VRS Observations Procedures

The attached spreadsheet allows you to copy and paste coordinates for three separate VRS occupation, then it automatically calculates the absolute precision at the 95% confidence level.  It sounds like you feel that the VRS is not accurate enough, but you're unsure of how to prove it mathematically.  If you're interested, click on J7, note the 0.5 value used in the formula, then read a bit in the attached PDF to try to understand why that value is used (then explain it to me 😉). Note that the spreadsheet must be modified if using more or less than three observations.

If you use Trimble Access, PM me and I'll send you a style sheet that allows you to do the same thing without having to manually enter the individual coordinates.

I'd caution against assuming you know more than your PLS or that your understanding of what is most important aligns correctly with his.  I fell into this trap due to youthful arrogance, and the excitement of learning higher math and wanting desperately to apply it. If the accuracy difference between base/rover and VRS would fit on the head of rebar or pipe, does it matter?

Regarding field standards, I think a 180 second burn with two more thirty second burns is all that's needed assuming the field tech looks at the deltas.  More important than that is knowing what constellations your VRS is capable of receiving and providing corrections for.  If most of the CORS stations are still only on GPS and GLONASS, you might want to do three 180 sec burns.  Regarding the TS setup, it will depend on how you process it in the office.  Using Trimble Business Center, it's ridiculously easy to process it via least squares and blend TS and GNSS.  Using StarNet, it's also easy but requires a bit more thought in coding/descriptions.  It also depends on whether you're trying to stake out items that day, or just locating features.  If the latter, I'd do one burn on two control points, set the TS on one, backsight the other, then reshoot the BS and rename the point in a manner that makes it clear that this was located via TS not GNSS.  I'd then use this point for subsequent BSs then merge it with the GNSS later in the office. If it's easy, I'll close a traverse, but with modern equipment and software it's quite efficient to get enough redundancy to run least squares and just report relative positional accuracy.  Since positional accuracy is the only acceptable way of reporting accuracy for an ALTA, it's just easier to get in the habit of using it.  It's also easier to explain to clients because I can just show them that the error between any given set of monuments is smaller than the last segment of my index finger.

Good luck, keep being curious and asking questions.

Wow, thank you so much for all of the information. That spreadsheet is awesome and seems like a quick and easy way to do some testing. I'll take a look at the .pdf as soon as I can. A lot of the sites we do are pretty small and might not necessarily need to be "adjusted" through least squares, but I'd like to get familiar with the software to run an analysis at a minimum. Of course ALTAs are a different beast. We unfortunately don't use trimble equipment here, but maybe one day! For those small sites, without doing any post-processing, do you think my method of averaging the coordinates before setting up the TS will work, or is there more to it than that?

I agree with being cautious about thinking I know more than the LS. I'll admit that I certainly don't know more than him, and have a lot of respect for his knowledge and experience, but he never went to school for surveying and seems to rely on me when it comes to the higher math / adjustments. I'm not sure he's ever done a least squares adjustment and has relied solely on closed loop compass rule. I took one college class a few years ago on it. I feel quite a bit of pressure to get this figured out, but I just know it will be worth it for me in the long run once I'm licensed.. I've considered working somewhere else where I can receive the mentorship I so desperately am wanting, but no matter where I am, I'm always loyal to my employer and take extreme pride in my work, which is why this is slowly killing me, LOL. 

It sounds like you're only concerned with horizontal position on relatively small projects.  If so, be careful not to try to apply whatever you settle on to large-scale vertical jobs.  (See NOS NGS 92 for more detail.)

Posted by: @jim-frame

↑

It sounds like you're only concerned with horizontal position on relatively small projects.  If so, be careful not to try to apply whatever you settle on to large-scale vertical jobs.  (See NOS NGS 92 for more detail.)

that manual can get you there for sure. In a VRS setting you need to have a wider gap in time between observations not shorter. You need 3 to 4 hrs. Now if you are only concerned with horizontal that vertical can tell you a lot. Trust me on this I watch my vertical  more than horizontal because if my vertical is close to good enough I know my horizontal is. Unless they don’t plumb the rod etc. Vertical is your weakest chain. If that’s good you can almost always bank on horizontal. That NGS pub 92 has some very nice information in it. That guides you through. It’s not a salesman it scientific and it works. The webinar NGS just hosted GNSS best practices this week once it’s on the NGS website watch that.  It will help. Now it’s not full constellation but gps + glonass. The publication and that webinar. I use a combination of all. NRTK and base rover. NRTK to attach to NSRS and such. Base rover to build up a network for redundancy and least squares. 

Posted by: @olemanriver

↑

It’s not a salesman it scientific and it works.

Well, maybe.  I ran a large project using the 3 cm RTN guidelines with an R8-3 (GPS + GLN only) on a Trimble VRS, and a lot my sessions failed to meet specs.  Wide-open skies at all 60 or so stations.  When I tried to backstop the RTN observations with PP obs, OPUS-Projects kept blowing up.  Despite a lot of well-intentioned assistance from NGS staff, I finally had to abandon OP (and bluebooking) and processed/adjusted the data myself in TBC and Star*Net.  Given the many unbillable hours I spent trying to make it work, it was a disappointing experience overall.  🙁

Maybe a receiver with more constellations available would have helped, but I didn't have one to experiment with.

P.S.  I now know a whole lot more about the GVX file format that I ever desired to know.

@jim-frame I am with you on that. So many other variables that come into play with GPS. Solar activity. Then with VRS or any network RTK solution. It’s all about how well that is maintained and designed. How stable the stations themselves are. We had a few base stations in our network we just flat out had to relocate. We could see the movement especially in the high temps and lower night temps. I have seen stations as I traveled over the years and could literally watch them move in the wind. There was a station in a network in one state that was an old line rod the old candy cane ones with zip ties to som e small conduit pipe. Latency on one or two of the stations in your area at times of survey. Antennas going bad. We had some of the older zephers literally crack and split apart. Latency on the cell network so many factors. I like to have a couple 180 seconds shots 4 hrs apart. And set a base on that and log static data. Hopefully 4 hrs until I figure out the network in my survey area. Oh the station spacing is another factor. I have followed the old gudelines turning 16 sets and still bust. It’s not 100% but it gets you most of the time to success. Not always for sure. But going less doesn’t help much either. I have developed my own checks utilizing the guidelines over the years and it gets me where I need to be most of the time. Now whenever I have done jobs for NGS specs I most definitely follow the guidelines to the T and sometimes two observations are just it enough so 3 or 4. Was you going for relative or just to be on datum fgdc specs. As they don’t do very well on the relative side.

Posted by: @olemanriver

↑

Was you going for relative or just to be on datum fgdc specs. As they don’t do very well on the relative side.

The original plan was to bluebook it as a 3 cm ellipsoid height project.  Station spacing averaged 7 - 10 km, with a few longer reaches.  I started with 3 each 5-minute RTN obs at each station, with 3-hour separations and at least 2 different days.  But a substantial percentage of them had PAGES residuals that exceeded the max, so I started adding obs, but still couldn't get all the residuals into spec.  That's when I started adding PP obs, and OP just couldn't handle those in the mix.  (The comment of one NGS staffer about the OP project was "everything causes it to crash and none of us know why.")  I beat my head against it for weeks, but finally had to deliver to my client so I took the homebrew approach.

@jim-frame yeah I bet they are scratching their heads.  I had my crew chief do a couple 4 hour plus sessions on a big Topo we were doing a couple weeks ago. He set up base to log as he RTK and such set more RTK controls he was there over the course of a week set up on 3 different points all tying back to his other RTK controls two of the three had good redundancy from opus and the other had 1 3 hr session and 1 2 hour. As I drug the opus results in to TBC horizontal was tight compared to RTK redundant shots several over that week also hit tight with a good traverse as well. But the opus results were all over the place way out of what I normally see in the vertical. I re processed them straight from CORS. Myself and got good results using the same CORS as opus used. Which doesn’t happen often but sometimes it does. I was scratching my head. My precision results from opus and such looked good just bad day I guess. Or days. It is frustrating at times but I try and remember I could be traversing miles just to get to a site . I re sent the same data to opus yesterday on a few files and dang if it was not spot on. Gremlins making me pull my hair out.  I do my best to set are projects up as close to NSRS as possible with what I can budget and such then it’s total focus on the relative for our work. Usually a couple 4 hr sessions while doing base and rover at a few points during the survey works well but not always. I sometimes miss doing the very tight tedious work and dotting my I’s and crossing my T’s like your project but then again I love the boundary work now days so much more. You were going for absolute accuracy to the datum then vs relative differences and that a different animal than every day land surveying for sure.

@murphy 

Posted by: @cb1997

↑

@murphy 

 

Hey! I just wanted to give you an update, I started having the guys set 2 or more points with rtk, backsight and store the backsight point as a new number to use going forward to keep the total station measurements "relative". then a few hours later, before noon, they tag the points again as an "aka" point number. If the horizontal or vertical varies by more than 0.1', a third observation is required. Then we can eventually start adjusting the data in the office after getting enough check ties and convincing my boss that we need to buy a starnet subscription.

The way i see it is, if they store the backsight as a new number, then re-backsight, the TS data is all relative and essentially "trig levels" the elevations from the first point they sat on and removes the horizontal errors I keep seeing when backsighting GPS control. Especially for jobs where everything will be shot with the total station, we dont necessarily want to incorporate the GPS errors into the control network, and this method should still hopefully get us close enough to the grid when we need to be. I still am having them write down the initial backsight errors before storing the third point. Am I thinking about this right? Like I said in the post, we don't build bridges or do much high accuracy design work, but I'm hoping this will get us on a much better track going forward for maintaining accuracy.

Thanks again for taking the time to reply and for helping me grow my knowledge on how to do a survey the "right" way. 

 

Ok so let me see if I am understanding you correctly. You set a couple of points with RTK. I assist is on grid state plane not scaled to the surface. You then set up on one back sight the other and hold the total station edm distance then run your loop. Let me ask you are you holding a surface distance or scaling to the grid using the c/f . So what you are doing is if holding the edm surface distance is your basically saying the point I am occupying is golden and the bearing or azimuth is golden. Here is what I do and I use rtk and total stations robots. Crew establishes a point via either NRTK or a HERE position just a rough raw position. For base and rover. We use Trimble so I am logging data at the base while setting other control points around the site. Not all are inter visible not needed. I try my best if possible to use first point as close to center of my project site. Not an exact science just guess. So all the base rover rtk vectors are radial to that base. We locate all property corners all even in canopy. (Bare with me on this) the first round of observations is to walk the lines looking for evidence finding the corners setting some additional control that is off site. After they have done this they have a pretty good idea of what is there what all other evidence and improvements need to be located for the map poor Topo portion. And where control pairs might be needed for robot or total station work. I like that base if logging to log for at least 4hours. They might set some additional pairs to locate things like houses barns shops etc.  this usually is at minimum of 4 hours. Remember this data begins to look like the spokes on a wagon wheel.  See NGS pub 92. Hub.  Then either that day or next they move the base to the next best open sky point they set from rtk base and rover outside the property. They log data again but use the rtk point number. We use Trimble so this might have to change based on your software. The now can run back through all control and prop corners. Even the ones in canopy that they were able to get decent RMS values on. If not they have set pairs in the open to shorten any traversing needed. All a minimum of 3 minutes or 180 epochs. I have been on points 5 to 10 minutes for it to get the precision down. Like almost the 5 to 6 minute the NGS pub 92 states. Also read through the NGS RTK guidelines by William Henney good information in that as well to support this. Now I have the 2nd RTK data done I will start evaluating the data as they map and locate improvements and locate prop corners they could not get with gps. All the 2nd observations from RTK have a 4 hour gap in time. Even if it is the next day. So say Monday at 8 am they locate point 1 Tuesday they need to locate that at 1200 or later in the day. This is all done on state plane for us. I send off the data to OPUS or process it myself attic to my hub and or any other point I logged data to as a check. All the traverse point data measuring rounds we use same point number store another in Trimble access. You get to see the error as you store the 2nd or third observations regardless of what method. This gives the field folks the ability to know hey. This works or it doesn’t. A third rtk might be needed or travers into and through to a station pair. Do not run loops we don’t on these projects. We are building a network on site. Always tying things together with all measurement tools. Static RTK conventional even levels if that’s required. Pay attention to the vertical on RTK more than horizontal even when vertical doesn’t matter. If you can achieve repeatable results within close to what you need horizontal 99% of the time the horizontal is better. That’s a little secret most know. It’s a 3d vector. After all control work and property corners have been located all that data goes into TBC for a least squares adjustment. You can design some good networks and set minimum control around a site. Usually as your property corners become part of the design most of the time. And here we have some crazy shaped parcels. We usually have minimal problems meeting.07 ft + 20ppms doing this. And usually much quicker than traversing. Small sites maybe not. That 2nd round crew chiefs should got the direct route not walking the lines except for mapping all that evidence and improvements the find from first walk through. So the 2nd time is quicker. Usually at the 2nd time a 2 man crew has split one measuring barns house making sketches doing robot work for location or even a 2nd rover mapping a field etc. teach them if they do not find a monument the first time to set something within approximately close like a nail for traverse maybe 2.  I will often set a couple 5 ft offsets to where I think corner should be I observe these like they are prop corners two measurements like all other control.  If the LSA all works then depending on job and all the opus or NRTK points worked I have achieved one a good idea of how close or uncertain I am to the datum as well as proved my relative precisions needed as well. If all is good we can scale to ground if that is required or leave it on state plane. Those points or observations that fall outside of the error needed can be achieved as they map or run smaller traverse runs through them all of this is all LSA together. It will seem longer on both office and field side until they get a few under their belts and get that timing down. But once the lightbulb goes off they will smoke sites and you will have qa/qc adjustments done very quickly. It took 3 good sites where I sat the crew chief down on each and walked him through showing him the vectors and site design at all stages and naming them the same point. Again might be a Trimble thing. But now he writes down on first observation and what he thinks if a point is not gps able so I know he is running a traverse through that. I prefer one job file for control and property corners and a different one for mapping just keeps it easier to see. I keep one office project for the control and once mapping starts I import that into a new project with control and property corners that are all adjusted. And job datum parameters. Etc. then I am just qa/acing vectors and set ups rod heights etc. for that data. It doesn’t always happen that way but it makes it flow nicely and quickly. Getting the standard errors pin pointed them taking care plumbing the rods and setting up tribrach and total station over points keeping gun in adjustments makes life better. After you are anyone else does several adjustments you will see when they need to be reminded of adjusting the gun and or they are getting sloppy.  A few other things when working in canopy with gps. If it rains that day or night before don’t attempt to much in canopy until leaves are dry it’s a crap shoot. Tell them to be patient in canopy and really watch the SNR make sure to have L5 band and be tracking about 22 birds minimum. Thats how it works.  I have set for 5 minutes before initialization then go for 5 to 10 minutes during observation. Watching the snr the rms values. Now common sense has to take place if I can get to it with total station quicker then I do that. However if I have to cut 2000 foot of line and know I will have to set up several times to traverse I know how long that takes so sitting on one point twice for 30 minutes that’s a rough number of walking in and out twice along with longer RTK observation times might make better sense and I am willing to take that gamble. Now the points I set as offsets to where I think the corner should be set. This sometimes saves a lot of time and when you have to go set a corner. I can tape it in or even set it with gps and tape and measure to offsets as a Ck . They were part of my adjustment I have achieved as the NGS pub states two observations minimum time on station which we cut down from 5 minutes to 3 minimum. And time between observations. So that pub is based on gps and glonass and for datum accuracies. Where I need relative and an idea of how close to datum I am. Use the 4 hr gap and longer observations for NRTK a lot can happen in that arena. Redundancy redundancy redundancy works wonders. We have every bird in our newer receivers and would not attempt canopy without that. Some jobs we never have to break out a total station or robot on. If I used the other brands I would set up and test how I could achieve something similar with their field and office software as well. Javad has a very good program built in that can achieve the relative precision needed and those I know that use it in canopy state the same thing set it to what desired results are and let it cook.  Don’t rush it. Sometimes a few minutes sometimes longer. Star net is a great LSA package you would have to not use same point number on but could still build a network with rtk and static and conventional even add levels if required. Learn all your tools and strengths weaknesses and use them to achieve what is required. Redundancy is key with any of them. Instead of just staking out to a tie or Ck point have crews observe them. As the project goes on so you might get more observations that aid in tighter control before the project is finalized for drafting and delivery. Once it’s done though hold whatever control as gospel before any design work is done. Because you have nailed it all down. Sometimes a good set up gun on the actual property corner in canopy and resect off of a few points set in open from the RTK works better than just occupying a point back sighting a point and turning and angle to the corner. For better stats . When using LSA. Because whatever error is in those two points is being expanded.  I always compile my deeds adjoining and subject property ROW rough it into based on eyeball imagery gets it close enough to look at in google earth as a kmz to help point crews to best open areas for base set ups and good control in and outside the site along with the property corners to aid in a decent network design.  But google earth in our area is not always clear for trees tall buildings power lines.  So the crew needs to aid in keeping a good strength of figure even if a point is way off the property to round it out. They can drive to it twice. Hope this helps. I do have some. Jobs smaller where we might set base up locate all corners once then set up gun on a point turn rounds to all others same point number and not rtk twice but I have redundant check by robot and the on datum has the same two minimum 4 hours session gap in time for opus etc. larger sites like thousands of acres more static is done and rtk to achieve things. 

I'm not following along with this thread yet so forgive me if this is completely irrelevant, but I watched an NGS webinar recently that got into best practices for GNSS survey and tangented off on the differing methods when using VRS.

https://geodesy.noaa.gov/web/science_edu/webinar_series/gnss-best-practices.shtml

They also pushed the relevance of info in the NOAA NGS 92 paper

https://geodesy.noaa.gov/library/pdfs/NOAA_TM_NOS_NGS_0092.pdf

@dave-o  That was a great webinar. The wife said it was one of the most attended ones in a while.