Strava max speed and graph don't agree

[Tim Hall]

I thought I'd have a go at logging my mileage this year and hae chosen to use Strava (free version) in conjunction with my Android phone to do this.  First go was today. I got a selection of  nice graphs at the end of the ride, one of which shows my speed and elevation. Screenshot below.


AIUI the blue shows spped, grey elevation. The max speed listed under the graph is much higher than the max speed on the graph. My stoker was recording the trip too, using some other gadget (fit bit and Android phone I think).  Her max speed was much closer to the blue line speed of the graph. What's going on?

[Feanor]

Sampling issue, I expect.

The actual dataset is fairly large and sampled at a fairly high rate compared to what can be shown on the little graph.

There may be one or two samples with the high speed, and these will be reported as the max values. But the graph will be heavily downsampled to fit the size of the image. The extreme values of the dynamic range will usually be missed in this downsampling, depending on the algorithm used.

Hello, Mr Nyquist.

[Kim]

Since it was recorded using the Strava app there's also the possibility that the max speed may have been determined using the instantaneous speed reported by the GPS chipset, which is determined using doppler shift to a surprisingly high degree of accuracy, while the graph will have been plotted from the positions and times recorded.

Meanwhile a GPX file from a Garmin or similar will only contain the position/time data.  (The Garmin's own trip computer is likely to use the doppler speed, thobut.)

[lissotriton]

If you view it on the Strava website, you can zoom in parts of the graph. Would let you see more details.

Can also check max speed for any segments, might give you a clue.

[telstarbox]

I find the max speed is occasionally something like 65-70km/h which would be unlikely for me, but the average usually "feels" right (slower for urban commutes with more traffic lights and faster for rural Audax rides).

[Ian H]

Strava (a free account) consistently records lower max speeds than my Garmin.  It does the same with climbing.  I can't say it's ever bothered me.

[Tim Hall]

Viewed it on the Strava website and, as lissotriton says, I can see the vMax of 47.2km/h, which happens at 23.0km along the ride.

I exported the Strava data as a gpx file and uploaded it in to ridewithgps.  That shows a vMax of 46.1km/h at 16.3km.

Looking at the gpx file in a text editor it has time stamps and position data, so presumably ridewithgps does Hard Sums to detemine the speed. Does the Strava website do that too?

I'm fairly sure my ambition/ability quotient isn't sufficiently high to get anywhere near 46km/h on a trike at present. I think vocal feedback from the stoker would have also served as a warning.

[Lightning Phil]

You’d have to intercept the data between the Strava App and their servers to see if app has already calculated and provided it, or the servers need to calculate it.

46 km/h would be hit by pretty much any bike or trike on a downhill.  Mostly you’d exceed it on any downhill longer than a minute or so.

[Lightning Phil]

I’ve done some sampling of one of my gpx files and reckon Strava averages over 5 to 10 seconds of time stamps.  The average speed varies, as does the maximum based on the length of period sampled. As does what Strava considers stopped time.