A thought:
The other night as I was laying in bed, unable to sleep, I was thinking about a trail near my home. This trail is rather steep and has pretty severe water bars--we refer to this trail as "Water Bar", though recently the forest managers decided to call it "Bonzi". Since they misspelled Bonsai, we still refer to the trail as Water Bar. But I digress.
As I mentioned, the trail is relatively steep, and straight and--except for the water bars--flat. This makes it fast, as you can well imagine. So there I was thinking about the trail and wondering, "how far do I travel in the air at each water bar? Now, I know that I can hit between 27 and 29 mph on the trail on a regular basis. I know that I slow a little for the water bars. So, I grabbed my calculator from off the night stand--what, you don't keep a calculator on your night stand?--and did a little math assuming that my launch speed was 25 mph.

So, assuming that I was in the air a single second, I would have traveled just shy of 37 ft. That seemed like a really, really long way to me. So I did it again, using a jump speed of 20 mph. This time I the answer was just shy of 30 ft. Still a long way. And with that thought drifted off to sleep. The next day I headed up to the trail and, since I
don't have a computer on my mountain bike any longer, thought I'd determine my air-time using the age-old method of 1-mississippi, 2-mississippi, etc. What I found was that I was landing right at the 2 of 2-mississippi. So, I was in the air, about 1 second. Now, there are some flaws with my reasoning, but I think that the errors cancel each other and the horizontal velocity is sufficient.
An observation:As I've stated, I use my GPS these days to record my rides. Recently I went on a long solo ride that had, according to the GPS, 4000ft of climbing. Not bad for 18.6 miles. I was mentioning this to T. since is familiar with the route and he expressed doubt in the veracity of the total climbing. Huh. So I went back and looked at the data. The program I use allows me to upload my rides and it will overlay the route with a topographic map to give two profiles: one from the GPS, and one from the topo.

As you can see, there aren't any really big elevation discrepancies. Sure, you can see where, while descending, the GPS didn't have a good fix--on the top graph is looks like a flat line, then a quick drop, while on the topo (bottom) graph, the slope is more realistic--so I dismissed T. as a naysayer and stuck to my 4000ft of climbing, thankyouverymuch.
Then it happened again.
On group ride I proudly announced that we had done something like 1600ft of climbing. E. said, "no way". Huh. Now, I was thinking about this. Could it be that the mighty GPS is off? I've checked the speed part and it is dead on. The mileage matches up, too. But could the vertical be off?
I then recalled that the better GPS's use a barometer for elevation. My watch has this feature, so on my
next ride I thought I'd use them both and see how far off they were. I expected them to be within 200ft of each other.
I was wrong. Way wrong.
My GPS logged 4500ft of climbing, while the watch--with its more accurate barometer based elevation measurement--showed only 3500ft of climbing. That's 1000ft or nearly 25% error! And it's not consistent, either. It all depends on how well the GPS is receiving the satellite signals, so under tree cover it's worse, naturally.
Long story short, you can't trust the elevation of the GPS using satellites alone. Now, if you are out in the open, and stationary, I think that the GPS does a fine job of absolute elevation. My watch has to be reset nearly every day if I want accurate absolute elevation, due to fluctuations in weather. Both technologies have their place, and neither is perfect. But for accurate total elevation gain/loss, use a barometer.
I guess my solo ride didn't
quite have 4000ft of climbing.
Update on the gps...
Last night on our ride my barometric elevation (watch) and GPS elevation were within 200ft of each other. To make matters worse, C. claims that the GPS elevation more closely matches that of his computer topo maps. Frankly, at this point I don't have any conclusions. If I were looking for government funds, I'd end this post with "more research is required to come to a conclusion." Anyone want to fund me?