AllWeatherJeff

I'm new to the Ride With GPS app. So far the calculated/predicted elevation gain never matches up with my recorded totals.

On the handful of routes I've ridden using the RWgps app, the recorded elevation gain is 15-25% more than what is predicted the app prior to the ride.

Which is more accurate? Does this discrepancy have more to do with service provider (Verizon) than anything else?

Thanks.

jimincalif

Lots of variables. Here in South Orange County, CA, RWGPS overstates the elevation gain as compared to Strava mapping, my Garmin and the bike computers of most everyone I ride with. However in other parts of SoCal, like the San Gabriel mountains, RWGPS elevation seems spot on. You'd think RWGPS and Strava would use the same mapping data, but apparently not, at least here. For example, here's a century I did earlier this month:

RWGPS route planning 97.1 miles, 7,820'
Strava route planning 97.8 miles, 6,730'
My Garmin 520+ recorded 97.9 miles, 6,138'

If you are using a phone app, that's yet another variable. I don't think it's the service provider, smartphones use GPS satellite signals for position (latitude, longitude & elevation).

Go figure.

clasher

Some phones apparently have barometers nowadays so strava or rwgps might use them. My bolt seems to be a few hundred metres short on rides so rwgps adds some back. Barometers aren’t always accurate either, I have usual loop I do and end up between 350-500m on it.

CliffordK

With RideWithGPS, for real-time tracking, you can hold any data field down, and change what is displayed. Sliding right or left, and you have 2 pages of current data to display.

Options include actual elevation, and real-time elevation gain. And, I find that the gain is off by a factor of about 2x.

Both RideWithGPS and Strava do post ride processing, so the post-ride data is closer.

If I count valleys to peaks, my climbing is usually more than I would expect, especially with a lot of flat riding. But, while riding, one might not realize that one is doing a slow climb/descent, or the effects of very mild rollers or dips in the road.

Nonetheless, GPS data is generally quite accurate for X/Y locations depending on the recording device, but relatively inaccurate for vertical data.

With RideWithGPS, I've been actually climbing (1% or 2%), and watched the elevation drop by a couple of meters.

Anyway, the climbing data is representative, but I do take it a bit with a grain of salt.

Some devices have built in barometers which significantly improve vertical data tracking, although I wonder if the barometers are also responsible for some of the strange anomalies that are found in a large number of Strava segments.

unterhausen

gps altitude is inaccurate and the cheap barometric pressure sensors they use in these devices are fairly inaccurate. In any event, the relationship between pressure and altitude in the atmosphere isn't exactly a constant either. When people asked this question, I used to post the ride I took where my house was 200 feet lower when I got back from a ride. It started raining when it got dark on that ride.

The thing that bothers me about RWGPS is they tend to smooth out climbs so it doesn't really show you the 20 percent sections. I am not a good climber, so i don't really like to pay much attention to elevation changes.

CliffordK

I've found that Strava also has issues with accurately calculating the steeper climbs, as well as constant climbing vs stepped climbing.

It would seem like a system could dynamically combine GPS and barometric data. I'm not quite sure the algorithm, but use the barometric data to level out the brief changes in elevation, while taking multi second GPS averages for the total elevation.

It might be easier to do with post processing, while combining data from multiple riders and cartography.

ksryder

When they released the route this year for Dirty Kanza they released it on Garmin Connect, and I think it showed about 9000 feet over 200 miles. When I plotted the same course into RWGPS it said 12000 feet. Strava's route creator said something else.

After the actual race my ride, recorded on a Bolt and uploaded to Strava, showed about 12,000 feet. But if you look at the rides on Strava from the hundreds of other people on the ride, literally everyone has a different number, even people using the same computer. I've seen everything from 8000-13000 feet.

So my takeaway from all of this is that elevation is really a giant shoulder shrug emoji.

atwl77

RWGPS typically over estimates for me. My recent 600k, for example:

Route built with RWGPS: 9464m
Recorded on Garmin Edge 1030 and uploaded to Strava: 6765m
Uploaded same ride to RWGPS: 6963m

unterhausen

A couple of years ago, rwgps redoodled their algorithm to match Strava because people were complaining about the mismatch. Strava always under-reports, and as the big player they can do what they want. RWGPS seems to predict a bit less climbing than other mapping software. It all depends on an algorithm, YMMV.

I'm a little surprised that Strava doesn't get a lot of complaints about elevation.

wipekitty

Around my area (Driftless area of western Wisconsin, southeast Minnesota, northeastern Iowa), the RWGPS mapping software generally underestimates elevation gain. I'd say it's off by 200-500' for a 60-100 mile route, compared to elevation gain measured using a Garmin or similar computer while riding the course.

It is generally not enough to cause significant issues, and it appears accurate enough for the calculations that really count (significant climbs).

joewein

When RWGPS makes an estimate for a planned route, it relies on map data. It looks up elevation on a map.

One problem with that are tunnels and bridges. If you go through a tunnel, the map elevation that RWGPS gives you typically is that of the mountain top that the tunnels helps you avoid climbing. The opposite applies on a bridge over a deep valley. The map data most likely will have you in the river far below. You can manually flatten these artefacts, but most people don't bother.

Another problem with using map elevation is that mountain roads run along steep slopes. Having you just a couple of meters off to either side raises or drops your elevation considerably. Map elevation data is not granular to the last meter but applies to rectangles or triangles. You see that in elevation diagrams that show lots of small teeth for what is really an uninterrupted climb or descent.

The only elevation gain numbers I somewhat trust to be reasonably accurate are from ride recordings with a device with a barometric altimeter, as shown on the device itself. Depending on the device and the service the data is uploaded to, the numbers may change again.

GPS-based elevation gain numbers from smartphones without barometric altimeter can be wildly inaccurate, especially on mountainous routes. The same is true for RWGPS estimates when one plots a route.

While barometric elevation is affected by the weather, especially on long rides, satellite based data is inherently much more inaccurate. That's because satellite signals allow for more precise triangulation of X-Y coordinates (latitude/ longitude) than Z coordinates (elevation). Strava's monthly climbing challenges are something of a joke because of that, since they reward the cyclists with the most noisy data.

Strava used to import the data from my Navi2coach GPS as is, with reasonably accurate elevation gains (i.e. same as my ride mates using Garmins). Then some time Strava started to post-process the .fit files and calculate their own totals, which routinely inflate the total by 20-30%. I can record the original number in the Strava comments but the annual or monthly totals are now worthless (though perhaps not quite as worthless as with a non-barometric phone).

GadgetGirlIL

When I listen to some of my rider friends talk about RWGPS elevation gains and then claim that RWGPS must be underestimating, I feel like I'm listening to fishermen talk about their catches.

I just pulled up 3 of my longer rides that are in both RWGPS and Strava:

300K (this past weekend) RWGPS = 6,204, Strava = 4,403 (I never needed my granny gear)
200K (Iowa) RWGPS = 6,623, Strava = 6,046 (some serious climbs as the route went over to the Mississippi river)
300K (Iowa) RWGPS = 7,671, Strava = 5,588 (Not as many serious climbs as the 200K in Iowa and the middle 50+ miles were on the Root River Trail which was fairly flat)

I don't pay much attention to the reported gains. Instead I consider how many times I had to be in my granny gear, if the climbs were more than a mile in length, and if I had to walk up any of them. In addition, I have an inclinometer (basically a little bubble level) on my bike so have a good idea how steep a climb really is.

pdlamb

+1

And I don't expect my numbers to match anyone else's to better than a couple percent.

njkayaker

Elevation gain as "predicted" and elevation as measured are not going to be the same.

There aren't many databases of digital elevation data. One common source is https://www.usgs.gov/centers/eros/sc...center_objects.

The resolution of this is about 100 feet. So, the predicted elevation gain might be missing little bumps. So, it wouldn't be surprising that a "good"measurement will often be greater.

GPS works well for horizontal positioning but is poor for vertical positioning. The fact that vertical movement being measured is much, much smaller than the horizontal movement makes it worse (people aren't generally concerned if they are missing 1000 feet of distance but see missing 1000 feet of elevation as significant.)

Using a barometer (air pressure) is considered as being much better than using GPS. Air pressure is a problem because it changes with elevation but also due to temperature and weather changes. It doesn't seem the Garmin barometers are temperature corrected.

It doesn't seem that these other changes effect elevation gain very much. They might cause "significant" differences in absolute elevation (which most cyclists don't care about). That the absolute elevation is wrong doesn't necessarily say anything about the accuracy of the elevation gain.

One basic problem is how small a bump should be counted. Different devices might have different smoothing procedures.

Getting accurate elevation gain on a flat ride is going to be a problem because the real measurement values are likely going to be close to noise.

================

The error in elevation gain measurement is probably much higher than people expect (on the order of 10-20% I suspect).

The elevation gain, of course, doesn't take grade into account. So. it's not a very accurate estimation of difficulty anyway.

njkayaker

What was the elevation gain? How long was the ride?

Changes due to weather generally fairly slow. It doesn't seem to effect elevation gain much at all.

That the absolute altitude isn't quite correct doesn't mean the elevation gain is going to be inaccurate. A 200 feet difference in gain is nearly at the level of noise.

unterhausen

it was actually 200 feet higher at the end https://ridewithgps.com/trips/1885037

There were a couple of nearly instantaneous changes in elevation on that ride. When the rain started there was a 100 foot increase @ about 42 miles, then a 100 foot decrease later on. Maybe when the rain finished, don't remember. The starting elevation is about right, from what I know about the elevation of the town center.

My point in posting that ride being that people shouldn't obsess much over these numbers.

njkayaker

So, it was a 200 foot error on 5000 feet. A 4% error. Basically, nothing. For what looks like a somewhat extreme case.

That is, the errors introduced by weather changes are generally minor.

But the absolute elevation being correct doesn't matter.

All that matters is getting the same number at a particular elevation (reproducibility) and measuring change in elevation accurately.

People also obsess about "calibrating" the devices because they think it effects the gain number (it doesn't).

The gain is a delta measurement (constant errors in the absolute elevation don't matter).

No, they shouldn't.

People tend to expect that the accuracy of the gain elevation is much higher that it can be. It's probably about 10-20%.

They shouldn't obsess about the number because it's not overly accurate and because it doesn't account for grade.

pdlamb

But sometimes it's part of the mental game you play on a long climb, if you know the gap or pass tops out at NNNN feet elevation. If you drove to the start, turned on your GPS and started riding, you could hit the calculated top 500' below the actual top -- and your motivation just ran out with no end to the climb in sight.

kingston 

I rode a 600k in the Driftless a few weeks ago. RWGPS said 21,347 and the two guys I rode with had 26,641 & 26,490 which is almost 25% more. My GPS doesn't keep track of elevation, but I could tell it was a hilly ride by the day-and-a-half of suffering.

unterhausen

the newer devices have some sort of number that shows you how far you have to climb. I never look at the elevation profile on mine, wouldn't help me at all. I only get motivated when I see the warning sign for trucks about the slope of the other side. I know people that look at the elevation profile, but usually it just results in them whining. I don't need any help with that at all, I'm a big whiner. I try to keep it to myself most of the time.

njkayaker

That's kind of unusual (nothing wrong with it). If you have a need to calibrate the unit, calibrate it. The issue is that people have a mistaken notion that calibrating it makes the elevation gain more accurate (it doesn't).

The Garmins can show an elevation profile and your progression on it. Though, it's not always aligned properly: your progression on the profile might be before or after your real progression.

Altimeters used for hiking need to be calibrated since they are used to navigate (so the measured elevation has to match the elevation on your map).

GadgetGirlIL

As for progression up a climb, I use a Garmin eTrek 20x so I see the contour lines on my map and can tell where I am on long climbs. The closer those contour lines are together, the steeper the climb I'm dealing with.

Sometimes to get through a long climb I end up distracting myself by counting telephone poles, expansion joints in the pavement, etc.

79pmooney

I find the elevation gained numbers highly amusing. I just got a GPS (Element Bolt). I can go for my basically flat out and back that I have been documenting as no elevation and get numbers like 2000' gained!
Its elevation vs known elevations isn't a whole lot better. My house varies over about a 100'. My unit has shown an elevation 270' lower than an elevation sign I was passing.

I find the elevation numbers "interesting" like the locations reported by the early GPS units (1980s and 90s). Numbers that are reported to the nearest foot but don't use those locations to find a safe harbor in a storm! Not if you want to live. I used to wonder at the huge climbing numbers riders boasted. Now that I have a GPS unit, I know where these high figures come from and just how much BS they are. I'd love to see the numbers acquired by a GPS. Strava, etc. unit over a course accurately surveyed. Then we could compare the two (computer and the surveyed) and make note of the significant digits on said unit. Sometimes I feel less than half the digits on my display are significant.

Ben

79pmooney

Best line in this thread! Thank you.

Ben

StephenH

One of the problems in all this is that elevation gain is not clearly defined in the first place. If you're riding along, the road rises six inches and drops back down, should that add 0.5' to your total elevation gain? Logically, it should, but on the other hand, climbing that 0.5' hill is zero effort expended over riding flat, so it's not really a measure of difficulty, which is what we generally want out of those numbers. Around here, a ride with 5,000' of climbing means it has a hundred 50' hills, which does introduce some difficulty, but is not exactly the same as climbing one 5,000' climb in the middle of the ride, either. Or, if I include all the little bumps in the chip seal, my normal rides all have about 20,000' of climbing.

Long ago, I got to measuring the slopes on some of my local routes, and on one route, the steepest single spot was one that I normally coasted UP, and it never had occurred to me that it was "steep".

One observation is that if there are three or four different numbers floating around as to how steep a hill is, everyone uses the steepest. So that 8% hill, you may take a 2' level and a tape measure and be unable to find any spot anywhere on it that measures over 5%, but it'll still be the 8% hill for everyone. I think that's similar to the fisherman effect. Ditto with on-bike temperatures, for that matter. And wind speed.