1. Use height data from GPS to calibrate barometric height on the fly
-> Here starts track
2. Calculate barometric height
3. Retrieve GPS based height and calculate difference between barometric and GPS height
4. Add the value divided by $COEFFICIENT (maybe 50) to barometric height and use the result as new base
5. Got to 2.
In this way you will start with a height that has the maximum deviation of the GPS signal received. With every measurement the barometric height will drift a bit towards the GPS height, but the jitter of the GPS signal will not show up in the data, provided the coefficient is high enough.
You basically do a continuous recalibration over time without having to wait for 4 minutes in one place. Additionally you constantly compensate the drift of the barometric measurements due to changing air pressure conditions.
An additional idea would be to increase the GPS polling interval and/or decrease the coefficient at the start of the track until the mean deviation between GPS and barometric height goes below a certain limit.
As bonus, of course DEM data could also be brought into the calculation instead of GPS (especially for the initial calibration).
Hopefully the idea is clear and has no significant flaws. I think it shouldn't be all to hard to implement and would make using the barometric height sensor a lot more convenient if there is no need to manually calibrate it. Especially for longer tours it might even prove to be more accurate than barometric measurement alone.
Regards,
Wolpertinger