I would appreciate some advice regarding selecting a GNSS device. I am interested in using it to accurately record planting locations for thousands of trees. I am looking for sub cm accuracy, primarily because I am hoping to use this to assist in planting on contour. Vertical accuracy needs to be optimized for this to work. I have been using a laser level which works great, but this would could be a great time saver if the accuracy is good enough. Tilt compensation sounds like a good idea for my purpose. I will use it primarily in North Carolina, USA. I don’t plan on using a base station and I am hoping this will work with the NC ntrip service. I am very much a beginner to this technology, and I don’t have $10,000+ to spend. Thanks for any advice.
The RTK devices at SparkPNT will work with the NC CORS network , but that particular service has a 1-time setup fee of $500. There are other options available to you for RTK, such as PointPerfect Flex.
The first thing that comes to my mind for your application is performing a continuous ToPo of the property first. That’s where you mount the RTK antenna to a 4-wheeler, UTV, etc and simply drive around - collecting points automatically. Those points are later used to create contour lines that can be exported back to your application of choice (such as SW Maps, etc). Then you can look at the map and literally walk along any specific elevation/contour in the field with your RTK device.
A ToPo isn’t necessary for what you are describing, but it’s a 1-time event that you keep forever and might come in handy. I thought it was worth mentioning just in case you find it interesting.
Personally, my 2 favorite field devices at SparkPNT are the ones at each end of the price range : The TX2 ($1250) and the Facet FPM ($2100). The formfactor of the TX2 seems bulletproof for a precision device. It’s just a great unit and a workhorse. The Mosaic X5 chip used in the FPM is the best performer, but not everyone will use it in a manner that highlights the differences. The larger antenna element in the Facet lineup is also helpful.
Seems like there might be an interesting backstory with planting trees to sub cm vertical accuracy
Hi Dan, welcome! Ryan already covered the correction options and gave you a good idea with the continuous topo survey.
I’d only add a little caution about vertical accuracy. RTK vertical measurements are normally less accurate than horizontal measurements, so I wouldn’t plan around getting reliable sub-centimeter elevations every time. It should still save you a lot of walking and help you find or follow contours, but I would keep the laser level for checking the final elevations before building the swales and berms.
The setup is fairly simple, the receiver goes on a survey pole, your phone runs an app such as SW Maps, and the NC CORS network or PointPerfect sends corrections through an internet connection. That gives you RTK accuracy without setting up your own base station.
Since you mentioned tilt compensation, that is available on the -T versions of the Facet models, but not on the TX2. It could save some time when you’re collecting thousands of tree locations because you wouldn’t need to level the pole carefully at every point.
sub centimeter accuracy is just not going to happen. In striving for cm level, there are only two reasonable options. One is a professional Real-Time Network, and an example is MassDOT’s RTN, run on Leica, with NGS-class control. The other is for you to put in a local base. With the local base, you will get excellent relative accuracy, and absolute will depend on how good you get at base coordinates. But presumably you care about relative within your site.
You definitely should work in NAD83(2011) epoch 2010.0. That is fixed to the north american plate.
Using the X5 (vs F9P) should get you better accuracy, and 3 bands is also better. Even if you got 2x FPM-T (really only one needs T with base/rover), I am very skeptical of reliable sub-cm vertical. 4 cm I can believe.
It’s really too bad that the lowest cost device is $1250. The RTK Express kit used to be only $600 (F9P and antenna, bring your own pole like the others).
(Please reply in this topic and don’t start new ones on the same subject.)
If I did get a local base, how much improvement would you expect over using a service? Excellent relative accuracy would work for most of my puposes. I have read a bit about the lengths surveyors go to for absolute accuracy and fortunately that is less critical for me. I may need to modify my goals if I can’t get better than 4 cm relative vertical accuracy. The horizontal measurements are still useful, but I might be better off with a less expensive unit if it is not capable of cm vertical accuracy. Hopefully I got the reply correct this time. Thanks very much for your interest.
@DanS , a very small % of people can actually produce a sub-cm accurate position. Even when the equipment can reasonably do it… most people can’t There are dozens of things that commonly trip people up and they never even know it. I’m 30 years into this and I still make my share of mistakes, and still learning everyday.
The Good News: In my area (Southeast US), PointPerfect Flex (RTCM, OSR) can produce repeatable points to <1cm across long periods of time with a correct workflow. I state that as precision – a permanent point’s coordinates are reproducible week after week with PointPerfect Flex. I can’t swear what you’d get in NC, but I would expect similar results for a $15/month PointPerfect plan.
Ryan, Sounds like you have a great deal of experience and I appreciate you sharing a bit with me. I have a question about your suggestion to drive around with the device to create a topographic map. I can download lidar maps that approximate contours on the farm. What would be the advantage to creating my own? It would be interesting to try this, but I am not sure how it would help me. Please let me know if I am missing something. Sounds like PointPerfect is a good starting point for me to try. Thanks again for your help.
Dan
LiDAR surveys are commissioned to be used for floodplain mapping over large areas. It was never intended for your 1cm vertical application…so keep that in mind. However, it might be useful as a visual guide in the field, for reference. I’ve used LiDAR on Civil projects many times with good results and it can be a legitimate shortcut verses a topo, but it really depends on what you’re doing.
One more thing to remember about RTK is… whatever you decide: it’s not a final decision. You can start with a FPM as a network rover with PointPerfect or your State CORS. If neither of those match your goals, that FPM(Mosaic X5) can still be your rover for a local Base/Rover pair. If you eventually commissioned your own base with a 2’nd Mosaic X5 base, you will have many additional options to play with inside the configuration.
But it’s also wise to remember we are talking about natural ground shots. Having the equipment and sticking to a mm workflow in the field and office isn’t really justified for collecting points in the “dirt”.
@DanS: It’s really hard to say about local base. RyanF’s comments about having to get a vast pile of things right is truly wise. And, it sounds like the “PointPerfect Flex” he is using is a network solution with a dense base network, rather than what I expected (a wide-area solution with far less dense bases).
RyanF: I’m assuming that your sub-cm repeatability is about horizontal, and I wonder if after establishing a stable control, and setting up on it 10 times on 10 days, what the resulting plot looks like. I’m guessing you have a proper survey rod with a level, that’s been adjusted, and a rod tripod (or a real tripod with an optical plummet??), and that you then take averages. perhaps even rotating the rod and taking another average, and averaging those for your “observation”. If you are able to plot something, and say what you did precisely to get it, that would be very interesting. Not just the horizontal plot, but also a plot of the vertical positions.
A while ago, I saw a plot of RTK-derived positions from I think 5 occupations of a control, not just on the screen, but ink dots on metric graph paper. Seeing it at truly 1:1 was stunning.
I think DanS is wanting to make measurements in a much less fussy manner, and my feeling is having vertical positions within 1 cm of true (even relative true) is not compatible with any sort of speed. But I would love to hear the details of RyanF’s experience. (I vote for new thread, sort of “Trip Report from striving for 1 cm with Point Perfect”.
In MA, in addition to contours from LIDAR, the raw LIDAR is available. It’s said to be 0.5m cells, with 0.1m vertical accuracy, but it seems to have a vertical resolution of a cm or two. It’s really interesting to zoom the false color scale to only a meter or two in a swamp to see what the drainage is. Or rather, was on that day.
I have not yet tried to measure a flat area with RTK and compare to the LIDAR height.
DanS will also need to be very careful about NAD83 HAE (ellipsoidal height) vs orthometric heights (e.g. NAVD88), and my advice is don’t get hung up on the forthcoming new NSRS, as you’ll have >=5 years to migrate before you have pain from not doing so, is my unsupported-by-evidence prediction.
@gdt , I performed a lot of testing when PointPerfect started offering a RTCM/OSR product. At the time, I was also doing a lot of testing on the LG290P. The test rig was a semi-permanent antenna (tripod) outside my office. Throughout each day, I would randomly log a RTK position. I would also restart the gnss whenever I thought about it. The purpose was to try to be random. Several trials were run from weeks to months. 3D positions generally fit inside a golf ball. I published some of the results on the Sparkfun forum.
Thingstream doesn’t publish a map of their ground network or it’s density, so I try to be careful about drawing any conclusions across the US. But it works well everywhere I’ve used it in MS, AL, FL, LA, GA, AR. I don’t use PointPerfect for Controls Surveying, but it is an easy button for general purpose RTK ( in my area ).
Thanks - that is impressive, to be talking about single RTK points vs averages. Fair enough on fixed antenna - that quite properly removes setup variance from the equation.
I’m guessing the network is in EPSG:6319. Otherwise, you’d probably get a cm of plate motion over 6 months, and that would start to show up!
I can’t promise if those were single epochs events or 10 seconds of averaging. I did a lot of testing. I do remember that I could not statistically identify that short occupation averaging was beneficial. My rationale (after the fact) was that in the scope of a 10-30 second occupation, nothing much is changing with the major error sources. IE: if receiver got the AR wrong, it’s a busted position no matter what. It’s my experience that another occupation is more beneficial than point averaging, as it helps to reduce the random errors we humans love to introduce (centering, antenna alignment, plumbness, etc). I love these discussions, but we are obviously outside the scope of topo shots
PointPerfect is ITRF2020 (current epoch, quarterly updates), so it more likely EPSG:9989 ?
Agreed about short observations, mostly, except that there is also per-measurement noise on the phase. It seems that’s a minor contribution.
With ITRF2020 (not sure I understand what “current epoch, quarterly updates” means, guessing "our network isn’t really ITRF2020; instead we change the configured coordinates in our base stations to the ITRF2020 coordinates as of the beginning of each quarter), I think you’re going to see 2 cm/y velocity and it would be really interesting to establish a monument, measure it really well, and in 6 months, 1 year, 2 years do it again.
Agree that we’ve drifted, but I think reading this is also helpful for the OP to realize how hard 1 cm is, to the point where people that aren’t deep into it probably have trouble following conversations about the nuances of getting there. It’s really not helpful how SparkFun’s marketing tosses around the accuracy claims from the chipset manufacturer, without putting it into system context, or doing tests in system context. It makes no sense to talk about 1 cm and then a few minute survey in, for example.
Here’s my understanding of PointPerfect’s Flex’s Reference System (RTCM):
It’s published as ITRF2020 (current epoch, updated quarterly).
So today, it’s realization is defined as ITRF2020(2026.5).
To me, that means my official metadata to document the base or “correction” service would be ITRF2020(epoch = first day of the quarter), because that’s when the network realized the reference system to update the frame.
Again, this is just my working assumption:
Q1 (Jan 1 – Mar 31): 2026.00
Q2 (Apr 1 – Jun 30): 2026.25
Q3 (Jul 1 – Sep 30): 2026.50
Q4 (Oct 1 – Dec 31): 2026.75
As you already know, there are Pros and Cons to this. But just like any GNSS position, the metadata is the key.
Amen brother… to produce a point with 1cm precision takes a few seconds. A point with 1cm accuracy can take a month and a WHOLE lot of office time for QA/QC.
Much of your discussion is above my comprehension, but it is interesting to hear that you are pushing the limits of the equipment and sharing data on your own research. I am continuing to educate myself and appreciate listening in on the educators. For my purposes, absolute accuracy is not so important, but relative accuracy across our 155 acres is what I need. It is good to know that measurement precision is very good and can be quick, but that absolute accuracy is much more difficult.
Gdt, I am reading your comments about using the raw LIDAR data. You stated that it “seems to have a vertical resolution of a cm or two.” That sounds useful. I envision walking my farm with my gnss device on a pole, stopping every 20 feet to flag a spot to plant a tree. Tree spacing of 20 feet. I move the pole up or down slope slightly to find a constant vertical measurement. When I find that vertical measurement I stick a flag in the ground, walk another 20 feet and repeat the process. This is what I did with the laser level, but now I don’t have to move the tripod. If the gnss vertical measurement accurately is + or - 4 cm, can I use the raw LIDAR data to get greater accuracy? Thanks for your help.
I’m really not clear on what you are doing and why. I get it that you are planting trees, but a tree is such a large and irregular object that talking about 1 cm doesn’t make sense to me. When talking about where a car is with RTK, there’s the (newish) concept of “Vehicle Reference point”, sort of like “Antenna Reference Point”.. How would you define “Tree Reference Point”?
I am also puzzled by 1 cm and ground. My ground (New England, and I think it used to be sheep pasture but not tilled), is not flat at the 1 cm level. It sounds like you have land that is tilled and thus much smoother (See “Reading the Forested Landscape” by Duane Wessels), and you are trying to roughly have an arc of trees along a contour line and to move up/down slope at approximately some interval along-slope to get to the same elevation to plant. Is your ground really smooth at the cm level? What if you just scratched a little later, or put down a cm of dirt? How is what you are doing achieving your goal? It’s hard to know how to give you advice when I don’t understand.
As for LIDAR, that’s the measurement from some specific day in the past. And my LIDAR data and your LIDAR data are surely different. The thing to do is to get the data for your area in look at it. Learn how to use QGIS and visualize it. That’s a lot of work too. The idea that you can use LIDAR to find horizontal coords of planting locations and then stake those out, and get cm level agreement in vertical – does not sound super likely.
Sorry my previous explanation was not clear. Lets see if I can do better and give more details. First, our pasture is not level. After marking contours where I want to plant trees in curving rows, I will use a tractor subsoiler to cut multiple two foot deep, approximately one inch wide narrow furrows. The furrows decompact the soil and assist rain water and tree roots to move deeper into the soil. Next I will use a bottom plow to turn the grassy layer over, down slope, creating a shallow swale, about a six inch deep ditch on contour. The down slope area is now my berm, it is higher because of the six inch grassy layer I just turned over on it. I do several more passes on the tractor using a tiller attachment to widen the berm and increase its height a little more. I use a auger to dig holes in the berm every 20 feet and plant my trees. Hopefully, I will be successful and paste an image below. This isn’t me, but this is close to what I am creating. The bottom of the swale/ditch should be as close as possible to level, but me and my tractor can’t do perfect work. As long as the variability is less than the depth of the swale, water flowing down slope is captured and infiltration is greatly improved. I can also purposefully design a 1% slope off contour to gently move water out of a valley and send it to the ridges. Our ancestors did this long before we had lasers and satellites, but I like technology that makes my work easier. I’m happy to do more brain work and less physical work these days.
I don’t have a dozer, but some contractors use gps guided bulldozers on massive projects to create similar swale and berm projects at a much faster rate. I have heard they aren’t as good at staying on contour, but it likely depends on the contractor and their equipment. I have no plans to do it that way and my tractor does not have gps guidance. Hope this is more clear about what I am attempting. I already did about 5 acres using the laser level. My laser level claims to be within 1/16 inch of level which is great, but it is slow. I am hoping to do it faster and good enough with a gnss device. I am OK with being told this isn’t realistic if that’s the truth. Thanks for helping me.
@DanS - I know you said that’s not your tractor in the pic, but we can get a RTK mounted to the 3-point hitch on that ole Ford for the final grading pass
Heck, I’ll come help if you got a Massey Ferguson 135.
There are dozens of things that commonly trip people up and they never even know it. I’m 30 years into this and I still make my share of mistakes, and still learning everyday.
For my purposes, absolute accuracy is not so important, but relative accuracy across our 155 acres is what I need. It is good to know that measurement precision is very good and can be quick, but that absolute accuracy is much more difficult.
