View source for Detailed Statistics for GPS Running Watches

The tables below are a more detailed analysis of [[GPS Accuracy]], showing the devices under different conditions. This is rather a lot of data, but it allows you to see where various devices seem to have problems. 
=Conditions=
I've divided the laps into those that go under the bridge, those that are immediately after a lap that goes under the bridge, those that go next to or under the power lines, those that turn around and then all the rest (normal). I've also included a 'mixed' category that reflects all the sections and a category for the straightest segment. The count field is how many measurements I have for that combination of condition and device, with each measurement being a quarter mile distance. 
{{:GPS Accuracy-breakdown}}
=Density Plots=
Below are the density plots for each device. 
==Mixed Condition==
[[File:GPS Mixed.jpeg|none|thumb|1024px|Density plots of the GPS accuracy under the mixture of conditions.]]
==Normal Condition==
[[File:GPS Normal.jpeg|none|thumb|1024px|Density plots of the GPS accuracy excluding going under the bridge, turn around, etc.]]
==Under the Bridge==
[[File:GPS Under Bridge.jpeg|none|thumb|1024px|Density plots of the GPS accuracy from the segment that goes under the bridge.]]
==Turn Around==
[[File:GPS Turn Around.jpeg|none|thumb|1024px|Density plots of the GPS accuracy for the turn-around condition.]]
=Theoretical and Observed Accuracy=
I calculated the theoretical GPS accuracy for the visible satellites at various different elevation masks and plotted that against the observed errors. The conclusion: GPS provides sufficient coverage to give great accuracy, and the errors are not related to satellite position. 
* The observed error is plotted along the X axis. Devices with lower error rates have fewer points towards the right of the graph. 
* The theoretical error based on satellite position, observed accuracy, Dilution of Precision, etc., is plotted vertically. 
* The Z axis (appearing to go further back) has the plots for increasing elevation masks. You see fewer points with higher elevation masks as there are insufficient visible satellites to compute the error. (The elevation mask is how far above the horizon a satellite has to be to be detected.)
* The color indicates how many satellites are visible. 
* I was expected the recorded error to be in some way proportional to the theoretical error, but as you can see, there is generally little relationship between the two. 
* The Suunto Ambit2 R, and Samsung Galaxy have slight positive correlations; when the theoretical error is high, the observed error is high. However, this is fairly slight. 
* The Garmin 620, Garmin 10, Garmin 310XT, and Polar RC3 have negative correlations, indicating that the higher observed errors are occurring at lower theoretical error levels. 
* Details of the calculation can be found in the description of my [[GPS Testing Methodology]].
<gallery widths=500px heights=500px caption="Theoretical and Observed GPS Accuracy" perrow="2">
File:GPS_ Suunto Ambit2 R .jpg
File:GPS_ Garmin 610 .jpg
File:GPS_ Samsung Galaxy S3 .jpg
File:GPS_ iPhone 4s .jpg
File:GPS_ Garmin 10 .jpg
File:GPS_ Garmin 205 .jpg
File:GPS_ Fenix 2 .jpg
File:GPS_ Garmin 910XT with Footpod .jpg
File:GPS_ Garmin 310XT no Footpod .jpg
File:GPS_ Polar RC3 GPS .jpg
File:GPS_ Garmin 310XT with Footpod .jpg
File:GPS_ Garmin 620 .jpg
</gallery>