Related
Options

Running Distance / Pace calculation is flawed / needs rework

Adrian s
5 months ago

Hello, A while ago, I reported that my Fenix 8 watch under-reports distance by a large (from a training athlete's perspective) amount, resulting in wrong running pace reported. 

This affects someone who is actively training both during running - pace is under reported, as well as post training and in planning. 

On today's 7km run - as reported by the Fenix 8 - the recorded GPS track distance was actually ~7.55km. So the distance is actually being recorded correctly, but the watch applies (as I've learned in the previous post) some corrections / calculations based on stride length and cadence, to figure out distance, and purportedly ignores the GPS track distance. 

Garmin has also recently removed the ability for users to specify stride length manually - so you are stuck with what the watch "thinks" is your stride lenght based on accelerometer data, etc. 

Now - to me as a runner who is actually training - distance is distance, based on map data and gps track length - not a calculation you are stuck with and have no control over. In today's run as reported by my F8 my running pace was 5:05min/km over a distance of 7.02km. But based on the recorded GPS track and MAP data - I had a running pace of 4:44min/km over 7.55km. These are two VERY DIFFERENT running paces / running zones, pertaining to DIFFERENT TYPES OF RUNS for DIFFERENT END GOALS within the training block. They cannot be used interchangeably!

Which one should I - an actively training runner - take as correct & accurate when running and planning my training block, with my top-of-the-line flagship sports watch? 

Has anyone ever found a fix or workaround to have the ACTUAL track distance/map distance reported on their activities instead of useless calculations? 

Also a note to Garmin - shouldn't you put a better use to your "athlete intelligence" technology - to analyze post-activity data and apply corrections based on hard facts (map topographic and survey data for distance / elevation) and IMPROVE such a CORE FEATURE of your devices like Running, instead of using it for shiny "golden badges" and an "intelligent insight" that tells me that my recent stress of 23 is lower than my average past stress of 25??? Slight smileSlight smileSlight smile Current features like HILL SCORE and ENDURANCE score are utterly useless (algorithms are wonky and elevation data gets skewed by the slightest change in ambient pressure or temperature)

Why are athletes penalized when this is a "SPORTS WATCH" and we are getting newer and fancier "smartwatch", "wellness" and "social" features instead of focusing on having a ROBUST and ACCURATE training experience on a thousand dollar Flagship sports watch??!! 

  • 0 1 month ago

    Today's training session:

    10k base run with 3x4min tempo effort intervals, then the rest of the run at tempo pace. 

    Distance as reported by the activity: 10.01 Km. (I already knew it's going to be lower than the true distance, since at the [measured] 3km mark of my run the watch was showing 2.7 Km)

    GPS Track distance: 10.74 Km. (verified by recreating the track with Garmin's course builder and checking the recorded GPS track in google earth, both distances were spot on).

    AVG cadence: 180 (On flat terrain I basically run at this cadence all the time, only varying stride length to increase/decrease running pace).

    Initial AVG running pace: 4:53 min/km

    Corrected AVG running pace: 4:33 min/km

    I planned my 3x4min intervals of tempo effort to be around 4:30 min/km, but instead they were actually around 4:15 to 4:20, resulting in overworking. The whole run ended up being more or less around my planned interval effort. RPE was skewed since it was really hot outside, so hard to gauge, but relying on the "false" data displayed by the watch resulted in overworking today. 

  • 0 1 month ago

    Had an interesting observation these days. I've ran on the same route on two consecutive days. 

    Day 1 - "Tempo" run - at the 3 km point on my route the watch showed 2.65 km. (avg pace 4:27min/km)

    Day 2 - "Easy" run - at the 3 km point on my route the watch showed 2.95 km, surprisingly close!. (avg pace 5:10min/km)

    Conditions were the same each day, clear sky, almost same cadence - 180 on day1, 179 on day2. 

    Taking a step back it almost seems like the watch gets confused on tempo efforts and varying stride lengths, since the cadence was virtually the same. 

    We can speculate the running dynamics calculation algorithms are tweaked for slower runs? Which would be weird if you asked me. 

  • 0 1 month ago in reply to Adrian s
    Adrian s said:
    We can speculate the running dynamics calculation algorithms are tweaked for slower runs?

    The algorithm is based on the avg. stride length, which is then multiplied by the cadence, and there is an additional factor taking in respect that the stride length is not a constant, and rather variable (longer at high speeds, and shorter than average at lower speeds). However, the speed/pace is not used directly for the calculation (since that's together with the distance what the algorithm is actually trying to calculate), and rather the acceleration peaks are used to detect your effort. So in praxis, the more vigorously you swing the arms, the more the algorithm increases the stride length (hence the speed and distance). And vice versa of course too. So it very much depends on each individual, on his running style and on other conditions (incl. the strength and quality of the GPS signal). If you are lucky, you may get quite accurate results, but I saw deviations exceeding 10% in other cases. Using a chest strap usually improves these errors significantly.

  • 0 1 month ago in reply to trux
    trux said:
    So in praxis, the more vigorously you swing the arms, the more the algorithm increases the stride length (hence the speed and distance).

    I think this might not be valid for me, I don't have a huge arm swing, but rather keep them tucked and only move them with my torso, which also moves 'as needed', slight rotations left and right. Sometimes I run with the arms tucked on the torso and the forearms slightly extended for balance, but again, the swing is minimal and it's the same at most 'steady' paces. I think there might be more to Cadence tracking than arm swing, since the watch has several accelerometers. 

    I've worked extensively to train body movement (including arm swing) and cadence so the body is as efficient as possible when running, part of an ample plan for ultra distance running efficiency. My cadence is almost always the same (180) without thinking, and I'd say arm swing is almost identical on every type of run except heavy sprints or intervals where ample arm swings are biomechanically necessary. But I usually focus on HR rather than pace or distance for those more niched workouts. 

    trux said:
    Using a chest strap usually improves these errors significantly.

    Unfortunately that's also not the case. I have 3x HRM PRO Plus (which I rotate and keep in different bags) and a HRM RUN (older model) which I use on important workouts where HR accuracy is crucial for tracking effort / planning subsequent efforts. I use the strap in 'dumb mode' (pace/distance only set to indoors) and I experience the same issues with under reported distance. If I use pace/distance 'always' from the strap, I end up with way over reported distance, sometimes 1-2 km extra. I've tried deleting calibration data and recalibrating countless times, at different stride lengths.

    You have some good observations but I think the issue might be more complex, unfortunately. 

  • 0 1 month ago in reply to Adrian s
    Adrian s said:
    I think this might not be valid for me, I don't have a huge arm swing, but rather keep them tucked and only move them with my torso, which also moves 'as needed', slight rotations left and right.

    And that's why you get the distance underestimated

  • 0 1 month ago in reply to trux

    Why do you think that is? 

    I had the same 'slight' arm swing in both runs mentioned before.

    I've experimented with different types of 'arm swings' and the results are always the same. Even full on arm swings like when walking.

    Also note that the watch records flight time, derived from individual steps from Cadence. That also has nothing to do with arm swing, and might be a more plausible source of deriving the stride length. 

    IMHO arm swing would be a very crude way of estimating stride length, and AFAIK is used in cheaper watches / 'fitness bands'. That would be a huge fail from Garmin. 

    (BTW you can run with your arm straight and still have all running dynamics, try it!) 

  • 0 1 month ago in reply to Adrian s

    If you could post links to the concerned activities, or the fit files, it might reveal some clues. Anyway, if you get different distance deviation at high pace and at slow pace, then there is some disproportion in the algorithm evaluating the accelerometer data and combining it with the GPS data. It is clear that Gamin's algorithm does not fit everyone, and can cause quite significant deviations from the true values. Practically, the only thing that works relatively reliably to mitigate the problem, is using the HRM-Pro or HRM 600 chest straps (or the discontinued Foot Pod). Those eliminate the inaccuracy added by the complicated arm movement, making it more difficult to estimate exactly the stride length.

  • 0 1 month ago in reply to Adrian s
    Adrian s said:

    Day 1 - "Tempo" run - at the 3 km point on my route the watch showed 2.65 km. (avg pace 4:27min/km)

    Day 2 - "Easy" run - at the 3 km point on my route the watch showed 2.95 km, surprisingly close!. (avg pace 5:10min/km)

    Conditions were the same each day, clear sky, almost same cadence - 180 on day1, 179 on day2

    This is a clear proof that the distance doesn't come from GPS alone. When you ran slower you did more steps since the cadence - number of steps per minute - was the same but a slower run took longer. It is clear that Garmin considered your step length to be the same in both cases, and that is why the onger time translated to the longer distance. 

    I wonder what satellite mode do you use and whether distance accuracy would change of you force the most accurate All satellites + multi-band mode? 

  • 0 1 month ago in reply to silentvoyager

    I always prefer accuracy to battery life and I have set the watch for the best possible accuracy - including all + multi band gps and every second recording. But this is not a case of GPS accuracy, since the GPs tracks themselves are always spot on with the ran distance. The issue is with the corrections that the watch applies to the accurate GPS distance recorded, resulting in lower activity distance. 

  • 0 1 month ago in reply to trux

    The HRM pro plus bands that I have all give out +1-2km when used to track pace and distance.