Thought some of you may be interested in these two blog posts by Rob Jordan - you might even understand more of it than I did. Plenty of time on his hands as he has even made his own units:

https://robjordan.github.io/gps-accuracy/findings

https://robjordan.github.io/gps-accuracy/

Interesting project but this is the biggest issue:-

The base maps used to create the planned route are accurate; specifically that the path of a road as described on the map accurately represents the path of that map in the real world.

Looks like he is using OSM most of which is derived from GPS data in the first place. I need to sell him some OS Mastermap that has been ground surveyed ;-)

Was the Mastermap ground survey carried out using GPS Dave?

I've got about £80k's worth of GPS surveying gear on site at the moment (Leica and Trimble). I'd say the standard error on those, without using RTK streaming, is about 2m in each axis. RTK gets it down to about 12mm.

Interesting read Blackhound, cheers for posting. Impressive what he's done with the self-builds.

In Reverse wrote: ↑Wed Dec 18, 2019 1:08 pm Was the Mastermap ground survey carried out using GPS Dave?

I've got about £80k's worth of GPS surveying gear on site at the moment (Leica and Trimble). I'd say the standard error on those, without using RTK streaming, is about 2m in each axis. RTK gets it down to about 12mm.

Interesting read Blackhound, cheers for posting. Impressive what he's done with the self-builds.

A mixture of ground survey and highly accurate GPS/groundstation devices. They have their own GNSS stations (OS NTET) getting them down to 8mm route mean square. They claim less than a few mm in a 10km square.

I think the accuracy of plotting the route is likely to have more impact than the actual accuracy of the map used.
From the image of the planned route vs recorded log it looks like the planned route follows the centre line of the road. If he was riding along the left of the road in reality, that could account for more than the ~2metre median error that most of the devices showed.

Dave Barter wrote: ↑Wed Dec 18, 2019 2:15 pm
A mixture of ground survey and highly accurate GPS/groundstation devices. They have their own GNSS stations (OS NTET) getting them down to 8mm route mean square. They claim less than a few mm in a 10km square.

Ta. Looks like the RTK network we use is piggy-backed onto their OS NET. I've just been down a right rabbit hole of links off their website. The amount I don't know for a(n alleged) professional is ridiculous. When I started studying this at uni 25 years ago we still measured with a theodolite and calculated coordinate errors longhand, which I can still do. Suffice to say I have not kept pace with the pointy end of the industry.

Reading the results seems to indicate that deviations of the recorded track are all within an acceptable margin of error whereas whenever I load a shared gpx on my etrex the deviation of the displayed track is greater.
Would this be caused by the use of different base maps/device used to create the original gpx?

Mariner wrote: ↑Wed Dec 18, 2019 11:33 pm Reading the results seems to indicate that deviations of the recorded track are all within an acceptable margin of error whereas whenever I load a shared gpx on my etrex the deviation of the displayed track is greater.
Would this be caused by the use of different base maps/device used to create the original gpx?

Yes or poor plotting with only a few track points

In Reverse wrote: ↑Wed Dec 18, 2019 4:35 pm I have not kept pace with the pointy end of the industry.

The majority of the industry is very much not at the pointy end