Another Garmin Rant (Read 1057 times)

Well, that's 6.6% high, which is well outside of what you should expect: Normally, the GPS data sort of self-corrects as you go -- positional inaccuracy is not cumulative, and errors don't compound the way they might with a footpod-based device.

My experience puts an error of 6.6% far above what I normally see -- but not outside of what I have ever seen. Trees, hills, curves, and even clouds can cause problems for the Garmin. Trees, clouds and hills block satellite signals which then reduces the Garmin's locational accuracy. Hills and curves make getting the location more important to producing an accurate distance estimate also reducing the Garmin's ability to correctly measure distances traveled. While vertical locational errors will cancel over a run, horizontal measurement errors do not. That is horizontal locational errors are cumulative. Imagine you run a perfectly straight line for a mile. The best estimate has you exactly along the line. However, due to locational errors the Garmin will believe that you ran a zig-zag route from one end to the other. Each zig and each zag adds to the total error. There is no cancellation. The only good news here is that the triangle formula a^2+b^2=c^2 helps out a lot. You actually travel along "a" and that is what you want the Garmin to produce. But it thinks you went along "c." (In this example the Garmin got the vertical distance traveled right. This is immaterial as vertical errors do cancel.) So, the error here is the "b" term. If you do a few examples you will see even seemingly large errors are not so bad. Suppose the Garmin has you running 200' (the "c") but is off at the end point by 40' (the "b"). How far did you really go? Answer: 195.96' and error of only 2% too long even though the vertical error was a whopping 25% of the distance traveled!

While vertical locational errors will cancel over a run, horizontal measurement errors do not. That is horizontal locational errors are cumulative.

Just to be clear, you are calling the direction run as vertical, and side-to-side (with respect to the runner) as horizontal, right? In which case what you says makes sense.

It also depends how good of signal/how many satellites it was picking up. My handheld navigational Garmin actually tells you on acquisition what ther +/- error is according to current signal quality; the running versions don't! For example, last time I booted it up I had clear sky and an error margin of 86 feet! The time before it was only 15.

Time to get to know your Garmin. The running versions do tell you the +/- error. Its buried in the menu's under Navigation. You can also change your data fields to add the +/- accuracy to add it to the main screen. Thats what I do, and I don't start a run unless its synched up and measuring less than 30 feet. I'm willing to make a bet that Jim didn't have his accuracy that low when he started his track workout. Not quite enough satellites synched up. Or maybe there are some trees at Jim's track? When I run on the track, I usually get between 0.24 and 0.26 per lap and nearly always 0.25 unless I had to pass someone on the curve. Yet, when running on a track I use my Garmin as if it were a stopwatch because I know it has its accuracy limitations. I hit the lap key every lap and look a the lap time to gauge my pace. Also, since I use my Garmin in miles, yet the track is measured in meters, Garmin measurements are off by 2.25 meters a lap anyway. The nice thing about the Garmin is I don't have to keep track of the laps of a long workout in my head. If it reads between 3.4 and 3.6 miles, I know I've done 14 laps.

Isn't that pretty much what globule said?

"...it draws a straight line between the points, making it look like you cut the corner." I read that as meaning less accuracy.

"...it draws a straight line between the points, making it look like you cut the corner." I read that as meaning less accuracy.

GoBlue needs to clarify what he was trying to say. This is what I think he was trying to say: (which I agree with and qualified earlier) "If there are less samples (aka-Smart Sampling) then it will draw a straight line between points making you look like you cut the corner. " MODIFIED: Ive come to the conclusion that I have been staring at my computer screen too long today, basically I believe I may have confused what GloBlue refers to as "its" . I have my P's and Q's mixed up. Sorry Jeff.

Please stoppy bad talky about Garmin! I just (finally) ordered one yesterday. I know it won't be perfect, but hey, the idea of relevantly accurate distance info (how accurate do you think your car odometer is?), heart rate info and skin cancer protection for my wrist was enough to sell me.