UKH

Just back in from a fantastic trail run on the west coast of Orkney on a glorious day - picked a nice day for leave for a change.

Was trying out a reconditioned Garmin Forerunner 45 I got at the weekend.  At the moment it’s just the watch in its own as my phone is too ancient to support the app to analyse/upload any info.

I set it running as I left the carpark, alongside Strava on my phone, which is my usual method of measuring speed and distance.

Theres a huge discrepancy between the two.  By 4km in on Garmin distance, Strava was only at 3km.  By the time I finished, I’d travelled 12km by Strava and by Garmin over 15km.  Obviously average speed was massively different as well.

Owing to the nature of the route (clifftop/coastal, lots of ins and outs around inlets and coves) the time-honoured method of OS map and string isn’t going to give me an accurate answer.

Anyone else had this, anyone any idea why and what I can do about it, other than just chill out and enjoy the route (which I did anyway

Can you have a look at the route on Strava and see if it seems about right?

If you have a laptop, perhaps you can plug your watch in via it's USB lead and get the .FIT file for your run off it that way.  (And then upload it to Garmin via the website, or perhaps convert it to a .gpx file to upload elsewhere.)

Or is this entire post actually just a humblebrag that you're on Orkney on a nice day.  (If so, mission accomplished - I'm quite envious!)

I believe that during any set period of time different devices receive quite different amounts of data from satellites, which produces differences like this.  A few years ago somebody I knew did a systematic comparison of various GPS devices (including at least one mobile 'phone and a garmin gps unit I think) and when analysing the data in detail found that the rate at which the devices communicated with the satellites was very different.  After various experiments he came to the (perhaps not very groundbreaking) conclusion that devices that are more specifically designed to create GPS tracks do it more accurately than 'phones for this reason.  On that basis, although there are things that can mess with the GPS on your watch too, I would usually trust the garmin over a 'phone.

Ha ha, you wouldn’t have been so envious pretty much every day this past week - it’s been really windy and usually raining as well.  I was feeling totally celebratory to wake up on a sunny day with only a light breeze on a day of annual leave

Haven’t managed to get Garmin Express on my laptop to connect with the watch.  My macbook is one where Apple helpfully did away with the usb port, so I have to use an adapter.  Garmin express “can’t find my device” tells me to check connections and try again.  When I unplug it, I get a message telling me I’ve not ejected my Garmin device properly!  The one it can’t find!  Grrrrr.

Had a look at the route on Strava and it’s pretty hard to tell from the trace.  I think it could be simplifying and straightening out the route a bit because I probably wasn’t running quite as directly as it suggested.

I will run both over the next few weeks and see how it goes.

Have also got a more up to date phone heading my way, so can install app.

Possibly not quite dealing with your question, but if you weren't aware (as I wasn't for several years) you can have Strava 'correct' activities for distance after the event. It's somewhere in the 'Edit' tab for the specific activity.

There are various sources of error in GPS tracks, but in your case I suspect that the main culprit is a too-long interval between position measurements. If you are doing a lot of wiggles in your route something that measures your position once a minute (say) is going to come up with a shorter distance than something which measures once a second. I don't know about Strava (don't use it*), but with my phone/app I have control over the sampling interval (fast for accuracy vs. slow for low power consumption). Might be worth a dig? Anyway, if that is the problem, then the longer distance and higher speed) will be the more accurate.

* When I did use it, I found it much more suited to cycling than (slow**) running, which would tie in with a longer sampling interval.

** If I was going slowly up a steep hill, it kept saying that I'd stopped and then a bit later decide that I'd sprinted ahead again. Made any stats pretty useless! That was a few years ago - perhaps they've fixed that now.

Check out the 3D speed setting. Seems to be lots of confusion over how it measures distance. You could always test it by running on a flat surface then compare results. Then try running on hills and twisty trails.

https://forums.garmin.com/outdoor-recreation/outdoor-recreation/f/fenix-5-s...

There is also Ultratrac mode. It takes fewer satellite readings to save power, at the expense of accuracy.

Trust the Garmin over the phone. I use one second recording on my Garmins, goodness knows what the phones use, and up until just now I used GPS + GLONASS satellite data. I have now turned off the Russian system but expect plain GPS to be perfectly fine. I don't use Strava, and I have heard of discrepancies between Garmin and Strava before. Have you seen the DC Rainmaker videos about the 45? A mine of information.

It’s most likely proportional to the sampling frequency of the GNSS signal. The route will be an interpolation between a series of points, each one sampled instantaneously by your device (in this case, watch vs phone). Now imagine drawing a circle over the course of exactly 10 seconds. If you sample the location of your pen once a second (ten times) you end up with a decagon. If you sample every 0.1 second (100 times) you end up with whatever the name for a hundred-sided shape is. Anyway, the latter is clearly a better approximation of a circle, and a longer distance to boot. Chances are your garmin samples more regularly than the Strava app your phone, so you capture more of the detail, and get a longer and more representative  route.
 

TLDR: Use the watch, unless there is some wild error that’s obvious when you look at the trace.

Addendum here for terms which rile me up when used incorrectly (Aka all the time):

GNSS: Global Navigation Satellite System. A constellation of satellites which is used to provide a position. There are multiple.

GPS: One GNSS. Only one. Not all of them. Not the catch all term. Yes, it’s the original, which is why people use it as a generic term. Run by the US, and actually the least accurate IIRC (at least the non military frequency).

GLONASS: don’t know this acronym, but it’s the Russian GNSS.

Gallelio: the EU GNSS. The most accurate.

In practice your device will use a combination of these systems. This doesn’t make it more accurate than a single system, rather just gives it access to more satellites to triangulate from - you need a minimum of 4 to even get a position, but ideally 8 or 9 for a decent signal. Each constellation is around 30 satellites, half of which will be blocked from you by the earth, and some which will probably be blocked from you by terrain/buildings etc.

/rant crossed with TED talk.

Just back from a 4km jaunt to the loch near my house in the wind and rain (usual Orkney weather reasserting itself).

The route is made up of straight lines.  Both devices measured pretty much identically.

So I reckon it does suggest that the Garmin measured more accurately over a wiggly route owing to more frequent plotting.  

Yay, I ran further and wasn’t quite as slow as Strava said!

Noted. But GPS rolls off the tongue easier than GNSS, so I shall continue to refer to ‘my GPS’ and to ‘doing the hoovering’

(though probably not in related sentences, unless Dyson come up with a self-propelling vacuum that uses GNSS to work out where in the house it is)

I had a similar experience recently when I was trying out my new (to me) Garmin Instinct Solar alongside my phone tracking app (RunnerUp). In my case though, the Garmin calculated a shorter distance than the phone to the tune of ~800 metres over 10km.

Looking closely at the routes plotted by both devices, it's clear that the Garmin is doing a better job - its line is much truer and straighter than that plotted by the phone,  which waves around considerably.

My phone is apparently capable of using GPS and GLONASS but I don't know which of the two systems the RunnerUp app uses. Worth noting that I carry the phone in-hand too which isn't ideal.

I have the Garmin set to use GPS & Gallileo.

Just back from another run (which was mostly on fairly straight roads and included the route to the loch I walked yesterday with no distance discrepancy).  Miles out again.  Garmin measuring 10.8km, Strava 8.2km.  Very annoying as I’m trying to keep an eye on pace and have no idea which to trust.

My next experiment will be to do a run with the watch set to “walk” mode, as it gave an accurate reading yesterday.

Should get an opportunity to pair Garmin with a more modern phone soon, so will see if that sheds any light.

I’m the opposite when running on trail - my trace should be a very wiggly line as I’m avoiding boggy bits, hopping over tussocks, detouring to an easier bit to climb over a fence etc.

I also need to check both devices to see which satellite set up they are using, if I can set them the same, I will.

I didn't explain very well. In some parts, I.e. very narrow forestry tracks that meander considerably around trees, the route I follow is also very "wiggly". What I mean is that despite this, the trace plotted by the phone adds a significant additional meander of its own when compared to the Garmin.

I've taken a couple of screenshots to compare. This section of the circuit follows a narrow, winding track through woodland, as described earlier.

Phone:

https://i.postimg.cc/G350GXgc/Screenshot-20220316-112431.png

Garmin Instinct Solar:

https://i.postimg.cc/pVGvkWrM/Screenshot-20220316-113511.png

It seems that GNSS inaccuracy has more than one form.

Edit:

Two more screenshots, this time of a straighter section of the circuit which follows an almost direct line through a field.

Phone ("I'm not drunk, I can't find my seat."):

https://i.postimg.cc/VvbstvMw/Screenshot-20220316-120348.png

Garmin Instinct Solar (ignore the one stray line. I didn't switch the watch tracking off when I finished):

https://i.postimg.cc/yNt880gj/Screenshot-20220316-120412.png

Gallelio: the EU GNSS. 

If we’re doing pedantry: Galileo.

Alopogies.

why not just run with only garmin as it exports to strava anyway?

yes, i think strava measures less often and if you ran a smallish circle it would in theory create an octagon (not tested).

Can you see the frequency with which your phone app is recording? It's possible (particularly if it's an old phone), that it's recording less frequently than the Garmin, so is cutting corners and recording a shorter distance.

I've always set my Garmin to record every second, in order to improve the accuracy.

Strava is not a device that receives or processes a GPS (I disagree that this is not an acceptable generic term) signal - it simply displays an input provided from some other device; in this case your phone.

As such the discrepancy is not a problem with Strava, but most likely with your phone. You would (almost certainly) notice the same issue if you used any other location based app with the same level of precision.

Running watches have better GPS receiver chips than phones (generally speaking) so the data will be better. I would always assume that the watch is giving the more accurate result.

Not quite true. Applications decide when to request a location fix from Android (Google Play services to be more precise) and the most likely reason for poor tracking is that the App (e.g., Strava) is not requesting a fix often enough to capture the 'wiggles' in your path. In various apps I use there's a setting in the app which control how often a fix is requested - anything from 1s (high accuarcy) to >1min (lower power consumption).

dammit

Loving alopogies- I think I’m going to add that into everyday speech!

I have now managed to upload my tracks onto Garmin Connect so I could see the maps.  There were no maps for the runs but there were maps for the walks.  So I’m wondering if I had GPS enabled at all!

Have now had a fiddle about with watch and a quick jaunt down the road and managed to make a map of a run.  No time or energy to do a proper run to test it tonight but I’m hopeful I’ve fixed my problem.

Similar topics have come up in the past.  There are a few moving parts to this.

Couple of key points:

1.  Strava will apply its own elevation map [1] if either your garmin does not provide elevation data or it doesn't recognize/like it.  In your case, a forerunner 45, doesn't have an barometric meter (correct me if I'm wrong) and as such elevation will likely be calculated using Garmins elevation map.  The same is almost certainly true on your phone - it too is unlikely to have barometric meter, but in its case - its elevation will be calculated using Strava's elevation map.  There is likely to be differences between the two and as such differences in distance will appear.  (Yes - you can calculate elevation using GPS satellites by the margin of error in the vertical axis is much greater than the margin of error in the X-Y.  That plus the fact that for large parts of the world we have relatively accurate elevation maps means it makes for a more accurate elevation reading to simply take the X-Y location from GPS to a elevation map lookup and take elevation from there.  That is what is happening for most consumer grade GPS applications.)
2. Strava, by default, uses GPS-based distance calculation calculated in real-time.  "Since this is a GPS-calculated distance, a flat surface is assumed, and vertical speed from topography is not accounted for. Also, some accumulated distance may be lost as straight lines connect each GPS coordinate, instead of an arc. This method of calculation does not capture variations in the route between GPS points and may vary further when battery-saving features are enabled."  It does not have to be this way though - Strava will also support you to perform what they call Strava post-upload approach:  "After GPS data is recorded, and uploaded to Strava, the data is parsed into streams of data and analyzed. At this time, a calculation can be run on the GPS coordinates to get the distance. This is how Strava determines distance for any uploaded file that does not include a distance stream.  Post-upload GPS-based distance can eliminate problems like stuck points (see above) and create smoother, more accurate distance data than the device equivalent".  Strava also says the following:  "If you suspect there is a problem with the distance recorded by your device you have the option to override your device's distance with the Strava post-upload approach. Click on the three-dot action menu in the left sidebar and select the option to "Correct Distance." This can improve the quality of uploaded data by eliminating outlier GPS data like inaccurate GPS points and data that is clearly inconsistent with the file. If you change your mind, click the button again to revert back to the original distance."

My recommendation:  Start by asking Strava to correct your data.  That will likely bring it closer to the number recorded on your Garmin - but they will still not match.  There after ... I simply ignore it.  As a recreational athlete, if you use the same device every time you're out - it does actually matter ... as long as it is wrong by the same amount every time.  You can build your baseline off that.  I don't think a watch vs phone is a fair test.  Borrow someone else's gps watch, stick both on your wrist and see if you're still getting differences ... I will guess that the difference will be much smaller than you are seeing vs the phone.

OT:  I wouldn't recommend the phone for Strava.  That's your emergency get out of jail if you get into bother or come across someone else who needs assistance.  Turn it off, stick it in a ziplock and bury at the bottom of your pack.  

[1]  https://support.strava.com/hc/en-us/articles/115000024864-Strava-s-Elevatio...

[2]  https://support.strava.com/hc/en-us/articles/216919487-How-Distance-is-Calc...

Ended up not needing to do any of that to get things working.

Upon uploading Garmin data to my new phone which would run the connect app, I realised it wasn’t mapping at all whilst on “run” mode.  Managed to fix that and all is well.  

Ran both Strava and Garmin simultaneously as a test yesterday on a short hill session near my house and found they were reading near identical figures.

All is good.  Thanks for the help and comments everyone.