Hello Geoffrey and others,
Have a good and sportive New Year to you all.
Geoffrey, I can understand that the slope you wanted to implement in the software is giving problems to your team. I did read more about it and I discovered that it hardly impossible to do with only GPS and not having a barometer inside. I found maybe interesting link to read more about it. http://www.gpsplusins.com/Documents/Papers/PAC.pdf
This could be used to correct the data slope. I found out that using GPS data for riding a track with a virtual trainer is not possible due to having to many errors in the altitude, even after using correction plug-in from Sport Tracks. Aldo if you look at the graph it looks very reasonable, but the little bumps in it are hard to ride and not realistic. So if you can’t put the slope in forget it and implement it in the next generation together with a barometer.
This is it for the moment. 
Best regards,
Albert
Hi Albert,
Thanks for giving the excellent paper! I will take a more detail look.
Yes, the most error comes from muti-path, doppler effect, PLL offset, athorphere....
I look another paper which said: To use more sampling to estiminate the speed error because the error factor is normal distribution. It needs over 20 sampling points!
I think it also works on there but only optimize. For my personal thought on slope, there are two ways:
1. Like ascending speed: Slope=[Alt(t)-Alt(t-5min)]/ dist(5min);
Update screen every 5 second;
2. Alt(n)= Average [Alt(n):Alt(n-60sec)] <----- there are 60 points totally!
Alt(n-10)= Average [Alt(n-10sec):Alt(n-70sec)] <------- There are 60 points also!
Slope= [Alt(n)-Alt(n-10)]/dist(1min)
Update screen every 5 second;
Now we are only working on improvement of Pace and indoor training, since we are almost occupied on GH-505.
Have a nice new year.
Geoffrey