Satellite display on 66i

You can call the top one calculated horizontal accuracy if you like. As usual, the number is tied to a probability. "Your true position is within x feet of the reported position with probability p."  And Garmin isn't going to tell you what probability they use. Smaller is better. You can't compare them across different models, even from the same vendor. Certainly not cross-vendor.

I believe the bottom one is elevation.

I agree with twolpert.

Discussed in another forum in relation to 66sr and one participant obtained this response from Garmin support:

"In regards to your enquiry:
We typically quote the 2-sigma (95th percentile) accuracy value.
So when we say your accuracy is 3.65m, we are saying that 95 percent of the time, you would be within 3.65m of the true location."

Legacy devices traditionally applied CEP 50%.

Interpreting that quote calls for someone with a background in statistics and geometry.  If someone reading this has that kind of education, please chime in.  I would like to know more about what the accuracy number means.  Putting aside spatial position, I assume (though without much knowledge of GPS geometry) that a standard deviation in the measurements must vary a lot depending on the axis of the variance.  In other words, I see no reason to believe that a standard deviation of 10 feet along say the northeast-southwest axis would be the same along the northwest-southeast axis.  Interpreted literally, the 3.65m in the quote would mean the maximum of all axes?  But the fact that the quote doesn't address the issue makes me suspicious that it's strident.

First time I've ever seen the number quoted. I would have expected 90% to 95%. Not sure what that equates to as CEP.

There's a few industry standard methods, a couple quick comparison summaries I just looked up that explain the differences: 

https://junipersys.com/support/article/6614

gssc.esa.int/.../Accuracy

the term 'accuracy' is in fact incorrect, as this should be used only when comparing a measurement to exactly known and defined value.

As a GPS receiver alone can not know such defined vale, we can not speak about accuracy.

We might rather use the term precision, which compares number unknown values, as position calculated by the GPS receiver with other unknown values, again positions calculated by the GPS receiver just statistically. This means the result will give you approximate calculation of probability how close will be the next measurement to the previous measurement(s), not any kind of map position!

However since the term accuracy, thought misleading, find its way into consumer world, meaning all the smartphones etc, it is used incorrectly here as well.

Note that GPS receiver is not able to tell you how far his calculations are from any kind of 'true' position, even if people still keep dreaming about this.

Using that argument Precision wouldn't be correct either unless you physically plotted each and all the points on the ground so they were also known and then did the measurements and calculations from them. 

In this case the accuracy being referred to is not literally accuracy, but probability of accuracy, 95%

Non corrected devices may not have a true ground reference but they have been extensively tested under various signal conditions and carry a reference model that tells them under the current operating parameters what the performance is.

For a consumer device with meters to spare in any direction I wouldn't be splitting hairs and loosing sleep over it.

do not what argue too much, but precission is the correct terme, as the results of individual measurements are collected and from this the potential dilution of position can be permanently calculated and updated. So measurements are compared to previous measurements. So yes, the device physically plots the data over certain time to be able to use precision term.

And as this time is not infinite, it says: if you make a measurement and calculation within next defined time, your result will be probably 6m of the currently calculated position.

In smartphone field, to make the picture more pleasant to the consumer user, 4 calculations are made, from them an average calculated and this will be then passed to temporary data base as entry for calculation of the potential dilution of position etc. (thats why such system may not be used for any serious navigation)

Accuracy term can not be used with any simple GPS not ground reference based.

As it is not know to what the measured value can be compared with. GPS receiver does not know where it is, so it can not carry out measurements leading to accuracy.

We can start talk about accuracy with ground based differential GPS possibly, when we clearly define what is the point we are referring to.