Getting Correction from Two or More GNSS Base Sensors

We tried the Trimble's VRS system which uses multiple reference stations to create a single virtual reference station somewhere 'near' you.

We found the results were all over the show and much less reliable and predictable than their iBase single base station system.

If we are 10km to 15km from an iBase station we will use that - further than that we use a second GPS as a real base station

We also do static sessions where we process against multiple stations.

We no longer use VRS at all

Without getting complicated the easiest option is to switch bases at each occupied point. To do that you have to stay connected to both bases or reinitialize and take more time. Rover set to Base A, get solution without moving rover switch to base B get a position. Move rover to second point, gather data from B then A, move again. You want the continuity of lock to A or B while moving.

Alternatively you could use one rover antenna, a splitter and 2 receivers, one locked on A the other on B. Higher equipment costs, but twice the observation solutions in the same time.

VRS algorithms vary, the best being one that uses available stations that put the rover inside a triangle of stations and compute a correction for the station location. In order to get the best correction the VRS algorithm should use at least 15 minutes of data prior to the first rover reading to get the best correction.

The best VRS routine should be like RSGPS as used by OPUS-RS. It utilizes up to 9 base locations, tries to surround the rover and uses at least 15 minutes of data before and after the rover file to get an excellent correction.

A VRS that holds all bases fixed can skew the results especially in elevation for a point that is outside the base figure. I have see this in static observations holding 2 bases fixed and had to switch to 2 completely independent and meaned solutions . In a similar light I have gotten OPUS-RS positions that were outside the polygon. By forcing OPUS-RS to use a CORS that was well positioned but more than twice as far away as the farthest initial CORS I have improved my results.

If you are paying for a VRS service you should have the option to switch and use multiple bases.

A third alternative is to store all observation files and post process against a different base or better yet a single post processed VRS position using data that precedes and follows your rover file.

More GNSS data at the base does you no good if you do not have equal data in your rover.

Once L5 is fully operational it will allow better correction data useable by you even without rover L5 capable receivers.

Paul in PA

With Trimble you have the option to transmit from your base on 1/4, 1/2, and 3/4 second delays. So you can perform what they call Multi Base by putting both bases on the same frequency but with different Station Indexes and with one of them transmitting on the zero and the other on the half. You have to change bases and re-initialize, but that's only a matter of a few seconds, and it gives you a pretty independent check on the point (I say "pretty" independent because the constellation doesn't change much in those few seconds. But you wouldn't get the same wrong answer off both bases). But you're only ever using the data from one base at any given time.

Our experience with VRS has been much better than Jim's; I can't speak to his but I can say that the LSU Gulfnet is very robust, has good station spacing, is well maintained, and has stations bounding the entire state. The only time I ever saw a major problem was when one of their hosts was uncooperative and the station had very bad latency - a friend of mine that was working very close to that station got some really bad data. But he was checking himself on knowns and so easily caught it. I believe that station has since been relocated.

As I understand the VRS, it uses the data from six (?) surrounding stations and generates atmospheric corrections that are specific to your rover location. It also creates a virtual reference position and so simulates short range RTK. But of course, there's a lot more that can go wrong in that scenario than there is in sticking a base on a known point and working in close proximity to it. The bottom line is that just like any other GPS measurement you have to check it.

I was told many years ago that GPS networks can model the atmosphere so accurately that they can be used to predict the weather.

Same can done with Topcon Magnet and TopSurv. If you have multiple bases that are broadcasting on the same freq and protocols, the software will show each base. Just choose one base, take an RTK reading , then choose the other base and take reading on same point ID. The software will give a weighted average for the coordinate and show residual errors.