Variations in mapping websites

dc2

I plotted a route in Wicklow using both Map My Ride and Ride with GPS and total elevation is completely different - see links:

http://www.mapmyride.com/routes/view/134089627

http://ridewithgps.com/routes/1687875

Anyone know why there are such discrepancies and which one is correct?

Sundy

Neither is correct. Both are inaccurate due to the elevation model used.

dc2

Sundy wrote: »

Neither is correct. Both are inaccurate due to the elevation model used.

So what elevation models are they using? Is there any mapping site using an elevation model which will give accurate results?

smacl

I'm guessing both are using SRTM data for base elevations, though the ride with GPS profile looks more detailed than the map my ride one, suggesting it contains more data and hence might be a bit more accurate. The base data isn't really good enough to give total cumulative height ascended to a reasonable degree of accuracy, as the amount of height error at each interpolated sample point is liable to be comparable to the height difference between it and the next nearest point on flatter areas. To get around this, different mapping programs use different smoothing algorithms, hence the large discrepancy in the results given.

If you can get something based around OSi models, you're accuracy will improve as it is significantly better base data. I don't think this is available on any free sites at present.

rab!dmonkey

Here's an article from Strava on how they calculate elevation. It seems the only realistic way to improve the accuracy of this measurement is to invest in a device with a barometric altimeter.

TheBlaaMan

rab!dmonkey wrote: »

Here's an article from Strava on how they calculate elevation. It seems the only realistic way to improve the accuracy of this measurement is to invest in a device with a barometric altimeter.

And since barometric altimeters in most consumer GPS is decidedly inaccurate (IMO, based on Garmin 500), even this approach is going to have a fairly large margin of error. Any one of them is a good approximation, with RidewithGPS better than MmR in my experience and my GPS data probably best of all

dc2

TheBlaaMan wrote: »

Any one of them is a good approximation, with RidewithGPS better than MmR in my experience and my GPS data probably best of all

Probably a stupid question but when you say "my GPS data" are you referring to your GPS unit or a website called "my GPS". I couldn't find any site called that.

TheBlaaMan

Nah, just the data from my Edge500.

Specifically, I'm referring to measurements like elevation compared against OSi maps etc and more subjective aspects such as maximum gradient on a climb. For both of these, Ridewithgps and mapmyride can be 'off' to varying degrees

blorg

TheBlaaMan wrote: »

And since barometric altimeters in most consumer GPS is decidedly inaccurate (IMO, based on Garmin 500), even this approach is going to have a fairly large margin of error. Any one of them is a good approximation, with RidewithGPS better than MmR in my experience and my GPS data probably best of all

If it is decidedly inaccurate after giving it time to calibrate it is broken and you should get it replaced under warranty. I've had it break in Garmins (probably due to moisture, or the thermometer breaking) and become inaccurate, but when it is working it is accurate down to a couple of metres; often it is bang on.

TheBlaaMan

Thanks Blorg. Yeah, I've though of this.....and posted elsewhere about my own Edge500.

It generally is fine, but sometimes it simply will not either pick up elevation points or for some other reason goes 'postal' for a day or two before righting itself and being fine again for a few months.......too much bother for a return for such an intermittent issue, so I live with it.

Sorry, OP...off topic!

smacl

TheBlaaMan wrote: »

And since barometric altimeters in most consumer GPS is decidedly inaccurate (IMO, based on Garmin 500), even this approach is going to have a fairly large margin of error. Any one of them is a good approximation, with RidewithGPS better than MmR in my experience and my GPS data probably best of all

If you're talking about inaccuracy, it's worth considering the types of error involved, specifically differentiating between absolute and relative errors. Absolute errors on the barometric altimeter on my Garmin 76cs are quite high. I've observed 20m height differences reported by the unit sitting on my desk over the course of 24 hours. However, relative height differences (ascents or descents) measured over short periods tend to be quite good. Garmin state that the accuracy of their barometer is +- 10 feet when calibrated, which is really just a statement of relative accuracy, where absolute accuracy could be ten times that. Commercial grade autonomous GPS, such as in the Garmins, have much better absolute accuracy, but worse relative accuracy due to various interference factors. Depending on the unit, and where you are in the world, you may also get corrections such as EGNOS or WAAS which will dramatically improve the accuracy of GPS. Using web based Mapping/DTM products, such as MapMyRide etc... have absolute error only, and unless the programmers tweak the algorithms in use, provide a repeatable result.

Bottom line, if I wanted to measure cumulative height gained over unknown terrain, I'd use a barometer, which I'd calibrate where possible by re-visiting the same point more than once. If I wanted to know my elevation and was unsure when by barometer had last been calibrated, I'd use a GPS. If I wanted to measure my change in performance over the same course over time, I'd use a mapping product. More related ramblings here.

Diarmuid

Surely the Garmins with the barometric altimeters use the GPS to calibrate the absolute elevation in the barometric altimeter regularly?

smacl

Diarmuid wrote: »

Surely the Garmins with the barometric altimeters use the GPS to calibrate the absolute elevation in the barometric altimeter regularly?

Maybe, but I'd doubt its something they'd do automatically given the nature of random errors in GPS. You're better of calibrating against a good known elevation, such as an OSI trig piller or statically observed GPS point.

Diarmuid

smacl wrote: »

Maybe, but I'd doubt its something they'd do automatically given the nature of random errors in GPS. You're better of calibrating against a good known elevation, such as an OSI trig piller or statically observed GPS point.

yeah but who is going to do that on every ride?

The random errors in GPS are relatively low. Modern GPS chips do a CEP of 2.5m.

So you can calibrate the absolute gps elevation pretty well with gps but then switch to the baormetric altimeter for the recording of the ride

smacl

If the Garmin GPS was delivering an absolute vertical accuracy of 2.5m, why add a barometer that has a claimed accuracy that is lower (10 feet = 3.048m), and is prone to drift?

Edit: The following Garmin faq may be of benefit. From that faq;

The Edge 500 calculate elevation data using a barometric altimeter as opposed to satellite data. Once the device is properly calibrated, the standard area of accuracy will be +/- 50 feet. To ensure that elevation data is accurate at the beginning of an activity you have the ability to calibrate the elevation manually.

I'm guessing this +/- 50 feet is absolute accuracy, whereas the stated +/- 10 feet is relative accuracy.

Diarmuid

smacl wrote: »

If the Garmin GPS was delivering an absolute vertical accuracy of 2.5m, why add a barometer that has a claimed accuracy that is lower (10 feet = 3.048m), and is prone to drift?

the chip in the garmin is probably not as accurate as the ublox but the real reason to add a barometric altimeter is to get more accurate sequence of elevation points. If you record 10 points in a second, while standing still at 100m with a CEP of 5m, your elevation could look like 103,95,105,100,90,98,102. etc... For a calibrated barometric altimeter it would be 100,100,100,100 etc... (If it wasn't calibrated it could be 70,70,70, etc..)
If you get the cumulative elevation gain/loss for both, they are very different, but the barometric one is much more accurate, even if it wasn't calibrated.