control as others have suggested. You have already said you are new to GPS. No sense in you opening that can of worms at this point.
Call me any time. Number on my webpage.
LeeGreen.com
A couple of things.
Norman's route is a good idea but it is critical that you don't bust your HI for that first setup that you will send to OPUS. If you measure 1.35 but key in 1.53, that's a problem. One submittal to OPUS is a bit of open situation.
Working on state plane / NAVD88 is not a problem. Understand the effects of the use/non-use of a geoid model in the DC. Get the metadata.
Be careful not to void the existing control. That's one of the shortfall of the OPUS-era we live in. Sometimes, we have a perfectly good geodetic monument available and we choose to set our own control point / OPUS to save a 1000' walk.
It really stinks when I'm told to localize sometimes. In doing a project for a city, i was told to localize to their control network and dial into their base...the control checked really bad. At the time i was a newbie so my boss handled it. However, here I am today about to work in the same city and told to do the same thing. Do you have any advice on how to manage this?
Why cant everybody just pay us to traverse everywhere?
> Why cant everybody just pay us to traverse everywhere?
When doing control work you can get much better results using GPS. Here is a clip from the CALTRANS manual. It's based on FGCS standards and specs. You will notice that a total station traverse doesn't even meet the specs for higher order work.
