I have an established 16 mile mountain bike loop I've been riding for a long time. I recently purchased a Magellan Cyclo 315 bike computer which I load files to Strava Premium from. Everything works great except the distance data. The distance value on the computer is always very close to 16 miles, but when I upload the data to Strava (.gpx file) the distance is between 15.2 and 15.4 miles. Any suggestions on how to correct this problem?
-
Thanks for your response. Unfortunately, this computer doesn't allow changes to response resolution. I have a calibrated speed/cadence sensor, and I think that's where my distance reading on the computer (accurate mileage) is coming from. The file going to Strava may be picking up the GPS distance.– pisgahrichJun 29, 2017 at 0:41
-
Have you tried testing the device in a straight route? If you travel in straight line the discrepancy generated by "missing" points of a curve becomes (almost) zero. So, if the discrepancy persists, then the culprit is the calibration of the speed sensor. Also, testing in circuits of different length will reveal if the error is proportionally the same (0.8 out of 16 is a 5% error)– JahazielJun 29, 2017 at 4:04
-
Try loading the strava activity in SNAP strava-tools.raceshape.com/snap and see if it "snaps" to the road better. You generate a better GPX file using that site, then upload it to strava again and delete the original. GPS coverage might be spotty in a mountainous area leading to straight-lines between points, not nice curves.– Criggie ♦Jun 29, 2017 at 8:05
-
1@Criggie since we are already editing .gpx files, this would be a good time to run it through digital EPO!– Rider_XJun 29, 2017 at 19:33
-
1@Rider_X Strava rounds to the nearest GPS point which has a best case resolution of 1 second. So already, any two rides can be inaccurate by a relative difference of 2 seconds. To add to that, Strava makes no use of speed sensor or direction information - to ensure that all data is dumbed down to the same level in interests of fairness, and also to Strava computing costs down. Also, all modern GPS units use some algorithms to smooth the GPS data. This is one reason why you see tracks often heading off the road around a bend. In summary: life ain't fair and Strava isn't either.– Craig HicksJun 30, 2017 at 16:02
4 Answers
It might have to do with how frequently the computer writes your GPS location to file. To optimize storage some write out a location to file once every few seconds, but carry out its own calculations on a higher resolution dataset in-memory before purging. If your route is quite twisty this can result in a shorter distance as Strava (and everyone else) assumes a straight line between locations in a .gpx file (this is really the only workable assumption).
Check your bike computer settings, some allow you to specify the recording resolution (e.g., once per second). The downside could be that your computer may not be able to store as many tracks as a lower resolution setting (e.g., once every 5 seconds).
-
3If the MTB route is a tight technical track, even 1 second updates create discrepancy. (In really technical terrain, you get discrepancy if you ran two cycle computers one off each wheel).– mattnzJun 28, 2017 at 22:53
-
GPS precision is also much worse when there are trees and steep hills that block and reflect signal. Technical trails often have both :)– ojsJun 29, 2017 at 19:19
If you have the 315hc model with the speed/cadence sensor then there can be a discrepancy if your wheel size is entered incorrectly. Strava prefers wheel sensors over GPS because of GPS update frequency issues that the other answer notes has its own proprietary way of calculating distance (most likely also incorporating other people's data who have used the same route) that doesn't correspond to your wheel-distance or gps-distance as calculated by your GPS.
-
2Thanks for the feedback I have a calibrated speed/cadence sensor, and I think that's where my distance reading on the computer (accurate mileage) is coming from. The file going to Strava may be picking up the GPS distance. Not sure if there is a way to change the default in Strava. Magellan's software that uploads and analyzes the .gpx files (CycloAgent) shows the correct distance Jun 29, 2017 at 0:46
-
Strava will be ignoring the sensor data they use the gps data. I don't think there is anything you can do to change this. The only time they pay attentions to speed data is if there is no gps data like an indoors ride. Try using another website if you want one that pays attention to your speed sensor data. This is from my experience gained from writing an app uploading to over 10 different sites.– IforJun 29, 2017 at 16:59
-
2Strava uses GPS data. They have their own algorithm differing even from other sites. My Garmin uploads to the Garmin website and to Strava. The distances that both sites compute for one track are never identical, so is the max. speed, the average and the altitude.– CarelJun 29, 2017 at 18:07
Strava has a proprietary secret algorithm for internally calculating distance, which involves smoothing the GPS data internally, although the original GPS data points are still stored in your history. If that Strava-calculated distance varies from the device distance more than a threshold, Strava replaces the device distance with the Strava distance. The Strava distance is not calculated by mapping your course to road and measuring the distance along a road. Strava distance also does not account for elevation change - while that is negligent for shallow slopes, you will be losing ~2% of your distance will climbing a 10% slope.
Strava's distance algorithm tends to lose distance over windy roads, perhaps because it is by design blind to the actual road traveled, so the windiness is removed as noise.
It shouldn't be difficult to calibrate your distance sensor along a track of known distance. Then you can know the answer.
-
I don't doubt that Strava is doing some behind the scenes trickery in order to make all of the GPS data (of varying quality and resolution) compatible, but do you have any references for the assertions made or are they largely hunches? I ask because I am genuinely curious about the behind the scenes mechanics.– Rider_XJun 30, 2017 at 17:01
-
@RiderX - I became familiar with this because sometimes an intermediate ride on a train would be included in the distance (even though GPS was off) and sometimes not. Strava "help" explained that (including train distance) happens when Strava overrides the device calculated distance with their own calculation. Elsewhere, strava's own documentation states that elevation is not included, and that the data is "smoothed". Since they don't say it tracks to roads, so I am assuming it doesn't. However, your device may track to roads (e.g. at least Garmin Edge 705 and up have that option). Jun 30, 2017 at 19:26
There is a combination of 2 things that might be going on here. 1- The GPS trail you have has a certain degree of uncertainty (+- 1 to 10m depending on signal quality). Depending on the sampling frequency it can make it look as though you are zig-zagging and add a bit of distance vs your actual path traveled. Some algorithms can smooth this out but it is not 100% accurate.
2- Your cycle computer's readout may based on the speed sensor which uses the rotation of your wheel & the diameter of that wheel. It multiplies the number of rotations by the tire circumference in order to obtain total distance traveled. There is a setting in your cycle computer to change the tire circumference and make it more accurate. The circumference depends on your wheel size, tire size, inflation and wear. If it is not accurate, it would give you a faulty distance measurement : https://www.cateye.com/data/resources/Tire_size_chart_ENG.pdf lists some of the circumferences for different wheel and tire sizes
-
In terms of (1) if the error is random, the average distance measured should be an unbiased estimator of the true distance. (i.e., a single journey may show a different, but if you average all your single journeys your average should converge on the true distance as your sample size goes to infinity). The OP saw a consistent and persistent negative bias in the measured distance for the same route suggesting that random measurement error is not the source.– Rider_XJun 30, 2017 at 19:12
-
Agreed, I read "recently purchased" and assumed it was not replicated many times. If the sample size is large enough you are correct that this would likely not be the issue. Jun 30, 2017 at 19:20