climbing vs. elevation gain

I think that climbing is just the difference in elevation btwn the start and end of the ride.  Whereas accent(garmin) and elevation gain (MB) is the sum total of all the "ups" in the ride.

If this ride started and ended the ride at the same place climbing should be close to 0.   A possible reason for a high climb value in a loop ride would be if you started riding before the gps locked onto enough satellites for an accurate 3d reading.  It requires many more than just the 3-4 for good 2d data.  So even though the gps looks ready to go it'll still take more time for good data.

MB and Garmin and any other program that you upload the raw data to will give different values for accent.  It all depends on how it deals with the data.  The raw data will have some random spikes and such, good programs will smooth this out and not and it to the total accent.  Also each program aslo has to decide how much climbing you have to do before it gets added to the accent value.  In other words, if you ride a rolling road with 10 3 foot swales some programs will count that as 30 extra feet of accent otheres will dismiss it.

Then there's the argument that a gps without a built-in pressure altimeter never gives accurate altidude changes (the FR 305 does not have such).

x2.  Data from my TRAINER ride last night:  elapsed time 1hour.  Total ascent 3874 feet.

Right.  That's correct -- and completely consistent with the post just above yours.

ETA:  But the error on the Garmin is, I think, somewhat worse than +-10 feet.  I just looked at the elevation data from my trainer ride last night, and while the average elevation is about right (around 280), there are plenty of spikes, some as low as 150 and some as high as 400.  As the other poster said, those add up.

You should turn off your GPS indoors.. your trainer info is completely irrelevant.  As for garmin accuracy on elevation, it is terrible. BT is way off.. the closest thing I have found to accurate is the elevation correction plug in for sportstracks. It uses something like 1 or 5 foot contours. This actually matches with mapmyride.com which is also pretty close.

Yeah, I think you're right.  Error outdoors (according to my logs, comparing with elevations from mapmyride.com -- of course there's some error there too) appear to be anywhere from +-20 feet to +-75 feet, apparently depending on where I was riding.

I use the Garmin for cadence and 'speed' (magnet on the back wheel of course, not GPS) on the trainer. For some reason, when I turn the GPS off, the Garmin also refuses to use the magnet on the rear wheel to determine my 'speed' (although it will use the magnet on the crank to determine the cadence).

I know that the 'speed' it reports is not accurate -- I use it within a workout to compare how I'm pushing at different points in the workout.  Example:  "last interval I was holding 24mph, so let's see if I can do that again". I know that the actual number has nothing to do with traveling 24mph, but it does allow me to compare my level of effort at different points in the ride.  (I have no power meter.)  

I can't completely agree with everyone...

I don't find the Garmin THAT bad.  Yes, there are errors, and yes, you have to take them for what they are worth, but many time my out and back routes have me ending withing 50-100 feet from start elevation to start.

Plust I have back checked my Garmin with the BT elevations, or Mapmytri.com, and found they all agree within 10% or so...

I think the garmin is good for ballpark.

ALSO, the faster you go, the more acurate is seems to be.  So using it for running is FAR more worthless than bikind.  Using it on a motor cylce or car is even closer.

No I didn't.  I was responding to D.Z.'s post.  My ascent descent numbers are very close when I get back.  AND the ending, starting elevations.  They also are close to BT....

This is only for Bike rides, though.  If I run, I'm way off.  Thats because the garmin takes a reading at every time interval (with some error) and then add/subtracts that from the previous point, whatever it is.  If you move only 30 feet, obviously you probably haven't move up or down that significantly, so the GPS error has more significance.  But if you have moved 100 yards on the bike, the reading probably caries more weight since you probably did move more vertically.

Take that to the extreme and don't even move.  All teh garmin adds up is error, so it's 100% wrong.  OR, if you moved from base camp to the top of everest in 1 second, It could be very accurate within a couple feet - granted, if there is a lot of rolling hills in between, it won't catch them.

Bottom line, my total ascent and desent have never been "way off".  Probably within 10% from each other.  Also probably within 10% of mapmytri, mapmyrun, BT, whatever.

BT mapping by hand is not bad, but when you import from the garmin, which has a point every few seconds it gives completely wrong numbers.  An example
http://www.beginnertriathlete.com/discussion/training/map.asp?routeid=34253

 BT lists as 2500 feet climbing, it is about 1000 ft or so. When I get home I can tell you what sports tracks uses for its contours.

I read somewhere that the higher elevation and lack of deciduous trees in our area would mean more accurate readings.  I've checked my garmin against topo maps and it's usually pretty close.  That's one benefit to living where there's less O2.

Marma explained it to me.  When you put the route in, it "reaches out" to google maps, I think and grabs the data for each point at a time.  For some reason, there is just a lot error that gets introduced when BT goes back and forth across the internet and google maps.  The more points and longer you take to put it in, the worse it gets.

When you hit the reset elevations, all the points are in BT, so it does a faster or mass grab of vert info from google.  Somehow fixing it...

To be fair, Marma gave me the layman's explination some time ago, and I'm paraphrasing it....  But yes, they know there is a problem with communicating with Google, so the added the reset feature.  I do that after loading every route, and it fixes it to be comparible to all the other mapping programs out there.

The 2500 feet is with reset elevation values.. google? you sure they are not using USGS data?

My original elevation was worse (even higher) before the elevation correction ..  sportstracks (which I find to be the closest to what I view reality to be) elevation correction  uses srtm 1" viewfinder panoramas for US topo and srtm 3" viewfinder panoramas for the rest of the world

http://en.wikipedia.org/wiki/Shuttle_Radar_Topography_Mission

This is supposed to be the most accurate digital elevation model.

Smoothing differences between online application also introduce or increase error in elevation