Dither on Strava tracks

[Russell]

I've noticed that Strava often over records milage and the images show why.  They show a snippet of the ride where we stopped for a lunch break (outside) and its obvious that the Strava track is all over the place.  The clean track is from a phone running Viewranger.  On that ride Strave was 2ml longer than the bike computer and Viewranger.  The dither is about 1.3ml so adding that to a tea stop as well easily adds up to the 2ml difference.

Strava Odiham Lunch by Russell Wiles, on Flickr
Viewranger Odiham Lunch by Russell Wiles, on Flickr

Strava on a Motorola G4 Play
Viewranger on a Motorola Moto E

Is this a Strava issue or a phone set up issue.  How do I go about investigating it?

Thanks

[grams]

Do you have autopause enabled on Strava? From the Settings button on the record screen.

It looks like Viewranger has run a very aggressive simplification algorithm over its track. It would likely have started with the same raw GPS data as the Strava track.

Neither track is more correct than the other one. Working out what a track like this *should* look like is hard.

[Little Jim]

It might not be Strava.  I use a Garmin to upload to Strava and also use my phone with the Strava app running as a back up for just-in-case.  The Strava track from the Garmin is consistently about 1 mile shorter on a 50mile ride than the equivalent from the Strava app.  I have no idea why.

[Feanor]

Sample rate, perhaps?

[Russell]

Yes, autopause is enabled and I have seen it display autopaused on the screen when we stop.

From the timestamps Viewranger is recording every 5 secs but there is definately some kind of filtering during the upload as the resolution looks much better on the phone screen when reviewing the track.  Viewranger displays a stop as a series of track direction arrows (so still sampling) but all very closely spaced as you would expect if the device is stationary.

I'll look at running Viewranger on the 'Strava' phone to see what happens.

[frankly frankie]

You say your lunch break was outside, but were you under cover - garage canopy, or shade of some sort.  It's normal for a GPS to get a scatter of inaccurate fixes in that circumstance.  Removing them from view is then a software thing - some will, some won't, sometimes you want it, sometimes you don't.

[bobb]

Yes, autopause is enabled and I have seen it display autopaused on the screen when we stop.

Auto pause doesn't always work very well IME. Whenever I stop (coffee, lunch, whatever), I always pull out my phone and hit Stop. When I'm ready to go again, I hit Resume...

[PaulF]

^this. Unless there’s an accelerometer in the device the only way it can tell if you’re not moving is if the position from GPS remains constant.  If the GPS is hunting for a fix....

[Russell]

Outside on the lefthand parkbench.  Pretty much open sky but shady!

https://www.google.co.uk/maps/@51.2531361,-0.9406296,3a,89.7y,177.8h,83.38t/data=!3m6!1e1!3m4!1sHl-HouGZxSYFuYrQ5Qcdew!2e0!7i13312!8i6656?hl=en-GB

Thanks Bobb'n'Paul

That makes sense but is a pain.

However here are two screenshots from the phone showing us picking up the shopping in Sainsburys.  Both apps were running at the same time on the same phone, using presumably the same GPS.  The Viewranger track is nice and smooth and shows no dither when stationary for just a few minutes, the east west short section just above  the cross hairs, but the Strava track has the distinct signs of dither, hunting, while we were stopped.

Viewranger by Russell Wiles, on Flickr


Strava by Russell Wiles, on Flickr

Sort of begs the question, if Viewranger can do it why can't Strava?

I'll experiment further when we next go out.

R

[TimC]

I suspect the answer to that is that Viewranger is optimised for walking, and probably has a wider velocity 'null zone' to avoid that kind of dither, along with possibly a better vector resolution. The downside is that you probably lose reliable recording if you're a particularly slow walker.

[grams]

I note the large radius curve on one of the Viewranger lines on your way either in or out top left. On that map, Viewranger looks like it's decimating the track and then reinterpolating with curves to make it look more "natural".

I don't know if that's good.

[Lightning Phil]

What is the stopped time according to view ranger , does it match how long you were stopped for?

[Russell]

I have deleted these tracks as they were in the car (Strava got upset!!) so I will answer your question next time we go out on the bike and have a proper stop.

[nikki]

Hi Russell, what's your ideal end result here? Understanding what's happening or aligning your Strava data so it more closely resembles actual distance travelled?

A quick-ish fix to the latter is to use the split function within Strava to break your ride into 3 separate activities: pre-break, break, post-break, and then to delete the messy break activity. You'll now have two activities rather than one, but their total distance will be more accurate. You can download and combine the gpx files with a bit of extra work if having a single activity is important enough to you to warrant the effort.

For the former...
Much as others have said, to a certain extent what you're seeing is just how GPS data are and different software and settings will have differing effects on how much the numbers gets tidied up. Other variables to add into the mix include satellite configuration in relation to the receiver and environmental structures at the time of the reading, orientation of the GPS module, whether your body is between the receiver and the transmitting satellites and, even, whether there is dew on a nearby buddleia bush or not!

I've done extensive work collecting gps data with 'identical' receivers, and they never give the same results - each has their own character it seems.

Looks like a nice spot for sarnies, anyway! 

[frankly frankie]

   

2 apps on one phone - seems to me you're getting the best of both worlds! 

The frequency the sats are polled for a 'fix' - that's a constant (hardware).
The frequency that positions are written to memory - that's software and is likely to be different for different softwares, and may or may not be user-configurable.  In your first image for example it looks like about 10 points have been written - in your second image it's probably over 35 in the same time / distance.
Each written position is invariably averaged from 2 or more fixes, software again and may or may not be configurable, but if all else is equal one might expect that if points are written more frequently there will be less averaging (not sure this is actually so, though).
The line-drawing software then also does a 'best fit' which is very obvious in the first image, and different software may use different algorithms for this.

This should be in the GPS sub-forum.

[quixoticgeek]

Yesterday I visited a friend, while my bike was in her hallway, (with 3 floors above us), my wahoo managed to add an extra 19km... I had auto paused, but it kept on unpausing. I had to pause it manually.

Not had that happen before

J

[Lightning Phil]

Yesterday I visited a friend, while my bike was in her hallway, (with 3 floors above us), my wahoo managed to add an extra 19km... I had auto paused, but it kept on unpausing. I had to pause it manually.

Not had that happen before

J

Stay long enough and you’ll be able to complete an audax in your friend’s flat