Thread: Accuracy figure of horizontal position fix on GPSMAP 66 satellite screen?

Originally Posted by Bushwalker8

The answer is simple, they are two different sources or error and you need to add both together to understand the total accuracy or distance to a known point.

There are indeed multiple sources of error to take into account when calculating the distance to a known point and the accuracy of that calculation, including:

1. Current GPS reported position errors.
2. Map errors present at the time the current map location is determined.
3. GPS location errors present when the 'known point' position was calculated.
4. Map errors present at the time the 'known point' map location was calculated.
5. Map errors in all calculation due to changes (movement) in the Earths crust over time.
6. ...and the list goes on...

However, these are all referencing the activity of land surveying, and have nothing to do with the GPS receivers reported position accuracy.

Originally Posted by Bushwalker8

The device gives you one, a horizontal accuracy number based on its calculations to where it thinks it is in the 3D reference frame. The OP’s question related to this calculation to better understand the accuracy.

Yes, 'The device gives you one,' is correct.

The devices 'Accuracy of GPS' data field displays the margin of error for the calculated location 'in a 3D reference frame', a reference frame that is based entirely on it's relationship to the satellites being used to make that calculation.

After reviewing this data field on many Garmin GPSr devices, a common theme has become apparent. The GPS Accuracy data field always displays one of the following text options based on the satellite constellations in use:

1. 'GPS' (when only GPS is enabled)
2. 'GPS + GLONASS' (when GPS and GLONASS are enabled)
3. 'GPS + Galileo' (when GPS and Galileo are enabled)
4. 'GNSS' (when Multi-GNSS is enabled)

Can you see the common theme?

They are always indicating the reported position accuracy based on the satellite constellations enabled.

This data field is exactly what it says it is: "Accuracy of GPS".

This data field has no relationship to the accuracy of the maps or map datums being used.

Originally Posted by Bushwalker8

However, the device also transforms that 3D position into the 2D horizontal coordinates of where it thinks it is in your chosen local datum. But it’s often actually somewhere else because the transformation is simplistic, and it does not calculate additional plate movement over time. The known point is actually further away.

The 'GPSr' is not 'somewhere else because the transformation is simplistic'.

The GPSr is exactly where is says it is.

The map data displayed may be incorrect for the current GPSr location due to any number of potential map data errors.

Again, the 'Accuracy of GPS' date field has no relation to the accuracy of the map being used.

The [Only registered and activated users can see links. ] states:

"If you are navigating and comparing the GPS coordinates to a map, chart, or other reference, set the map datum in the GPS unit to the same datum as the map to ensure the most accurate navigation."

The [Only registered and activated users can see links. ] states:

"You should not change the position format or the map datum coordinate system unless you are using a map or chart that specifies a different format."

If we look at a Garmin GPSr that did not provide on-board mapping capabilities, we can see that the [Only registered and activated users can see links. ] states:

"You may select each datum applicable to your region until you find the datum that provides the best positioning at a known point."

Originally Posted by Bushwalker8

So that additional error needs to be added to the device provided accuracy number to give you the total distance to the known point if you want to understand the real accuracy of your position.

Once more, the GPSr reported position accuracy has no relationship to map data errors.

One must be able to understand that these are two different things. The GPSr calculated position in reference to the satellites being used will have some amount of inherent error, as will the maps used to display where that position falls. The GPSr uses the selected 'Map Datum' to position the map data under the devices calculated position as precisely as possible.

Originally Posted by Bushwalker8

So if the question was how does it calculate the accuracy, then one short answer is that it does it insufficiently because it leaves some of the error out.

Only if one assumes this value is supposed to be including map errors in the first place, which I do not believe it does, as I explained above.

The accuracy field is based on where the device it thinks it is in the frame, but it’s reported as a 2D horizontal number. The number relates directly to the ground coordinates it provides you, and importantly those numbers are in whatever datum you have chosen so the datum is intrinsically linked and a significant part of your overall error.

GNSS devices report horizontal accuracy and typically with a confidence level. A common method is the already mention CEP or Circular Error Probability 50%.

Note very carefully the word Circle, it is obviously 2D. It doesn’t say Spherical or anything else in 3D for the reference frame or individual satellites.

In other devices and software it’s much clearer that the accuracy number is horizontal, the software even draws it for you in 2D.

You appear to be confused with other things like satellite specific User Range Error which is quite different to the User Accuracy we are discussing here. Here is a simple overview with a picture that may make it clearer for you: [Only registered and activated users can see links. ]
Note that the User Accuracy is a 2D circle on the ground, around the coordinates the device is giving you, which will be in the datum you have selected.

This has nothing to do with land surveying, it’s the basic ABC’s of GNSS, maps and navigation.

You have quoted recommendations from Garmin advising to use local datums so that should tell you it’s fundamental and important because of potentially large discrepancies in error. And if you do so the Garmin will add a simple general offset for the datum to take out the worst of the error but it's definitely not precise because the datums are far more complex. I gave an earlier example NZ datum that shows they can require far more computation.

And if the datum is important then it follows that if the datum moves then that is also important.

So if you want to understand the accuracy number, the topic of this thread, then you should understand that the Garmin doesn’t track that movement of the datum over time. So it’s reported position – the location of person in the middle of the circle in the picture referred to above - increases in error over time. The plate moves, and with it goes you, the datum and the map. They are all connected, the Garmin isn’t because it's reference frame is left behind in your wake. The coordinates it’s giving you are no longer correct.

No-one has said the drift is in or should be added in the accuracy number. The point being made is simply that it's not, the Garmin doesn’t track it and so you should be aware of it.

Garmin clearly understands their datum transformation is initially simplistic, and that their device isn't tracking it's subsequent drift over time. They provide the raw data in the RINEX file to allow you to do the processing yourself to correct the position. The correction includes a far more rigorous and accurate datum transformation computation, and adjustments for drift using precise reference station data. That is the only purpose for the RINEX.

You don’t have to take up the correction option, it’s optional. But if you don’t then accept you won’t be correct and you won’t know by how much or in which direction.

The points here relate to the underlying basics and do not include the “map errors” you refer to which is an additional consideration.

And I have no idea what point you are trying to make by listing the GNSS combination selections, obviously whichever combination you have selected the device will still provide an accuracy number. And if it thinks the chosen constellations give it a better horizontal position on the ground then the accuracy number will reflect that.

You still haven't said what datum you are using.

Originally Posted by Swall

if you have any information about Garmin's "datum transformation" and stuff, don't hesitate to share this information here.

I can elaborate a little more based on my local situation here.

It’s likely the devices are populated with the reference frame and local datum parameters as they were fixed prior to the devices release and not subsequently updated with firmware. For example, our current datum GDA2020 was released in 2017 and my GPSMAP78sc last firmware update in May 2019 didn’t get it.

My GPSMAP66i did come with GDA2020 but it’s been poorly and incorrectly implemented. If you look at any waypoint and switch the datum between GDA94, GDA2020 & WGS84 it’s clear the shifts are wrong. They are kind of the right distance and direction between GDA94 & GDA2020 (bearing in mid no decimals are displayed so it’s not precise), however the relationships to WGS84 are wrong.

In this example and using UTM and you can see they have GDA94 still locked at the WGS84 reference frame origin where it was when originally released in 1994 as the GDA94 & WGS84 coordinates displayed are identical. But this is incorrect because GDA94 has moved with the plate and is the reason that GDA2020 was released to re-align with the framework origin. Garmin didn’t do that and the error shows us they have simply tacked on GDA2020 and calculate it as a simple offset from out of date GDA94, and as a result the GDA2020 coordinates are incorrectly offset at higher numbers.

It’s the GDA2020 coordinates that should currently at this time be the same as WGS84, and because GDA94 has shifted to the NE, the GDA94 coordinates should appear lower than WGS84.

So now both datums are in the wrong place in relation to the reference frame and any improved position accuracy in the 66 devices has already been blow away if you use it with an Australian map.

Another consideration is that the Garmin shift would not account for all the variations around the country. These transformations are calculated using a complicated mathematical formula (7 parameter similarity Helmert transformation), or commonly applied in software by a shift grid. However, the official shift grid has over 5 million points and takes over a minute to load on my laptop so that is clearly not happening in the Garmin. It’s more likely they are simply providing the same average shift amount to any point.

If you want to have a look at a shift grid there are examples of our local ones here: [Only registered and activated users can see links. ]

I would be careful trusting datums in your own area until you have tested them.