I am wondering if theres a fix for this- 4 people leave at the same time from the same place and do the exact same ride at more or less the same speed, start and finish together. Garmin 820 shows feet climbed at more than 200 ft difference between two of the riders and none of them are the same. Mileage is the same within a tenth of a mile so its not a gps problem. I believe the garmin uses atmospheric pressure to measure climbing. Anythoughts?

It's likely the state the units were in separately at the start of the ride. If one of the units was stored indoors in climate control, it will likely have a different end reading from one stored in a garage or breezeway. My bike and my wife's bike live right next to each other in my workshop-- we've had rides where our ending altitude is the same down to the foot.

GPS favors horizontal distances and it's very accurate for that.

The accuracy of elevation gain is much lower. GPS is so bad for it that barometers are preferred. 10-20% variation is probably typical. And the variation is probably larger for small total elevation gain.

You said "200 feet" difference. But you don't indicate what total gains you are talking about.

Check the calibration settings on each unit. Barometric altimeters are more accurate than GPS if they're calibrated appropriately. Garmins have settings to calibrate it when you pass a saved location, or from GPS either continuously or just once at the start of a ride. Different settings on different units can make them disagree about how much uphill there was because they'll count differently.

I use a Garmin Edge 200 which is not barometric. After my last ride (28 mi.) the Garmin Edge unit showed 1677 ft. of elevation. When I uploaded the data to Garmin Connect it posted as 1218 ft. And, when it was sent to Strava it posted as 1364. I have no idea why or what the real data is.

Don't worry about it.

IME, anything those types of devices measure, except for speed and distance, is pretty much baloney.

The total amount climbed was 1500-1700 ft depending on which garmin you believe. It is also true that my Garmin consistently shows a difference between ascent and descent of as much as 200 ft on this ride, and is always different no matter the ride, usually ascent being more than descent.

Same here.

I don't think Garmin worries much about their software accuracy. For example, they could recalibrate the elevation when the bike was stopped, but my 705 never adjusts once I hit Start.

I rode in gusty winds today, and the 705 showed 2000 feet of elevation, when ridewithgps calculates 1100 feet for the route. Each sudden gust affects the 705's barometer. But, usually the elevation is quite accurate. I've done rides with lots of up and down climbing on the Blue Ridge Parkway, and the 705 is often within 20-30 feet of the posted elevation at the overlooks.

All these GPS recordings need smoothing of the raw data, and each upload site or software does it's own methods. I noted some differences in a thread post from 2014.

The barometric altimeter doesn't work well on rolling hills-- it's much better at long grades. I have a mountain climb that I've done half a dozen times, which gains right around 4,650ft in 20 miles. Every time I get to the top, my Garmin has been within 40 feet (+/-20ft). As far as I'm concerned, that is stunningly accurate for a consumer-grade device. The last time I went up (last Wednesday) my 520 and my buddy's 810 were within 10 feet of one another.

I wondered how small these barometers are.

For example, this $10 packaged altitude sensor to use with Arduino board computers is 1.2 grams (!) and resolves 0.3 meters of elevation, about 1 foot.

[QUOTE=rm -rf;19548817]I don't think Garmin worries much about their software accuracy. For example, they could recalibrate the elevation when the bike was stopped, but my 705 never adjusts once I hit Start./QUOTE]

Newer Garmins have a setting to do something like that. So do Suuntos.