Blue Circle

[+rogric]

Just purchased a Legend HCx and finding it works great, I have one question which is not answered by the Manual. When I zoom in a blue circle appears around the position icon. What is the circle and can it be switched of?

[+Walts Hunting]

That is a visual representation of the error radius

[eaparks]

To elaborate a little more... the blue circle correlates with your EPE (estimated position error) giving you a visual representation of the area that your waypoint's coordinates most likely are in; no guarantee, just most probable.

[+trainlove]

As an experiment, go to the smallest zoom level (or is it largest), cover your antenna and the circle should get bigger.

Make a tin foil 'hat' and the circle might go away and the arrow turn to an hourglass or something similar.

[+Sputnik 57]

Just to add for completeness, the blue circle is a function not only of the signal error, but also of the precision of the map you are using. If you have detailed maps, the circle will be smaller, given the same reception, than using the basemap, since the road positions on the basemap are not as accurate. Basically, the circle is saying, "I'm pretty sure you are somewhere inside this blue circle."

[+StarBrand]

I've found it to be roughly 2 times bigger than the reported EPE. I read it as "We are over 90% certain your true position is somewhere within this circle.

[+rogric]

Thanks to all replies, now why couldn't the manual explained that?

Thanks again.

[+Walts Hunting]

1. It has nothing to do with the map. It is computed based number of satellites, strength of signal, etc and would be the same if you had no maps. I do most of my caching w/o maps since all that matters is where the arrow points.

 

2. It is always exactly 2 times, not approximate. That is because the EPE is radius and the display is a circle and D=2R.

[+julianh]

1. It has nothing to do with the map. It is computed based number of satellites, strength of signal, etc and would be the same if you had no maps. I do most of my caching w/o maps since all that matters is where the arrow points.

 

2. It is always exactly 2 times, not approximate. That is because the EPE is radius and the display is a circle and D=2R.

The radius of the "accuracy ring" DOES include an allowance for the map accuracy. You can demonstrate this quite easily if you have access to one or more map sets in addition to the default base map. (My unit is a Summit HC - I believe these comments apply equally to all current-generation eTrex models, but I don't know about other Garmin models, such as the Colorado and Oregon.)

 

Load the highest resolution map set you have (e.g. detailed topos or street maps). Make a note of the claimed Accuracy / EPE, and the circle radius when you zoom in very close. (i.e. zoom in so close that the accuracy circle just about fills the screen.) E.g. right now, using MapSource MetroGuide Street Maps, I am seeing an Accuracy of 5 metres, while the circle has a radius of about 20 metres.

 

Now hide the detailed map, and display a low resolution map set, such as the built-in base map, or MapSource World Map. Note the EPE and the circle radius. Right now, I see an Accuracy of 4m (you get a little bit of fluctuation from minute to minute), but the ring has blown out to about 250 metre radius. With a low resolution map set, the unit adds a generous allowance for map precision, recognising that map features may be out of position by 200 metres or so.

 

Now hide ALL map sets and note the EPE and circle radius. Right now I see an EPE of 5m, and the circle radius is about 15 metres - slightly smaller than when I show my highest resolution maps. This is because with no map set to display, the only error it needs to report is GPS error, with no additional allowance for map accuracy.

 

Toggle between various map sets, and you will see that the circle radius changes to reflect the accuracy of the current map set, and it is quite repeatable. This is because the unit knows that if the accuracy of the displayed map is not very good, it draws a more generous circle to indicate my likely location AS DISPLAYED ON THE CURRENT MAP.

 

This has all been documented elsewhere as well:

 

Acc. Circle (Accuracy Circle) - This is a circle of uncertainty with regard to you current fix and its relation to the underlying maps. It takes into consideration the current EPE calculation, but mostly it considers the accuracy of the underlying map. To minimize the size of the database the underlying maps are stored with a certain amount of precision. This is why you often see "overzoomed" as a message on the map screen. The accuracy circle is another way to display this same information. If you turn the map display off you will see increased accuracy for this circle, on the III+ if you load a mapsource map for the area you will see increased accuracy, and also if you are using a dgps you will see a smaller circle. Note that it is on the line tab on the G-III.

 

http://gpsinformation.org/dale/screens.htm

Edited October 29, 2008 by julianh

[+StarBrand]

...

2. It is always exactly 2 times, not approximate. That is because the EPE is radius and the display is a circle and D=2R.

no - I meant the radius is 2 times the reported EPE - I often see an EPE of 12 feet or so an a circle around 48 feet across.

[Hertzog]

...

2. It is always exactly 2 times, not approximate. That is because the EPE is radius and the display is a circle and D=2R.

no - I meant the radius is 2 times the reported EPE - I often see an EPE of 12 feet or so an a circle around 48 feet across.

I once measured it on a 60CSx and got the best fit by assuming they multiplied the EPE by 3.4 and then rss'd in a 28ft map error to get the blue cirlce radius (this matches your data pretty well by the way: a 12ft EPE would translate to a 49ft radius). I doubt that they really go to the trouble of rssing in an error though - the math would be more trouble than it is worth for a small device. I'm guessing that they multiply the EPE by a factor and then simply add in a map error; in that case my best fit would be a factor of 3.2 and an additive error of 11 feet. I couldn't get the EPE low enough to actually see whether there is a "floor" due to rssing the errors. It occurs to me that I might be able to do some tests using the base map to generate large map errors and the power save mode to get larger EPEs; sometime if I can muster the energy I'll do that test.

 

After posting I noticed that you said "48 feet across", which would be seriously fifferent than my results (I was speaking of the circle radius)! I can only speak for the 60CSx (and earlier models that I have used), but I would be very surprised to see a 48 foot diameter circle with a 12 foot EPE.

Edited October 30, 2008 by Hertzog

[+julianh]

It occurs to me that I might be able to do some tests using the base map to generate large map errors and the power save mode to get larger EPEs; sometime if I can muster the energy I'll do that test.

Good luck! I put this one right in the same basket as other things we would like to know, but Garmin aren't telling, including:

 

1. How is EPE calculated?

2. What is the barometric altimeter auto-calibration algorithm?

 

In the absence of any authoritative explanation (which I DON'T expect to see coming from Garmin any time soon!), we just have to make use of these tools as best suits our purpose.

 

In my own figures posted last night (Summit HC, current software & firmware), with no map loaded at all (and therefore presumably no additional allowance for "map error"), I was seeing EPE of 5 m and a radius of about 15 m, indicating the circle radius for GPS error alone was about 3 times EPE.

 

When I had my highest resolution maps loaded (MetroGuide Australia v6), I saw an EPE of 5 metres with a circle radius of about 20 metres, indicating an additional "map error" of about 5 metres for this map set.

 

With my lowest resolution map loaded (built-in base map), the radius blew out to about 250 metres with an EPE of 4 metres, indicating the "map error" for the coarse base map is about 200 to 250 metres.

Edited October 30, 2008 by julianh

[+Team CowboyPapa]

 

1. How is EPE calculated?

2. What is the barometric altimeter auto-calibration algorithm?

 

Oh-oh, I learned earlier this year not to argue with Julian, so I'm passing on #2.

 

Now, for #1, assuming that the algorithm estimating EPE takes into account the map accuracy, exactly how does it know the map's accuracy attribute? From metadata attached to the map? For example the bit rate of an MP3 file where a value such as 128kbps is one of the "tags", or the pixel dimensions, such as 1200 x 1600, are attached as metadata to a jpg file? So the bit rate of a song or the number of pixels of a photo are somewhat indicative of quality, are they not? Is that a reasonable analogy?

[SiliconFiend]

 

1. How is EPE calculated?

2. What is the barometric altimeter auto-calibration algorithm?

 

Oh-oh, I learned earlier this year not to argue with Julian, so I'm passing on #2.

 

Now, for #1, assuming that the algorithm estimating EPE takes into account the map accuracy, exactly how does it know the map's accuracy attribute? From metadata attached to the map? For example the bit rate of an MP3 file where a value such as 128kbps is one of the "tags", or the pixel dimensions, such as 1200 x 1600, are attached as metadata to a jpg file? So the bit rate of a song or the number of pixels of a photo are somewhat indicative of quality, are they not? Is that a reasonable analogy?

Garmin maps have different numbers of bits assigned to the different zoom levels. The maximum is 24 bits. Assuming that means 24 bits for latitude and 24 bits for longitude, and taking a wild naive guess about Garmin's coordinate system, I think the minimum resolution would be the circumference of the earth divided by 2^24, or 40,075,160 meters / 16777216 = 2.38 meters. So that would be a grid with 2.38 meter spacing. Maybe.

[Hertzog]

In my own figures posted last night (Summit HC, current software & firmware), with no map loaded at all (and therefore presumably no additional allowance for "map error"), I was seeing EPE of 5 m and a radius of about 15 m, indicating the circle radius for GPS error alone was about 3 times EPE.

 

When I had my highest resolution maps loaded (MetroGuide Australia v6), I saw an EPE of 5 metres with a circle radius of about 20 metres, indicating an additional "map error" of about 5 metres for this map set.

 

With my lowest resolution map loaded (built-in base map), the radius blew out to about 250 metres with an EPE of 4 metres, indicating the "map error" for the coarse base map is about 200 to 250 metres.

Good point about "no map" eliminating the map error; I get the same results on my 60CSx. With the 60CSx basemap I'm getting a little higher radius than you: 320m; could be a difference in errors they assigned for the 60CSx and Summit basemaps, or maybe there are different errors for different locations.

[Hertzog]

Now, for #1, assuming that the algorithm estimating EPE takes into account the map accuracy, exactly how does it know the map's accuracy attribute? From metadata attached to the map?

They don't include map error in the EPE; it gets added when they calculate the radius of the blue circle. The map error could come from metadata attached to the maps or it could be just one number associated with each map product; I for one don't know. They may also include an error term for zoom levels, but I don't have a clue about that either. What we do know for sure is that different map products have different errors associated with them.

 

PS - Wish I had seen the Delorme thread before I went out this morning; I think I would have "wandered" by the Manhattan Beach REI store on the chance of seing the PN40.