Curious if anyone else has noticed this. On a fairly windy ride the other day I noticed my Edge 500 was showing some rather bizarre data going from - to + grades on a nearly flat section. Here is the profile for a 1 mile section on a normal ride.



And here is roughly the same 1 mile section on the windy day. I was riding into the wind at this point with gusts pushing 30mph at times.



My working theory is that the wind was causing pressure changes that affected the unit. What I don't understand is why that isn't an issue when descending when the "wind" can be 40-50mph. Or is it?

Here are the data files for anyone interested. The above section is at about 4 miles in.
Normal day - https://ridewithgps.com/trips/1383761
Windy day - https://ridewithgps.com/trips/1377992

Have you tried enabling Elevation Corrections? (I use Garmin Connect)

I don't use elevation correction. It always seem to overreport the elevation compared to the Garmin. The Garmin has a barometric which is supposed to be fairly accurate. I guess I'm not too concerned about the most precise data just consistency for comparisons.

In this case it's not the actual elevation I'm interested in but the effect wind seems to have on the data.

They are accurate most of the time, but because they use barometric pressure you are not going to get consistent readings. Anytime you have pressure changes, (weather systems moving in and out of the area, etc.) your readings will get funky. I ride the same routes constantly throughout the year, and my elevation readings are, 1) always way high, and 2) never less than 15-20 percent in variation from one ride to the next.

If you're looking for consistency, the corrected elevations are better data. I think one of the reasons everybody hates the corrected elevation is because most of the time (in my experience) it's lower than what their Garmin reads. Everyone wants to believe he/she really climbed 5,000 feet so they bash the lower number.

Wind affects the pressure reading, and therefore the elevation change data. I've found the internal altitude measurements of the Garmins to be (a) highly variable from unit to unit and (b) typically highly optimistic.

I usually see the opposite.

"Using the elevation data from your GPS, your trip has +1814 / -1803 feet gain/loss. Our elevation data shows this route has +2016 / 2031 feet "

Maybe I should start using the corrected data and get a free 10% boost. Yesterday's ride is even better.

"Using the elevation data from your GPS, your trip has +2693 / -2661 feet gain/loss. Our elevation data shows this route has +3757 / 3799 feet " Seriously? 1000 feet difference?

The inlet ports for any barometric pressure reading at speed must be carefully designed to minimize airflow effects on pressure. I doubt the Garmin's design takes that into account. I'm not sure it even has enough surface area to do so. Airflow must be smooth and uniform around the inlet, and airflow direction at speed needs to be perpendicular to the inlet - which might be why windy days have more effect than descents.

That's interesting. Most of the people I know around here get much higher readings from the sensor than the corrected elevation. Maybe it's our location. Either way, if you ride a lot of the same routes you're going to get more consistency from the data sets.

Also, if you normally start rides from your house, put it in as a known elevation. You google to find a site where you can get the elevation, but once entered it seems to help the garmin more accurately track elevation

Yes. The other day I was riding in the wind, and I could change the grade by cupping my hand in front of my Edge 305.

You should try ridewithgps.com - their corrected elevation data is like EPO! On a 85mi ride the other day (Thu) my 510 said 8600 ft; ridewithgps.com says 9700. God, I love that service.

Headwind definitely shows up as a negative grade on my 510, and tailwind (I wouldn't know because I always seem to have a headwind both ways ) as a positive grade.

I suspect that Garmin essentially measures stagnation pressure (much easier to measure than static pressure) and then uses Bernoulli's equation to correct for speed. Which is why it gives correct readings during fast descents (Bernoulli's equation will work as long as there's no wind) but produces weird results in high wind.

If that's the case, 30 mph wind would reduce apparent elevation by ~60 feet if you're standing still. It's more complicated if you're moving.

No. The ports are on the bottom of the unit, more or less perpendicular to flow, so would be static pressure. For stagnation pressure, the port would have to be pointed directly into the air flow.

For static pressure you need a sensor that is on the outside of the unit. Garmin has a system of channels inside the mount that terminate at the intake port. There could be lines of flow that terminate inside the port. I guess it depends on the exact geometry.

It should be very easy to test. If Garmin normally measures stagnation pressure (as mounted on the bike), it has to do velocity correction. If it does velocity correction, it should immediately show up as a jump in elevation if you turn it on inside a car that accelerates from stand still to the freeway speed (you may need to crack rear windows to equalize pressure).

What does happen on the 510 on occasion is on a headwind climb it goes seemingly bonkers and jumps to something ridiculous, like a 35% or 50% grade, then slowly reduces back to something sane. I assume this is when it's so off the GPS altitude that it triggers an altimeter reset. Curiously, this never seems to show in the post-ride data. Maybe it's identified heuristically as junk data and ignored.

I guess Garmin could do something bogus with the air ports and then try to correct in software. But the concept of static pressure for altitude and stagnation pressure for airspeed has been basic to and uniformly applied to aviation flight instruments since the earliest days of flight. It's not new. Somewhere at Garmin I'm sure there is someone who knows this because they've been making aviation GPS units long before they made sports units (I used my first Garmin aviation unit in 1991), and needed to know how GPS indications would correlate to existing certificated flight instruments.

One issue they do have have to contend with is thermal sensitivity of the pressure transducer they use. In basic flight instruments this was done with hardware. Not transistors, capacitors, resistors hardware, but tubes, bellows, gears, levers, bimetallic strips hardware.

I've never seen a static pressure sensor on the outside of anything with one. From aircraft to model rockets to garmin head units, static pressure is measured via a port located perpendicular to the airflow and ducted to the sensor.

Stagnation pressure would be measured with a duct facing forward, and there is nothing like that on a garmin or it's mount. That sort of system would be wildly inaccurate with even the slightest breeze.

Interesting discussion.

Now I'm not sure what to do. I read somewhere that the corrected elevation can be off too. For example riding next to a sharp drop off the gps could show you as on or off the road enough to jump your elevation up and down. Obviously barometric pressure isn't perfect either and can change over the course of a ride. It's not due to a specific site either. The above data was from ridewithgps. Here is my ride today on garmin connect. 4,887 feet with elevation correction and 3,520 without. I consistently get more elevation using the corrected data. Wish it could boost my average speed too

Garmin's sensor is not perpendicular to the airflow, it's more complicated than that (there's a duct in the center of the mount that leads to the sensor, it's not clear where it begins.)

I did a simple experiment that you can try to replicate. Take the unit out of the mount, close off the port with your finger and start accelerating. With my Garmin 800, I see reported elevation fall as I accelerate and return to the original value when I slow down.

I'd take "can be off" with a constant error set against a "will be off" with a variable error set.

Use the corrected elevation from the GPS data if you're that concerned with having the "correct" elevation data.