Accuracy GPSMAP 67

This reminds me of a 3 stooges skit where Moe and Larry's watch had a different time, Curly's watch had stopped altogether.  They couldn't figure out what time it really was.  That's what happens when you put consumer grade devices next to each other, they show different values.

Make sure all of your devices are using the same settings and have the same view of the sky.

Why does it matter that these are close, instead of exact?

Even 3 gps units the of the  same model would probably read different tracks. Take one gps unit of any model and try it on 3 consecutive days and the readings would be different too. At one time I had two 67i units and could actually see a difference in the map redraw rate between the two, very subtle, but noticeable. All settings the same. One was Moe and one was Larry.

Correct, every GPS has its error. And even a GPS can have a bad-hair day (Curly). 

I've got quite some experience with it. (I did some comparisons in the past before.)

What I notice is that Fenix and Edge are consistently in close agreement and the GPSMAP consistently is not. 

The GPSMAP receives more satellite systems, is multi-band and has the best antenna. It is the only one that indicates its error and it is about 2m. If it consistently shows deviations throughout the track of ~12 m  is not what would  expect.

So I am wondering: do other users notice this? is there someting I can do about it?

I did another test. I set recording of the GPSMAP67 to interval, once record every second. I cycled a short loop on the road 5 times. I made sure I followed the track on the road with an accuracy of about +/- 0.5 m.

red: Edge1040
blue: Fenix7
green: GPSMAP67

In this situation the result is more like what I would expect. No substantial deviations between the devices noticed.

n three locations I measured for each GPS the distance between most outying lines, which is a measure for precision. Position 1 is under trees, position 2 also and position 3 is more clear sky. The smaller the distance, the more precise is the GPS.

In this mode we see the error well within +/- 2m.

In this small test we see the Edge 1040 and GPSMAP 67 to be most precise, whereas the Fenix has some deviations.

Question is whether the intial problem is caused by the auto recording mode or whether the GPS just had a bad-hair day...
To be continued...

The use of a specified interval is what you want. All the time, every time. In particular, you want to stay away from "auto" (or whatever Garmin calls it on any particular device). In "auto" mode, the device magically determines when it makes sense to log a point. In my experience, the magic seldom works well.

When comparing devices, also need to be cognizant of the sky view enjoyed by each device. The Edge is likely on your handlebars (mostly good sky view). The Fenix is likely on your wrist. if you are cycling, this is likely pretty much equivalent to the Edge.

But where is the 67i? On the top of the bag behind your seat? Clipped to your belt? In your pocket? The point here is that your body is mostly water. It does a fine job of blocking GPS signals. This (and the need for a good sky view for iR messaging) is why the conventional wisdom is to wear inReach devices high on your shoulder.

Edited to add: There are lots of variables related to GPS "accuracy" or "precision" (lots of confusion among terms). The quality of the fix depends heavily on the configuration of the current constellation overhead, as well as on the presence of things like cliffs, steep hills, urban canyons, your body,. etc. (all of which can block a portion of the sky, as well as contribute to the possibility of multi-path reception, which can seriously skew the reported position).

Remember that the figure that the GPS reports (1.5m or 2m or 7m or whatever) involves a PROBABILITY. "Your actual position is within x meters of the reported position with a probability of x%." Garmin will not tell you if x is 90 or 50 or what. Conventional wisdom is that it's somewhere around 85%. Point is that no matter what it is, there is a probability of 100-x% that your actual position is farther than that from the reported position.