Ride with GPS elevation errors

[mattdwyerva]

I rode DC to Pittsburgh this fall and I had planned each day's ride on RWGPS. 

I have often used this tool around our home and it seemed to pretty fairly show grades and elevations.  However, it showed crazy grades on this tour.  On the segment from Cumberland to Hancock on the GAP rail Trail, it claimed the max climbing gradient was 13%, and said total climb would be nearly 3000 feet.  No no no....  Not sure what it should be, but Strava did not agree at the end of the day, for example.  At all

It makes me wonder as I work on a cross country plan partly relying on this tool....

[Pat Lamb]

Not unique to rwgps.  You know the saying, "All generalities are false?"  That goes double for elevation on digital maps.  It's all too easy for a digital mapping tool to shortcut over a ridge, especially if the map database doesn't account for a narrow railroad cut.  Whatever you're using, treat the elevation gains with a grain of salt.

And if you have a good granny gear, you'll appreciate it at more points than you might expect on a cross-country ride.

[mattdwyerva]

Fwiw

https://support.ridewithgps.com/hc/en-us/articles/4419010957467-Grade-Elevation-and-GPS-Accuracy-FAQ

[John Nettles]

Matt, the RWGPS/Google/Strava/etc. data typically are based on various sets of topographical maps.  As Pat indicated, the route may cut across a ridge/valley, especially on twisty and/or narrow roads.  While it is better on major roads (since wide roads take up more width), they too can be wrong, especially if they are new and/or go over hilly terrain.

To see what I mean, whenever you see a massive descent followed immediately by a massive climb (or vice versa), look at it via Streetview or via satellite view.  You will most likely see a bridge or tunnel or something that throws the elevation off. I "think" they are getting better since they are starting to incorporate actual data from the riders that show there was no massive change.  I have seen a noticeable decrease in the false climbs over the past few years.

Look on the bright side, these massive false climbs make the ride easier than the numbers indicate.

Tailwinds, John

[mattdwyerva]

I've mostly used the app in the mid Atlantic US.  But in my few years of use, I've never seen errors that remotely approach the scale of the GAP Trail errors.   Guess I've been lucky. 

It's dead on for the Skyline Drive for example.   

[BikeliciousBabe]

I've mostly used the app in the mid Atlantic US.  But in my few years of use, I've never seen errors that remotely approach the scale of the GAP Trail errors.   Guess I've been lucky. 

It's dead on for the Skyline Drive for example.

As noted above, what you might be seeing are not really errors but the results of high bridges and/or tunnels, which exist on the GAP.

A great example is the Hiawatha Trail in MT and ID.  To illustrate, I made this map:

https://ridewithgps.com/routes/45166223

It follows the old Milwaukee Road right-of-way from Avery, ID to Taft, MT (except for a short detour around a tunnel that collapsed) and includes the Hiawatha Trail.  IIRC, there are more than a dozen tunnels along that route.  There are also some trestles that are hundreds of feet high.   If you dig deep into the profile, you will see grades of over 10%.  Of course, that is impossible even for a railroad right-of-way crossing a range of the Rockies.  That last hump you see at the end represents the 1.66-mile St. Paul Pass Tunnel, which crosses between MT and ID.  Look at the grades to the west of the summit.  Nearly 30%.  Impossible.

BTW...If you ever have the opportunity to ride that stretch, it is spectacular. 

[mattdwyerva]

Good points and great example.   

[John Nelson]

On August 16, I rode from Ten Sleep, Wyoming, to a campground high in the Bighorn Mountains. The ride is all uphill. Here are six different calculations for climbing feet.

• 3878 feet: What the Garmin barometric altimeter sensor measured on my actual ride.
• 4174 feet: Ending elevation minus beginning elevation (using ending elevation from the National Forest Service).
• 4274 feet: RideWithGps predicted elevation gain minus predicted elevation loss.
• 4279 feet: Ending elevation minus beginning elevation (using ending elevation from RideWithGps).
• 4323 feet: RideWithGps predicted elevation gain.
• 4908 feet: What Garmin says after applying “elevation correction” (using survey data instead of the barometric data).

Garmin (and RideWithGps) say that the barometric altimeter sensor is the most accurate. But in this case, it can’t be accurate because that number is significantly less than the difference in elevation between the ending point and the starting point. The climbing feet can be more than the elevation gain, but it cannot be less.

[John Nettles]

After doing that climb, I like Option 6.  That was a long slow slog of a climb for me.

[mattdwyerva]

Great detail, thanks