How does a cycling computer know the gradient?

[ColinJ]

Even data derived from maps can be wrong. Memory Map has a grid of elevation points which it overlays on the digital maps. It then just interpolates between them.

I noticed on the Humber Bridge ride that it thinks we cycle across the surface of the Humber rather than using the bridge!

It is also very sensitive to lateral error on roads along the edges of steep hillsides where a few metres to the side would effectively have you down at the bottom of a cliff!

 

[Deleted member 121159]

Even data derived from maps can be wrong. Memory Map has a grid of elevation points which it overlays on the digital maps. It then just interpolates between them.

I noticed on the Humber Bridge ride that it thinks we cycle across the surface of the Humber rather than using the bridge!

View attachment 670830
It is also very sensitive to lateral error on roads along the edges of steep hillsides where a few metres to the side would effectively have you down at the bottom of a cliff!

That's hilarious. My local one is the Severn Bridge, I wonder if it's the same.

 

[Alex321]

It shows how unreliable it can be, that on two recent 35.93/35.95 mile rides I did over exactly the same route, the distance was 0.2 mile different (which could be explained over that distance just by variations in where on the road I was), but the elevation gain was different by 69 feet (2001 vs 2070).

And I frequently find that when cycling a loop from home, the elevation gain and elevation loss are different by 20m or more on my Wahoo.

 

[slowmotion]

GPS isn't very good at calculating elevation. Lookee here.....

https://gis.stackexchange.com/questions/234430/inaccurate-altitude-gps-data-with-my-android

 

[mustang1]

I thought barometer would require calibration like on an airplane where they have to set the air pressure several times during a flight otherwise the altitude is incorrect. Shrugs.

 

[DaveReading]

I thought barometer would require calibration like on an airplane where they have to set the air pressure several times during a flight otherwise the altitude is incorrect. Shrugs.

A known elevation at a given point may provide a datum.

Even without that, while absolute altitudes may well be incorrect, relative altitudes (i.e. the amount of climbing and descending) should be much more accurate.

 

[Ming the Merciless]

It must be map data as well, since it will show you upcoming gradients as well as what you’ve ridden.

Depends on the gps, not all have that feature or mapping but can still show the gradient.

 

[Ming the Merciless]

From what I understand, both barometric and GPS elevation calculation are problematic and not particularly accurate

They are pretty accurate used in combination. The problem comparing with the satellite survey values is fractals and granularity. If I walk up my steps in my garden I’ve definitely gained a couple of feet in elevation. But the data used my mapping only have values every 50m or so and won’t show that elevation difference. Similarly when when a gps records the elevation will affect the result, one gps make capture elevation at top of a hill, another may miss that and record 1m less. Thus the calculation of elevation gain and loss is problematic, but the individual values are likely pretty accurate in areas with good signals from multiple satellites.

 

[Alex321]

I thought barometer would require calibration like on an airplane where they have to set the air pressure several times during a flight otherwise the altitude is incorrect. Shrugs.

Aircraft tend to travel several hundreds of miles in one flight, which means they will be traveling trough a variety of weather systems, with differing atmospheric pressure.

That is much less of an issue on a bike - though changes in weather/atmospheric pressure probably do explain some of the variations in elevation gain on rides of a few hours.

It is also rather more important to have accurate altitude in an aeroplane - if your bike computer says you are at 1000 feet when you are actually at sea level, that shouldn't be too catastrophic, whereas with a plane it could be.

 

[Dadam]

I've got a Garmin smartwatch with (I presume) a barometric altimeter. At first it seemed quite accurate then after a year or so started to creep up or down. When I do a ride it seems to correct itself presumably from GPS data but at any given time I'm not on a ride it's often wildly inaccurate. For instance I live at roughly 475 feet but right now it's saying 22 feet.

One local ride it seemed to get the correct relative elevation, i.e the profile of the hills fitted, but on Strava the ride started at -60 feet and went up to the 120s, when it should have peaked at around 630.

 

[freiston]

I've got a Garmin smartwatch with (I presume) a barometric altimeter. At first it seemed quite accurate then after a year or so started to creep up or down. When I do a ride it seems to correct itself presumably from GPS data but at any given time I'm not on a ride it's often wildly inaccurate. For instance I live at roughly 475 feet but right now it's saying 22 feet.

One local ride it seemed to get the correct relative elevation, i.e the profile of the hills fitted, but on Strava the ride started at -60 feet and went up to the 120s, when it should have peaked at around 630.

I wonder if that could that be down to GPS reporting using WGS84 global spheroid as per the link in the post by @slowmotion above? Or maybe barometric calibration? I'm speculating in all directions without much in the way of educated guess

ADDENDUM: My phone usually misreports my altitude but gives what I believe to be reasonably accurate elevation profiles, i.e. it consistently adds or takes away to what I understand actual altitude to be. Currently it gives my altitude (on the first floor)as 170m but our house is on the 120m contour line.

 

[steverob]

I don't know if this is still the case, but certainly it was as of 2-3 years ago - even though there is a barometric sensor in all iPhones sold since about 2016, the Strava iPhone app only uses the lat/long reported by the iPhone's GPS and ignores everything else, instead choosing to calculate altitude from its own base map data.

Strava were challenged on this when it became apparent and asked why, but there was no answer - possible that they've fixed it since, but no guarantee.

 

[Dogtrousers]

Everything to do with elevation and GPS is a matter of black magic, and making the best of imperfect data.

Altitude can be worked out from GPS signals alone, but it's a bit hit and miss, as GPS is really all about giving you lat/long.
Altitude can be worked out from a barometer - but that needs to be calibrated, and can drift.
Altitude can also be derived from survey data, if you know where you are and have access to some kind of reference. This reference could be a map, or it could be the the elevations along a planned route.

Garmins (and probably other manufacturers, but I can only speak from experience with Garmins) will "auto-calibrate" the barometric altimeter by using GPS elevation and a black magic calibration algorithm.
Some Garmins can also force-calibrate the altimeter if you set a waypoint with a known elevation, when you ride past that waypoint it will calibrate to that elevation. I have several of these set on my usual ride-starting routes. Other Garmins allow you to manually calibrate.

When you download a route as a tcx file it contains elevation data of the upcoming route derived from whatever mapping software was used to plan the route. So my Edge 530 can see upcoming climbs because it knows the upcoming elevation changes along the route. But if there is no route loaded it doesn't have a clue - it doesn't use mapping data, only route data.

Some Garmins have a feature called "climb pro" that surveys your route when you load it and figures out where the hills are (it normally does a very good job of this, but can be fooled by particularly nutty routes). Then it alerts you when you start a hill and tells you how far you have to go and gradient remaining - all derived from the route elevations. This is normally pretty good, but sometimes it can beep you and tell you that you have just completed a climb while you are still struggling up to the top of a 15% gradient.

The GPS can figure out its own instantaneous gradient from altimeter and GPS data. If it has a route loaded it can also figure out a predicted gradient from route elevation. Hopefully these don't disagree too much and the Garmin "climb pro" uses a black magic algorithm to fudge this problem.

As @ColinJ says the displayed gradient lags behind reality a bit. I don't find it too bad, about 15 seconds or so.

There may be some particularly clever devices that use mapping data (not route data) to predict upcoming hills. I don't know. I don't have one.

Of course, the displayed gradient on your screen is never going to agree with the gradient that Strava or RideWithGPS tells you when you review your ride afterwards. And as we know from many other threads lots of different devices and methods and black magic algorithms will come up with different values for total climb.

Sometimes I think that rather than moaning about discrepancies and inaccuracies, we should applaud the lengths and black magic that the likes of Garmin, Wahoo, RideWithGPS, Strava and so on have gone to in order to give us some reasonably consistent data.

 

[SpokeyDokey]

I've got a Garmin smartwatch with (I presume) a barometric altimeter. At first it seemed quite accurate then after a year or so started to creep up or down. When I do a ride it seems to correct itself presumably from GPS data but at any given time I'm not on a ride it's often wildly inaccurate. For instance I live at roughly 475 feet but right now it's saying 22 feet.

One local ride it seemed to get the correct relative elevation, i.e the profile of the hills fitted, but on Strava the ride started at -60 feet and went up to the 120s, when it should have peaked at around 630.

Obviously no idea which model Garmin watch you have.

I have an Instinct and it is set to automatically recalibrate altitude via GPS at the start of each hike or bike ride.

It can also be set manually with a known value or by using the Garmin basemap (DEM - digital elevation mapping) that is built into the watch.

Seems to be pretty accurate too.

 

[Richard Fairhurst]

I noticed on the Humber Bridge ride that it thinks we cycle across the surface of the Humber rather than using the bridge!

The Two Tunnels Route in Bath (NCN 244) has a similar problem - routeplanners sometimes think you're going over the hill rather than under it! cycle.travel gets it right as does Komoot; RideWithGPS has a manual "flatten elevation" feature that you can use to correct it.