Noisy velocities calculated, displayed and exported to GPX-File

[hans1]

After tracking a trip, positions, velocities stored also.
Opening a track through the Statistics button, velocities can be displayed as a table at adjustable distance averages and as a plotted line in a graph.

The positions look perfect. And if I stand at one spot for some time, position data slowly and continuously drift in an area of about 5m diameter.

What I find is, that the calculated velocity data in the line-plot are quiet noisy. I.e. running at a continuous speed of 10km/s generates a plot displaying deviation of +/- up to 2km/h, changing within a second.  That doesn't improve if I change the GPS-time interval oder GPS-distance interval for storing the data. As the modification of these intervals lenghts doesn't make a difference, I guess the velocities are not calculated basing on the time to distance ratio of successing stored locations.
So there are some questions arising from this:
  >> How is the velocity computed
  >> Is there an option in OruxMaps to smooth/filter/average velocity data, to improve the readability of the velocity-plot ?
  >> Does anybody else observe these scattering of velocities?
  >> In case OruxMaps cannot provide smoothed data, are there tools around to do so?

Further: The axis-scale of the line plots would be better readable if I could change the text height.

I'm using OruxMaps 7.4.22 at an Android 10 Smartphone.

[hans1]

OK, after some research I got a general information about GPS velocity measurement.
A velocity information is provided by the GPS.chip of the smartphone. It has a typical  accuracy of about 0,2-0,5 m/s. It is not derived from the distance of two positions devided by the time span of the measurement but it is derived from the bias of frequencies caused by the Doppler effect of movement to frequencies.
So .... if Oruxmaps just reads this velocity information from the GPS chip of the smartphone, this information cannot be modified by extending the timespan. It is always an immediate computation for the moment of one measurement.

So .... I guess the noisy velocity information of my smartphone during running (+/-0,25m/s = 1km/h), displayed by Oruxmaps might be statistical noise or also might be the true representation of the real changes of my velocity while being in the moment of accelerating by unbending my leg in the phase of jumping upwards and the lower velocity in the moment of bending my knee after landing after the jump and decelerating before the next jump .... may be??

[hans1]

It therefore might be desireable if Oruxmaps might additionally offer the option of a low pass filter for the time series to smooth the function and elaborate the true horizontal velocity of running.