Your calculations for total climbing are hopelessly wrong. Quite understandable, considering that gps altitude data jumps around more or less randomly by at least 30m. The 15 flat kilometers to my dentist resulted in a 2000m locus climb... I wish we had that kind of mountains here :-).
You need to apply some sort of dampening algorithm on the altitude data before summing up the differences, at least if the origin is gps altitudes.
It works better with google heights, but that's not really useful in the mountains in a foreign country without data connection.
On a related note: how about supporting the pressure sensor of newer phones? That gives much cleaner altitude data when calibrated once.
altitude problems was already discussed here (//http://forum.asamm.cz/viewtopic.php?f=10&t=728&p=3980) - seems I planed to add selectable strength of filter, but forget on it ...