Skip to main content

Twice during my measuring career of less than 100 courses, I needed to compare Start and Finish elevations with a precision I could defend. I used an e-trex Summit taking 5 readings going back and forth between S & F three times averaging them. The instrument has a common 4' range of readings at a given elevation with a steady barometer. The other day after reading the elevation profile thread, I did a little more precise experiment. I took 10 readings with it on the floor and another 10 with it on a 15" high tote. The floor averaged 1,652.7' and the tote 1,654.2' for a difference of 1.5' or 18". On a windy day or with barometer fluctuations, there can be a lot of variability which is why it is necessary to go back and forth at least 2 times to get a credible elevation difference.

Does anybody have a better method?

Oscar Wagner
Johnson City, TN
Original Post

Replies sorted oldest to newest

Oscar, when you are using a GPS device with elevation determined by barometric pressure, you can't be sure using any technique. I have the same problems with my Garmin Oregon.

On devices where the elevation is determined by the satellites (okay, using the satellite data), leaving the GPS stationary for 10 minutes lets it get a very good fix on elevation. However, if there are buildings or large metal objects nearby, elevation will still be skewed by the objects, as the satellite signals will reflect off the objects, giving false time/distance readings relative to the satellite.

Maybe, just use GoogleEarth, zoom in very closely, and use those elevations. I would put more faith in those than my barometric elevations.
GE Elevation Question

I was using GE for course design the Mtn Home VA and noticed a seeming contradiction. Using the elevation profile function, there was a big dip that did not look right. I have been over that stretch on and off for 30 years. Then, I put the hand cursor over the same identical location xxx 18.386 xxx 22.510 and the elevation was quite reasonable relative to the surrounding terrain. For some reason, the hand cursor did not show up in the inset screen shot but it does show the same coordinates. What happened?

Oscar Wagner
Oscar, I don't have an answer for you. But I have encountered similar anomalies with GE that turned out, with analysis, to be apparent artifacts of imprecise mosaicing. It wouldn't surprise me if this is the issue here. One way to investigate is to look at other aerial imagery, such as Bing Maps, zoom in all the way, and compare with GE. You may notice that there are blurring or mismatched image boundaries. This could cause inaccurate or contradictory elevation readings.
Last edited by pastmember
I am going to respectfully disagree that GE or satellite based elevations are going to be more accurate than barometric reading ... if you take a tedious lot of them back and forth several times to establish a RELATIVE difference. My first experience with satellite based elevations was that they were only within 30 feet of what GE says. I have a 2000 vintage instrument so maybe things are better now.
As for GE, what are you going to do if there has been recent construction or the S or F is under canopy or in rare instances indoors?
"The minimum elevation difference was 0.4 feet, collected on September 26 with the GPS-only setting, while the maximum was 4.0 feet collected on September 30, also with the GPS-only setting. In other words, Oregon 600 can serve well if elevation differences are the main study purpose."

Note that they let their unit "warm up" for at least 30 minutes, and they took multiple readings over an extended time period at each site (like Oscar talked about). Also note that this is with a unit with a barometric pressure sensor. I think other GPS units will give you elevations based only on GPS and that is not nearly as accurate.
Oscar, I think we are trying for too much precision using methods that are not intended to be precise.

Using the barometric elevations on my Garmin Oregon 400, I don't trust it at all. I can ride the routes multiple times, then compare tracks, and all will be different for the Start and Finish point, even when it is a single point, not separated locations.

Satellite-based elevations could be more accurate, if the unit is capable of more accuracy. But, unless we get a survey-grade GPS, I don't think we can expect inexpensive devices, nor free online imagery (GoogleEarth, etc) to give us accuracy to a foot, or even 10 feet. But, why do we need such accuracy?

For a drop of less than 1 meter in 1 km for records, we don't need less than a meter accuracy, unless the Finish is right around 5 meters lower than the Start (for a 5k). When it is that close to the threshold, it may take locating a topo map with sufficient scale.
Duane I had the same experience with you when I used the elevations from my Garmin during a ride.
But the people in that study that I posted used the Garmin in a way similar to what Oscar suggests, leaving it in place for at least a minute to let it stabilize on a reading.

I should have noted in my previous post that the minimum of 0.4 feet and maximum of 4 feet that they mention were of ERRORS of all the points they measured.

So if you measure the elevation of points by letting the GPS device (one with a barometric sensor) sit on the ground for at least a minute, you can expect an error of less than 4 feet in the difference between start and finish elevation measurements.
I realize this is much more academic than something you are going to have to do more than a couple of times in a career. You can prove to yourself that this gets close. Tape measure from the the bottom to the top of a file cabinet so you know the difference. We know that even on really calm days, the barometer is going to drift and this must be accounted for so one set of readings on a gusty day is not much use. Set the device on the floor at one point and wait long enough for it get to ambient temperature, say 3 minutes. Then take 10 readings at 10 second intervals and average them. Then go to the other end of the stairs and do the same thing. Repeat this a couple of times more and see what the differences are. We know that these instruments are very sensitive. You can see the change just moving it from your belt to your chest. It is just a matter of lots of tedious readings to get the averages to work for you.
My opinions is that this is the best you are going to do short of hiring a survey crew.
That's not really much of a test though, because the conditions are identical at the two points. The real test is outside at two points that have significant separation, where the wind and temperature may be a little different.
But that's what those guys in the study did, and they got a maximum error of 4 feet for relative elevation, so I'm convinced. If I check the elevation drop on GE ahead of time and see that it is going to be close to -1m/km, I'll use my Garmin to see what it says.

Add Reply

Link copied to your clipboard.