Setting Corners with RTK

Two main issues with a single base are multipath interference and is the receiver over the point at the HI entered into the unit. Using a second base and locating each set point with one then the other alleviates these issues with the base (for the most part). However, it still doesn't mean the rover is receiving clean data. But if the point is set from one base, later checked from the second one, there is an infinitesimally small chance it's a bad location. This presupposes that the two base points are checked during their set-ups. We sometime ago drifted away from two base set-up staking, modern units have such good error trapping the bad days of a poor fix lasting for 1/2 hour as points are being set isn't a problem anymore. Not that I've seen. We did find a bad R-10 point last year, the first one ever that wasn't suspicious that we've discovered. Clearly it was a bad fix in a fairly benign environment.

If you take time to "prove" the base; check in points at the beginning and end of a session, check to all set points and some existing points, collect data and check your base with a static session, then you should be good to go with single base. 

I will run line on a section. The section corner, 1/4 and section corner will get set, then line between them run. Say the NW Section corner is 100 and the N1/4 is 101, begin a line, locate 100 with a line point that shows 0+00.02 station .01 left, you have a check on the NW corner, proceed along the line hit 101 which shows it "off.02'x.03' you know you're good. Leave enough points to check line visually and that proves the line. Visually will be good enough for fencing in the big country. 

As projects get more urban up to a downtown area full of 0 lot lines, your error budget shrinks and you may need to bring out the total station and get more precise. The RTK points will get you under a cm with care (the error statements for Trimble are good representations in my experience), but that's about as well as they do. GPS units are highly accurate from a world wide perspective, but not measurement devices to use in a machine shop.

Be sure you understand the error budget you're needing to achieve.  

I think I've been setting corners with RTK for at least 20 plus years. Even when I'm using a total station, the control that I'm occupying with it, I most likely GPS'd in. Thing to remember about RTK is that you're always going to have 1-2 cm. of error floating around which is well in excess of your 0.02'. The only way around that is to do some lengthy static sessions with multiple receivers collecting observations simultaneously and then post processing that static data, then run off that with RTK and checking in to other points in the network. Generally when I'm setting corners with RTK, I'll set a 60-d spike after first staking out to the point and take two 180 epoc observations on it with RTK, turning the pole 180 between shots and averaging the two. If the position is satisfactory, then I'll pull the spike and set the pin. I'll either then do a 10 minute static position on it, or return to shoot it again using the same methodology with a completely different satellite constellation. Two bases giving real time corrections isn't really a thing that I'm aware of, but doing static and post processing, I can have two independent vectors into the solution that provides some redundancy. I won't set corners using a network solution because I always have better options. 

Everything done in recent decades has involved black magic boxes and the belief that what shows up on a screen is gospel truth.  Never doubt the man behind the curtain!  Consistent garbage is still garbage.  Consistent truth is wonderful.  But, how do we know which is which.

Say someone worked stocking shelves in a grocery store.  They quickly learn the difference between a one-pound container and a five-pound container versus a 50-pound box.  A sense of truth is developed.  Everything that supplies numbers out of thin air is questionable as there is no common standard alternative to which a reasonable comparison can be made.  If one picks up a one-gallon can of paint, muscle memory suggests this weighs somewhere around nine pounds.  The black magic boxes, from handheld calculators to satellites, do not have anything but faith to rely upon.

A recent thread reported on EDM's providing false answers.  How does one know that every piece of data is correct?

In the survey world, I call it "Making the turd shiny".  If we implicitly believe every piece of data is true then the lesser the amount of data, the better the truth.  If we use Total Stations to assemble data around and across a mile-square section we have numerous bits of data, each with a certain amount of potential error.  Doing the same project with simple transits and 100-foot tapes aided by a plumb bob induces even greater possibility of the results being something less than truth.

I had a project once that asked for the southeast quarter of the section to be split in half and a ten-acre tract to be split out along one exterior line.  I set my Total Station on the bar at the east quarter corner and shot to the monuments at the other three corners of the southeast quarter.  Then calculated the midpoints of two lines and set monuments to split the quarter section.  Then I shot to the southeast corner of what was to be the 10-acre tract, calculated the other corners and set them.When I first told that story on an earlier version of this website I was condemned for not running a closed traverse with multiple setups.  How is that any different from setting corners by satellite ?

Posted by: @mightymoe

↑

If you take time to "prove" the base; check in points at the beginning and end of a session, check to all set points and some existing points, collect data and check your base with a static session, then you should be good to go with single base.

I wish I would've determined the base station control point coordinates with a static session. I was thinking about that this morning. The reason we're using a base is that the nearest network station is about 11 miles away and our residuals aren't that great. The base station control point was set with RTK, so I assume it throws any error into the base.

Posted by: @williwaw

↑

Thing to remember about RTK is that you're always going to have 1-2 cm. of error floating around which is well in excess of your 0.02'.

I assume it would be better to set pairs of points, static session on each point, then break out the total station. Nowadays, what is the recommended static session duration for what I'm doing?

Posted by: @holy-cow

↑

The black magic boxes, from handheld calculators to satellites, do not have anything but faith to rely upon.

I agree with you. Referring to this post, the first 2 corners I set were at a street intersection. One at the southeast corner, the other at the northeast corner. I really wanted to set up the total station, if I would have brought it, and check the distance between them.

I use static only. It was the only technology available when I started. Still valid.

Two antennas to collect data at ideally four-hour duration, and traverse through them with the survey instrument.

In New England it is hard to find spots where there is clear enough sky.

"Thing to remember about RTK is that you're always going to have 1-2 cm. of error floating around which is well in excess of your 0.02'."

Au contraire. Am I the only one who has actually taken RTK to a calibrated baseline and compared the results. Rather soothing, actually. More like to the hundredth horizontally and double that vertically. Conditions: Open skies, bipod carefully centered, five shots in quick succession for the Leica to average automatically, all shots facing the same direction.

I'm with Bruce on this one, we're lucky to be able to check mountains of DOT static run control, they're probably over the top with their surveys, but one thing for sure, dozens of control points checking less than .02' is eye-opening. It's improved greatly since the early days. I'm seeing just what Trimble declares for the accuracy of the units. 8mm+1ppm is very realistic for Trimble RTK. And consider we are usually checking into control using a range pole and standing there, no tripod, no bipod (a topo-point RTK check of 15 sec.). I've also checked a baseline near one of my jobs and it was the same. That time I used a bipod but it was all very close. This is open sky, getting results for difficult multi-path will degrade the accuracy. We don't do static on subdivision corners, it's all done RTK except control. 

Posted by: @bruce-small

↑

Conditions: Open skies

Those are essentially unicorn conditions, not the 95% of the areas I work in. Mr. Moe is in Big Sky country. I nearly always have to deal with multipath to some degree and more often than not I don't know how much error is floating around until I go back and occupy it with gun and my backsight check is off .1'. 

If you are uncomfortable setting corners with RTK, as I am, I suggest the following strategy. Set 3 well spaced control points in good sky areas with your RTK, then set up your total station via resection using those control points, and stake. Using the resection routines and 3 or more control points will tell you if there are outliers in the control. 

@norman-oklahoma I would add a fourth GNSS point, not used in the resection, as a further check

Back in the days when I was busy doing commercial properties of all kinds, like shopping centers and apartment sites, I had an efficient routine down pat: Set a mag nail at all reasonable bends in the drives around the site, with three little pink spray dots around them (little dots, mind). Collect all of them with the bipod and five shots each, averaged, and labeled CP in the data collector. That told my F2F to put that point number on the CP layer and larger than usual. Then I was off collecting the parking lot stripes, bumpers, utilities, etc. When I returned next day for the buildings I had a map showing the parking features, and the control points. Being different, of course, I used a white brick with a black vertical line for the reflectorless total station target. Shooting those control points were always on or under 0.02' difference and I'd carry on. More than that I'd go looking for my error. Some days just for the fun of it I'd record the R-M differences, and at the end of the day they invariably averaged 0.009'. Yes, I'm serious. 

Why a brick? Because that put the black line exactly on top of the mag nail, so no target error, and if stolen or run over, well there was fifty cents down the drain

Posted by: @field-dog

↑

Posted by: @mightymoe

↑

If you take time to "prove" the base; check in points at the beginning and end of a session, check to all set points and some existing points, collect data and check your base with a static session, then you should be good to go with single base.

I wish I would've determined the base station control point coordinates with a static session.

So how was that point determined?

If I'm doing a random boundary survey where I don't expect to be back to the area anytime soon I just do a here position and roll with that.  Having your base position dialed in with a static session doesn't have any affect on the quality of the vectors of the points you measure from it.

Posted by: @bruce-small

↑

That told my F2F to put that point number on the CP layer and larger than usual.

I do a very similar thing, except that the control points sort to a layer that prints red, and the boundary marks to a layer that prints blue. Check points sort to a layer that is very light gray, barely visible when printed on white paper. Got the idea from you, Bruce, several years ago.

Posted by: @bruce-small

↑

Why a brick? Because that put the black line exactly on top of the mag nail, so no target error, and if stolen or run over, well there was fifty cents down the drain

Yup. The good ol' brick backsights. I used to use a cinderblock.

@field-dog

As a fresh instrument operator on a construction site my party chief had a backsight figured for the 7th brick joint on the top row from the left corner of a bank building across the highway from the job. Setup, set 0° on the brick joint and start cranking angles. 

Posted by: @williwaw

↑

Posted by: @bruce-small

↑

Conditions: Open skies

Those are essentially unicorn conditions, not the 95% of the areas I work in. Mr. Moe is in Big Sky country. I nearly always have to deal with multipath to some degree and more often than not I don't know how much error is floating around until I go back and occupy it with gun and my backsight check is off .1'.