Garmin Total Ascent calculation

I ahe had my gamin for 3 years and found the ascent and descent features are not 100% accurate. I can go on the same ride 3 times and get 3 very different results. I just use it as a suggested amount.

I noticed the same thing with my Garmin.  Did a ride last month, Garmin showed 6500 ft. ascent, mapmyride.com showed 4500 and another site I used to map it showed 5500.

From what I have seen elevation is tricky to get right in mapping programs.

Slake707 is the local expert on how these things work, waiting for him to explain my ramblings...

I've used pressure-based and GPS based altimeters for many years in mountaineering. Here is what is happening.

In order to compute total gain, the Garmin must take point elevations often. Each one of those has some error in it. At first thought, you'd think those errors cancel each other out. But the math doesn't work that way.

Assume you're riding a perfectly flat road at 500' elevation for one mile at 20 mph. If the Garmin is taking point elevations every 30 seconds, it will take six elevations. (It actually takes them far more often). The Garmin will compute elevations (each probably with a little error) of say: 510, 505, 495, 500, 505, 500 which show gains of:

495->500 = +5'
500->505 = +5'
Total Gain = +10'

So the Garmin computed +10' of climbing when there should have been 0' of climbing. Over a long ride, this error adds up big.

Mapmyride on the other hand takes exact locations, and then looks up elevations in a database which are far more accurate. I think they're based on either 2 or 3 foot contour lines.

You can get more accurate with a Garmin by exporting the location points into an application like mapmyride which then looks up those elevations in the database. But there is still error involved that can add up big as well. Every time the Garmin computes a location there is also error. So when mapmyride is looking up the elevations that correspond to those slightly incorrect Garmin points in the database, there will again be incorrect elevations just like the example above.

Manually mapping a route on a map showing roads is more accurate but there is still error because the road maps don't align with the actual sat photo perfectly. EG. if you miss a bridge the elevation gain could include the elevation gain if you had ridden down to the low point under the bridge and then back up. It seems minor but when you're talking about 2-3' contours over a long ride, it adds up.

Bottom line is that (by far) the most accurate way to determine climbing gain is to manually trace a route on the sat photo. But of course that is very tedious and time consuming. This is how I do my climbing gains. I have all the climbs I do mapped in segments which I add up depending on which climbs my ride includes. Since my climbs are relatively long, I've double checked the accuracy based on USGS benchmark point elevations which is easy here because benchmarks exist at the bottom and top of most all my climbs.

ETA:
One issue that confuses many people is that they assume that because the Garmin's total ascent and total descent are relatively close that the device is accurate. But again, the math doesn't work that way. Take the same series of six elevations above. The total descent is:

510->505 = -5'
505->495 = -10'
505->500 = -5'
Total Descent = -20'

It appears that +10 and -20 are pretty close so the device must be correct. WRONG. The false descent is based on the same incorrect elevation series and therefore Ascent and Descent will always be similar regardless of how incorrect the elevation series actually is.



Edited by breckview 2009-06-19 8:23 AM

Having known about this inaccurate ascent tally for some time yet not understanding why, your response helps me grasp whats truly happening... especially on long rides where I've always been suspect of my total ascent reading. (calories are high too, I usually reduce by 30%)

I've been "mapping my ride" afterward and always feel more comfortable with their totals. The Garmin devices sure do make you feel like you bustin' arse though! For shorter rides, just think of the totals as "in the ballpark" and for longer rides you never can tell.

If you search around the 'net there's a lot of discussion of the algorithms people "think" Garmin uses to compute calories. There are even some remarks from Garmin programmers floating around. I can't remember any details because it's not in my area of interest but as I recall, the algorithms were bordering on ridiculous.

A Garmin-sized device will be eventually created that will compute accurate elevations using either the current GPS point elevations with some brilliant error correction algorithm and/or using the GPS system that the military uses that's far more accurate. The device would somehow look up elevations using more accurate point locations in real time using the database that it has downloaded and stored. I'm sure the database is massive so the device could only store very small parts of it depending on its location and the speed it's moving.

ETA #2 (BT wouldn't let me put this in the original message):
In terms of differences between say Maymyride and BT's Route Tracker with routes both mapped manually on the Sat photo, I think the differences are likely due to how the specific application handles it's interface with the elevation database.

For example, in order to speed up a mapping application it could be programmed to only look up elevations which are X feet apart. And/or the app could be programmed to ignore elevation gains that are less than Y feet.

Total elevation gains are greatly dependent upon the resolution of measurement.
EG. if your resolution is Y=10 feet (gain) and X=100 feet (distance), the results will be far different than if X=Y= 1 foot .

i knew mine was wrong when i did a track workout, and i had gained 500-700 ft of elevation. . . which was the first time i used my 305. . .

"Having known about this inaccurate ascent tally for some time yet not understanding why, your response helps me grasp whats truly happening... especially on long rides where I've always been suspect of my total ascent reading."

Exactly!!!  A lot of people do not realize that the total ascent/descent amounts are used by the Garmin to determine caloric usage.  The errors in the elevation data it is using cause the caloric usage calculation to be high.

In answer to your question as to 'why' (and to add to breckview's eloquent and excellent explanation), the GPS system is optimized for lateral navigation.  It uses satellites in orbit looking down at your position  from space.  Imagine if you were at the top of a 50-story building looking down at the sidewalk at a guy on a ladder.  It would be difficult for you to determine whether he was standing on the top rung 10' above the sidewalk or on the bottom rung.  But as soon as he stepped 10' away from the ladder, you would be able to resolve his new position easily.  Now, your GPS unit is actually receiving signals from 10 satellites (at least I think the 305 has a 10-channel receiver) and of course they are at different positions in the sky so that they can provide some additional accuracy as a result of the geometric benefits.  However, they are all still essentially dealing with the 50-story building viewpoint and the vertical navigation calculation is a tough one.

I also agree with breckview's comment that we will not get truly accurate ascent/descent data until the unit simply plots your horizontal location on an elevation database and records your position that way.  However, since the Air Force stopped dithering the GPS signals us civilians use 7-8 years ago and WAAS has been implemented, positional accuracy has been improved.  WAAS is an enhancement to the GPS satellite system that uses ground based transmitters that rebroadcast the satellite signals corrected for variations allowing for much better lateral and vertical positions to be calculated.  Unfortunately, and I don't know this for sure, but I don't think the 305 has a WAAS enabled reciever.  WAAS was an enhancement pushed forward by the FAA for use by aircraft to allow precision approaches to airports that had no other guidance to their runways.  However, all of the rest of us get to use it too!  I don't know if the new Garmin 310XT is WAAS enabled, but I have been wanting to find out.  If anyone else knows, please share... WAAS has become widely available over the last couple of years, but I think the 305 was designed by Garmin before then.

FWIW,
Greg

I posted not too long ago how accurate one 800ft climb was on the Edge 305.  I then looked at a route that had rolling terrain to it and it was much less accurate.  I know the Edge 305 I have does not have WAAS.