CYCLINGFORUM.COM - Where Cyclists Talk Tech --- Return To Home

 

    Register FAQ'sSearchProfileLog In / Log Out

 

****

cyclingforum.com ****

HOMECLUBS | SPONSORS | FEATURESPHOTO GALLERYTTF DONORS | SHOP FOR GEAR

Return to CyclingForum Home Page CYCLING TECH TALK FORUM
          View posts since last visit

elevation accuracy of mapping softwares?
 

Author Thread Post new topic Reply to topic
April
Joined: 13 Dec 2003
Posts: 6593
Location: Westchester/NYC

8/4/13 6:18 PM

elevation accuracy of mapping softwares?

I mapped a route using ridewithgps.com. It told me it's got 3800' of climbing.

After the ride, one of the rider send me his strava data. It's reporting only 3300' of climbing.

That's a 10% difference. And will make comparison between rides really hard to make. Rumor has it ridewithgps tend to over estimate the elevation data, which is good for inflating egos. Based on that rumor, I'm incline to believe in Strava's number. But if a rider ask me if my ride is more or less hilly than another one that has 3500' of climbing, not know how that elevation data comes from will completely skew the answer to the point of useless!

So, does anyone use altimeter and have comparison information regarding the different mapping software in their accuracy of elevation data?

 Reply to topic     Send e-mail

KerryIrons
Joined: 12 Jan 2004
Posts: 3234
Location: Midland, MI

8/4/13 6:58 PM

Data base issues

There can be a lot of data base issues in mapping software. If the route is very "rolling" then one package might not pick up the 50 foot elevation changes. It saves them effort to not have to calculate to that level of granularity. Since none of these companies provide the basis for their calculations it's very hard to figure out which one is most accurate. Look at how bad they are with calorie numbers.

 Reply to topic    

Nick Payne
Joined: 10 Jan 2004
Posts: 2625
Location: Canberra, Australia

8/5/13 12:47 AM

I've found, with my Garmin, that there is a considerable difference in elevation gain/loss between what the unit on the bike reports and what connect.garmin.com reports once the data is uploaded.

I've also noticed that if I create a kml file from a Garmin route, import it into Google Earth, and then tell Google Earth to show the elevation profile of the route, that some of the gradients that GE shows me are far in excess of reality - eg the ride we went on today has a max gradient of 8-10% in a couple of places, but GE is telling me that there are sections where the gradient hits 15% or 16%.

 Reply to topic    

dan emery
Joined: 11 Jan 2004
Posts: 6884
Location: Maine

8/5/13 4:27 AM

I've wondered too

I just have a Garmin 200 and it only took me about a year to realize that it had an elevation function (on the third screen in the "history" section of the unit, I don't upload anything). I've noticed that the ascent and descent figures are slightly different. Thanks to the erudite Kerry, I figure that has to do with the "granularity."

I'm guessing the reading is not all that accurate, but that's just a guess.

 Reply to topic    

ErikS
Joined: 19 May 2005
Posts: 8337
Location: Slowing boiling over in the steamy south, Global Warming is real

8/5/13 5:03 AM

GPS is not very good at elevation. It is a known issue with the system. I use Strava every ride and see fluctuations in altitude climbed over the same route ridden on different days.

I have also seen about a 1% difference between receivers and users.

 Reply to topic     Send e-mail

Tim123
Joined: 01 Apr 2004
Posts: 252
Location: Adelaide

8/5/13 5:09 AM

The Garmin has an inbuilt barometric altimiter which by and large is fairly accurate (as long as a large weather fromt doesnt roll through mid ride. There is a loop I do regularly that mine and everyone elses Garmin registers around 580m, once loaded into Strava or Garmin, if you apply altitude "correction" it adds around 300m to the altitude gain. It also shows many more ups and downs and downhills on roads that have absolutely no downhills whatsoever. I believe the mapping data that these corrected altitudes are based on are known altitudes nearby, but may not actually be on the route ridden. Unfortunately iPhone and Android users get the inflated altitude by default as there is no other reference other than the Strava mapping data.

Good for the ego and stats though.

 Reply to topic    

dfcas
Joined: 11 Jan 2004
Posts: 2815
Location: hillbilly heaven

8/5/13 6:40 AM

Ive had a Garmin 200, 500, and 510. The 200 does NOT have a barometric altimeter and uses GPS to measure altitude. I did a dead flat rail trail ride and it indicated 700 feet of elevation gain. If you are on roads in the database, when you upload the data you can have Garmin correct the elvation stas. The gravel roads and such are not in the database.

The 500 and 510 both have barometric altimeters and seem to be fairly accurate.

My old Avocet 50 was the first bike computer I had with altimeter function and I believe it was the best. It did not start counting elevation gained until you climbed 30 feet continuosly, with the idea being your momentum would carry you over a slight roller with little effort.

I'm currently using a Sigma 14.12alti which seems pretty accurate. I think I like it better than the 510, but I dont care about mapping.

 Reply to topic     Send e-mail

April
Joined: 13 Dec 2003
Posts: 6593
Location: Westchester/NYC

8/6/13 7:44 AM

How about distance accuracy? Does most mapping software seem accurate?

That is something that can be checked rather accurately with a well-calibrated cyclocomputer. Except I haven't re-calibrated mine after a battery replacement...

So at the end of the ride, I got all kind of result from my group and none agrees with what the software (say, a mile for a 50 mile ride). So I wonder if people didn't set up their computers right? Or the mapping software aren't all that accurate after all.

(contrast that with google direction for cars, the mileage seem to agree quite accurately)

 Reply to topic     Send e-mail

ErikS
Joined: 19 May 2005
Posts: 8337
Location: Slowing boiling over in the steamy south, Global Warming is real

8/6/13 9:14 AM

Bike computers that are set up on a wheel magnet are only as accurate as the install and the numbers people enter into them. If the person who installed them did a weighted roll out on the bike and used the mm you get from that to set up the computer they can be very accurate. If they just used the numbers provided in the instructions then odds are then speeds and distances will be inflated.

For our needs, anything less than a 1% error factor should be okay.

I will say my bike computer seems to read just a bit high but I have been to busy to push it down by a couple of mm and get it dialed in closer. I am comparing it to the GPS on my phone and the Strava data.

I think GPS can be so accurate that it sees your path on the road, and how you corner. Take the outside line all the time and you will have a higher distance than your riding partners.

Once again, less than 1% for our use is good enough. Anything more is statistically insignificant.

Elevation, is not this accurate.

 Reply to topic     Send e-mail

KerryIrons
Joined: 12 Jan 2004
Posts: 3234
Location: Midland, MI

8/6/13 6:53 PM

Distance accuracy

As Erik says, bike computers are only as accurate as their calibration. If the wheel circumference is 3 digits (like 210) then you can get 0.5% accuracy. If the circumference number is 4 digits (2110) then obviously you can get 0.05%. To get that level of accuracy you really are best to calibrate against roadside mile posts. Those are put in place by surveyors and are quite accurate. If your ride at least 10 miles and compare your computer reading against the mile posts you can correct the mileage. Just be careful because some computers continue to calculate after you stop and so the number can change a bit.

Regards GPS accuracy it mostly depends on how "turny" your route is and how frequently the unit updates. Lots of turns and infrequent updates means that the unit is striking a straight line through your turns and so cutting off some distance. Impossible to predict the effect without knowing the update frequency and the nature of the turns.

 Reply to topic    

ErikS
Joined: 19 May 2005
Posts: 8337
Location: Slowing boiling over in the steamy south, Global Warming is real

8/7/13 4:30 PM

A little experiment.

http://app.strava.com/activities/72912205

Three devices used. My bike computer is a little fast. I need to bump it down a few mm.

 Reply to topic     Send e-mail

April
Joined: 13 Dec 2003
Posts: 6593
Location: Westchester/NYC

8/7/13 4:38 PM

ok, so most riders computers tend to be on the high side. That explains why my average speed was always lower than the group even though we rode practically together. Doesn't quite solve the mystery as whether theirs are right (and mine low), or mine is correct and theirs too high though.

So I wonder how organized events base their cue sheets on. I've noticed I often disagree with the cumulative mileage on long rides. And it's not always err on the same side either! This didn't used to be the case BEFORE all the online mapping websites become available...

 Reply to topic     Send e-mail

Sparky
Joined: 08 Dec 2003
Posts: 19068
Location: PDX

8/7/13 4:45 PM

"So I wonder how organized events base their cue sheets on"

When I was a ride leader in NJ with the BTCNJ, pretty much by car odometer I was told. Granted some of the que sheets where a decade or more old, so who know initially vs updates. But with construction and changes over the years, I am thinking it may be the odometer of the ride leader on the updates being likely perhaps...

 Reply to topic     Send e-mail

KerryIrons
Joined: 12 Jan 2004
Posts: 3234
Location: Midland, MI

8/7/13 6:57 PM

Car odometers

Car odometers are a whole different story. When I used to travel a lot for business I would get bored on long freeway drives and check the odometers of the rental cars. Range was dead on to 2.5% high (5 times the error obtainable if you calibrate a bike computer against mile posts).

Our local rail trail has new mile posts put up by the county parks department and most of the original railroad mile posts. I assume the parks department posts were "calibrated" with a county pickup truck, because they are consistently, mile after mile, shorter than the mile posts put up by the railroad. Since I know the RR posts were done accurately by surveyors when the line was laid, I know that the county truck odometer read high just like so many rental cars I drove. We won't even discuss why the parks department couldn't just match the original mile posts (their zero-mile reference point was different than the start of the rail trail by 20 miles so they wouldn't have the same mileage but they certainly should match).

Our current Subaru Outback odometer is WAY off (6% low). I called Subaru thinking it was some sort of calibration failure and they tell me you can't calibrate their odometer and that it is "within spec." Nonsense.

 Reply to topic    

ErikS
Joined: 19 May 2005
Posts: 8337
Location: Slowing boiling over in the steamy south, Global Warming is real

8/8/13 6:12 PM

Our Tahoe reads low, even with stock tires. The Mazda is spot on.

 Reply to topic     Send e-mail

Steve Katzman
Joined: 12 Jan 2004
Posts: 452
Location: Orlando, FL

8/16/13 1:22 PM

Topo USA

I use an old version of DeLorme Topo USA to make the elevation profiles and the climbing data for our local Orlando club's annual century, "The Horrible Hundred". According to the software, it reports a total climbing for the century to be very close to 5000 feet. Friends with GPS have come back and said they measured between 3500 and 4000 feet.

Not sure who is right, but I would tend to believe the software, which uses a topographic map to calculate elevation.

 Reply to topic     Send e-mail

Sparky
Joined: 08 Dec 2003
Posts: 19068
Location: PDX

8/16/13 2:45 PM

"reports a total climbing for the century to be very close to 5000 feet"

The GPS may not add the cumulative climbs. All the little ups added without subtracting the down undulations nets more climbing than the altitude difference, yes?

 Reply to topic     Send e-mail

dan emery
Joined: 11 Jan 2004
Posts: 6884
Location: Maine

8/16/13 4:20 PM

GPS altitude

Recently I've looked at a couple posted D2R2 GPS files which have climbing figures that look to me totally goofy (way low). I figure like a lot of other things, you get a number, but whether that number means anything is highly debatable. I think you'd probably get better information by having an experienced rider rate it 1-10 based on perceived exertion.

 Reply to topic    

Steve Katzman
Joined: 12 Jan 2004
Posts: 452
Location: Orlando, FL

8/16/13 5:00 PM


quote:
The GPS may not add the cumulative climbs. All the little ups added without subtracting the down undulations nets more climbing than the altitude difference, yes?


Yes, I think you've got it. If the GPS calculates based on a finite number of data points per mile, then the ups and downs ridden between those points is effectively ignored, which would make a road with short rollers record low. I would think that a GPS would be more accurate on a long steady climb or a long steady descent.

 Reply to topic     Send e-mail


Return to CyclingForum Home Page CYCLING TECH TALK FORUM
           View New Threads Since My Last Visit VIEW THREADS SINCE MY LAST VISIT
           Start a New Thread

 Display posts from previous:   


  
Last Thread | Next Thread  >  

  
  

 


If you enjoy this site, please consider pledging your support

cyclingforum.com - where cyclists talk tech
Cycling TTF Rides Throughout The World

Cyclingforum is powered by SYNCRONICITY.NET in Denver, Colorado -

Powered by phpBB: Copyright 2006 phpBB Group | Custom phpCF Template by Syncronicity