123

GPS accuracy? (Read 1514 times)


Joggaholic

    I am currently trying out the GPS on my droid. When I review the recorded track on google map, I notice that the path is sometimes off by as much as almost 20 ft. Here's the map:

     

    https://maps.google.com/maps/ms?msa=0&msid=200154586983648982309.0004b5954c6410a05f885

     

    I was staying on the dirt/concrete track at all times but the recorded path was sometimes going through the parking lot or the soccer field. The GPS on the phone recorded 6.3 miles, but measuring the path on google map (the measure route feature on RA) yielded 6.15 miles (2.4% difference)

     

    Is this kind of expected? (If not, this may indicate that either my phone or my app is not that great) Do the GPS watches such as the Garmins give more accurate positions? I'm trying to justify if I should buy one instead of relying on a free phone app. I have read reviews saying that the Garmin watches are very accurate, but the comparisons are usually made against another watch (ie the 305), and I haven't yet found a zoomed-in map or a number that can tell me what kind of accuracy is to be expected (+- X ft or something).

     

    Thanks

      GPS watches seem to be more accurate (at least at running speeds) than phones, but there will always be error.  There are about seven million threads on this subject if you do a search.

      Runners run.

      xor


        The GPS in your phone is great for checking in with 4square.  Not so much for super duper running accuracy.

         

        Also, what Mikey said but I say it with a funny accent.

         


        Joggaholic

          GPS watches seem to be more accurate (at least at running speeds) than phones, but there will always be error.  There are about seven million threads on this subject if you do a search.

           

          Yes I have looked at reviews but I brought up this question specifically with my map because I don't know what "good" means from the reviews. I can't tell if the jaggy lines I've gotten from my phone is "good" or "bad" as in how does it stack up against the goodness of the GPS watches. Thanks for your response. I assume with a gps watch I should see a similar map with the loopy lines more tightly bunched together compared to what I have now?

            You can zoom in on  people's maps on here to see how closely a GPS watch follows the road. My droid phone, in comparison, makes it look like I zig-zagged all over the place when I have used it for mapping a run. Once, I even wrote my name with my GPS watch after a run.


            I'm back!

              Take a look at my log here. All my runs have associated Garmin maps. Actually, 20-foot accuracy is pretty good.


              Feeling the growl again

                Yes I have looked at reviews but I brought up this question specifically with my map because I don't know what "good" means from the reviews. I can't tell if the jaggy lines I've gotten from my phone is "good" or "bad" as in how does it stack up against the goodness of the GPS watches. Thanks for your response. I assume with a gps watch I should see a similar map with the loopy lines more tightly bunched together compared to what I have now?

                 

                Any of the current GPS watches will do better than your phone for this purpose, and none will be 100% accurate.  All will be sufficient for running, and all will occasionally make significant and noticeable errors.

                 

                In other words, calibrate your expecations with what such a device can deliver and you'll be ok.  Wink

                 

                When GPS needs to deliver absolute accurace (think road construction or surveying), it is coupled with additional land-based technology to enhance its accuracy.  If you read the fine print in the instructions of any GPS device, you will see that the variation you note is within the expected error range.  This is why GPS is not used to certify race courses and it's silly when people say "the race was mis-measured because my GPS recorded 26.4 miles for this marathon".

                 

                The online mapping systems suffer their own sources of error.  No measurement is exact.  All of them you are using are likely acceptable for your purposes unless you need to certify courses.  IMHO the big advantage of the watches is increased functionality (running-related and "smoothening" to average out some of the more significant GPS errors) and convenient operation, not accuracy.

                 

                It also depends where you are running.  Plot the track run under full tree cover or especially in a city with tall buildings against a map and the results will likely be comical.  Use that same device out on an open trail system or rural roads and the results will be much better.

                "If you want to be a bad a$s, then do what a bad a$s does.  There's your pep talk for today.  Go Run." -- Slo_Hand

                 

                  Regarding GPS accuracy.. it will be getting better over the next few years. 

                   

                  http://www.kurzweilai.net/next-gen-gps-satellite-upgrade-3-feet-accuracy

                    GPS watches seem to be more accurate (at least at running speeds) than phones, but there will always be error.  There are about seven million threads on this subject if you do a search.

                     

                    While on vacation last week, I had the opportunity to run with my wife, son, and father in law.  Of course, we all have GPS watches because we're all data junkies (to some varying degree)....

                    Prior to the run, we picked the course on MapMyRun.com and setup a 10k route.  On MapMyRun, it was truly 6.22 miles.

                     

                    My father in law and I both have Garmin 305's.

                    My wife and son have the same RunLogger App on their iPhone (I think it's RunLogger, but not 100% sure).

                     

                    We all ran the same flat course route at the same time (likely pinging the same satelites...)  It was a clear sunny day without any high-rise interference potentials.  We ran different speeds, though, but on a "there and back" route, we all knew our turn around point (in this case, it was the Atlantic ocean in Ponce Inlet, FL).

                     

                    Me: 6.22 miles

                    FIL: 6.18 miles

                    Son: 6.20 miles

                    Wife: 5.85 miles

                     

                    3 of the 4 were very close and "accurate".

                    My wife ran the same route with different results. ??? She ran the same speed as my son and they turned around at exactly the same spot at exactly the same time.  I'm guessing that there's a setting on her iPhone GPS reading relating to its accuracy and pinging frequency that was different from my son's settings.


                    Regardless, just as Mikeymike and SRLopez (and others) said, accuracy is in question with all GPS technology.

                     

                    MTA: My biggest question regarding the GPS results relate to the elevation.... when running at sea level, how the heck do they calculate the 150 foot of climbing hills during a given mile when I'm no more than 10 feet above sea level at any given time???

                     

                    Cheers,

                    Brian

                    2014 Goals:

                    #1: Do what I can do. <DOING>

                    #2: 365 Hours training <NOPE, INJURED>

                     


                    Joggaholic

                      From comparing bhearn's maps and reading all your responses, I gathered that my phone's GPS was at least not too far off, though not as good as a gps watch I suppose. I'm not expecting something perfect, I was just trying to figure out where I stand now with my phone and set a proper expectation with getting a watch (eventually). Thank you all!

                         figure out where I stand now with my phone 

                        Pun intended?

                        Failure is a good place to start.


                        Just a dude.

                           

                          MTA: My biggest question regarding the GPS results relate to the elevation.... when running at sea level, how the heck do they calculate the 150 foot of climbing hills during a given mile when I'm no more than 10 feet above sea level at any given time???

                           

                          Cheers,

                          Brian

                           

                          I run the GPS system for a construction company that uses it for bulldozers and graders. We have to be accurate to roughly an inch, and our equipment does a lot of tricks to get that to happen.  So I know a thing or two about how GPS works.

                           

                          Elevation is the hardest thing for GPS to get right. An accuracy of 20 feet horizontally is pretty good, yet it may still be 100 feet off vertically. Its mostly because the satellites are more above you than to the side. Depending on the day and where your horizon is, this can be better or worse.

                           

                          It's funny to me that we care so much about 1 or 2% accuracy on most of our runs. I remember as a kid we drove the course in the car and used the odometer. I doubt we were ever that close. I found a course I ran in high school to be off nearly half a mile in 8. We just thought one course was fast, or another course was slow. We only cared about place in cross county. Times pretty much only mattered on the track.

                           

                          -Kelly

                          Getting back in shape... Just need it to be a skinnier shape... 


                          Feeling the growl again

                            While on vacation last week, I had the opportunity to run with my wife, son, and father in law.  Of course, we all have GPS watches because we're all data junkies (to some varying degree)....

                            Prior to the run, we picked the course on MapMyRun.com and setup a 10k route.  On MapMyRun, it was truly 6.22 miles.

                              

                            Me: 6.22 miles

                            FIL: 6.18 miles

                            Son: 6.20 miles

                            Wife: 5.85 miles

                             

                            3 of the 4 were very close and "accurate".

                             

                            MTA: My biggest question regarding the GPS results relate to the elevation.... when running at sea level, how the heck do they calculate the 150 foot of climbing hills during a given mile when I'm no more than 10 feet above sea level at any given time??? 

                             

                            First, MapMyRun has associated error as well....so while concordance between it and 3 GPS units indicate it is "about" 6.22 miles, mapping it online does not mean that the online measurement is 100% accurate.  Wink    (I know you knew this but lately I've been hearing more people swear that the online ones "must" be 100% accurate).

                             

                            The poor performance in elevation makes sense if you think about how GPS works and how the errors occur.  Each GPS reading will plot you at a point in 3-dimensional space.  Let's make the relatively safe assumption that errors are random; in other words, it is equally likely to plot you a bit left of where you really are as right, forward instead of behind, etc. 

                             

                            As you run, the forward and back errors essentially cancel each other out over time.  This is one reason GPS average speed readings (at slow speeds) jump around a lot at first, but are more dependable the further you go...there is more data points there to average out so the forward/back error pretty much disappears. 

                             

                            Now, the left and right errors do not disappear.  They "zig-zag" your path, and add distance.  Since the distance we run is typically miles, if accuracy is good this will add up to only a small percentage of total distance...perhaps not even enough to really notice, the difference between your FIL's 6.18 and son's 6.20

                             

                            The up/down error is like the left/right error, it is cumulative rather than canceling.  Each error is recorded as a small "hill" you ran over.  However on relatively flat courses, even a small amount of error appears significant.  If that 150 feet over a mile you mentioned had been left/right error, it only would have added 2% to your total distance -- about what you can expect from well-performing GPS.  However when you are running flat courses, 150 ft of elevation change over that mile seems like a lot because you were expecting, say 10 ft or less.

                             

                            MTA:  Kelly and I were typing at the same time....what he said about horizontal vs vertical from the satellites also makes sense to me, as even if you simply monitor elevation rather than tracking cumulative change I have found its accuracy to be so bad that I quit looking at it years ago.

                            "If you want to be a bad a$s, then do what a bad a$s does.  There's your pep talk for today.  Go Run." -- Slo_Hand

                             


                            I'm back!

                              Elevation is the hardest thing for GPS to get right. An accuracy of 20 feet horizontally is pretty good, yet it may still be 100 feet off vertically. 

                               

                              That's for sure. Check out this 1000-ft hole I seem to have run in and out of. Coincidentally or not, this is very near UBC's particle accelerator. And I usually get elevation anomalies in this spot. Hmm.

                               

                              What's even worse about Garmin-reported total elevation change is that all the errors strictly add, as spaniel says. 10 feet low here, 10 feet high there, that's 20 feet of change. This is actually different from left-right errors, which also do not disappear, but don't add the same way. If I'm plotted 10 feet left of my true path here, and 10 feet right a few hundred feet later, that adds much less than 20 feet to my recorded distance.

                               

                              So yeah -- certainly you want to ignore "total elevation change", and recorded actual elevations are also very noisy, just because of the way GPS works.


                              Feeling the growl again

                                 

                                What's even worse about Garmin-reported total elevation change is that all the errors strictly add, as spaniel says. 10 feet low here, 10 feet high there, that's 20 feet of change. This is actually different from left-right errors, which also do not disappear, but don't add the same way. If I'm plotted 10 feet left of my true path here, and 10 feet right a few hundred feet later, that adds much less than 20 feet to my recorded distance. 

                                 

                                Yes, due to trigonometry the component that adds will be a function of how far you ran from one data point to the next.  Apologies for not distinguishing this completely as I was focusing on the additive part.

                                "If you want to be a bad a$s, then do what a bad a$s does.  There's your pep talk for today.  Go Run." -- Slo_Hand

                                 

                                123