NTRIP. What does the recommended 6 miles really mean?

GPS/GNSS Development Boards
1

So, I’ve been using my pair of Facet’s to survey/measure property lines in rural Texas. Cell coverage is very spotty, but I’ve managed to use one Facet as a NTRIP client on a PPP-processed +_2mm ground reference point I defined. I’m getting a large error in my measurements vs the 2 paper surveys I have in hand. >2 foot difference over 1200’ marked distance from Facet vs professional survey distance.

The setup looks like this:

Facet PPP NTRIP server -->wifi to house -->WIRELESS internet service provider ( I think their tower is ~5 miles away but they’re based near Dallas 80 miles away) → Internet → EMLID NTRIP service → Internet -->Cell tower–>my iPhone as a hot spot → Android device using SW Maps bluetooth’d to Facet NTRIP client.

Physically, the two Facets are never more than 3/4 mile from one another. Electronic communication-wise, though, that’s a lot of hops, distance travelled via radios, and distance via wires and fiber. So how does this relate to the recommendation for the NTRIP client being placed 6 miles from the server? Is this where my error is coming from?

I’m going to try measuring with the Facets using radio. This is less than ideal, because I the devices can’t reach each other through the woods. My idea is to create multiple PPP locations and move around that way due to the radio los limitations.

Thoughts?

2

While I’m not certain, that would be my guess (latency issues + curvature → inaccuracies); run through our accuracy guide https://docs.sparkfun.com/SparkFun_RTK_ … ification/ and see if you can achieve similar results

Has the accuracy of the reference point being used been verified?

3

Your setup sounds pretty sound. I doubt there is significant latency in your radio link (cell phone->hotspot->etc). As long as your getting RTK Fix, the GNSS receiver is happy with the age/validity of the RTCM corrections being received. To put it in perspective, I’ve seen a receiver drop RTK Fix to Float when the RTCM data was 15+ seconds old. If you’re getting regular 1Hz delivery of 1s old RTCM (I bet your better than 1s delay) then you are likely seeing good RTK Fix.

I’m with TS-Russell - check your coordinate system. I am not a surveyor, but when folks being to talk about how inaccurate their readings are, it’s often due to incompatibilities between reference frames. All things SparkFun RTK is in the WGS84 coordinate system. Perhaps your professional survey is in a different system.

Please let us know if you figure it out! We’re here learning as well.

4

Good points above about coordinate systems. When I hear about errors around a couple feet, that’s the first thing I think of, because the difference between WGS84 and NAD83 is in that order of magnitude. What datum are you using?

We’re all assuming that your rover is actually providing an RTK Fixed solution. :slight_smile:

How are you calculating the distance between your two points using the GNSS receivers?

Did you try setting the base at a near by random location, letting it survey itself in, use your rover on your two points, and calculating the distance between those two sets of coordinates? This might help sort out datum issues. You don’t need the PPP solution for this.

Did you try getting PPP (eg CSRS) or OPUS solutions for your two points and calculating the distance using those coordinates? That might help eliminate RTK communication issues.

You could also get a 200’ tape, layout two points on level ground, use your rover to get coordinates of both points, calc the distance with the coordinates, and compare.

Are you using Lat-Lon, UTM, or state plane coordinates?

What’s your elevation? Don’t forget distances between GPS coordinates are on an ellipsoid (“geodetic distance”) or a projected plane (aka “grid distance ”) that is somewhere down around “sea level”.

Though 2’ in 1200’ is more than I’d expect from elevation issues. But I’ll go on anyway.

GPS measurements of distance rarely equal the distance you’d measure with a total station, chain, or tape. Rays starting at the center of the earth and passing through two coordinates on the grid get farther apart when they go through the ground surface at an higher elevation.

So, if you’re at any elevation, the distance between the points on the ground will be longer than the distance of the points’ coordinates down on the ellipsoid or grid. If the surveyor used a total station, chain, or tape to measure the distance, the distances will be ground distances. You need to calculate the elevation factor to convert grid to ground distances or visa versa.

You might need to ask your surveyor if their distances are ground or grid. They can help you convert ground to grid. There might be other corrections too.

5

Here’s a link about elevation factors and the other corrections I mentioned above that might be needed.

https://www.e-education.psu.edu/geog862/node/1815

6

One thing to add, if you’re calculating the distance between your two sets of coordinates using ECEF XYZ coordinates and Pythagoras, you’d get the distance of a straight line between the two sets of coordinates. It would be a chord vs. a tape on the curved earth surface, but it would eliminate elevation factors.

We’re also assuming you’re able to accurately locate (center) your GNSS receivers over the two points, and that you’re using the same points that the surveyor measured. It’s not uncommon for a corner to have a hard-to-find marker on the actual corner and a more obvious monument 2’ away that is only a witness to help you find the actual marker.

7

Ok. Great responses. I’ll do some testing this weekend to see if I can address some more of the suggestions. I’m rained out for a bit.

I did verify a few of the points with the Facets connected via the Holybro radio link vs WiFi/Cellular NTRIP. They were within a few millimeters of each other for the points I could reach with the radio. I’ll try between the two PPP points as well to confirm this weekend. It feels like the hardware is working.

Regarding GPS coordinate systems, does it really matter here? The metes and bounds survey has a quadrental direction and a distance in feet. So if I went to Mark 1 (iron rod) and Mark 2 (iron rod), the GPS calculated distance whatever coordinate system I’m in would still be the same as long as both GPS collected coordinates were in the same system, no? Perhaps I’m having a fundamental misunderstanding.

8

If your base is on a random control point, and you use your rover on your two iron rods, your coordinates for the two iron rods will be in the same coordinate system, and the distances should be good. You also should be able to put your base on one iron rod and the rover on the other iron rod and get good distances, but that assumes there are no errors in setting up your base. Roving to both iron rods eliminates a number of possible errors.