Bicycling mag vs Strava

[bartsie]

I feel that usually Bicycling is so full of infomercials and inconsequential stuff (like stories about a hot dog someone had mid-ride somewhere) that an issue can be entirely consumed during one toilet visit. Nevertheless, the latest issue contains a large well-researched piece about Strava and the protagonists of the two tragedies that everyone was talking about last year (both linked to downhill KOMs). Recommended!

There's also a Castelli ad showing David Millar on a training ride with a frame pump on his bottom tube. Bye-bye then, Rule #30!

 

[CAMPYBOB]

Does the article explain 'why' STRAVA's altitude numbers are so much higher than Training Peak, Cateye Atlas and Map My Ride?

"...an issue can be entirely consumed during one toilet visit."

You might want to use toilet paper. Just saying.

 

[bartsie]

Quote: Originally Posted by CAMPYBOB .
Does the article explain 'why' STRAVA's altitude numbers are so much higher than Training Peak, Cateye Atlas and Map My Ride?


No it doesn't. However having thought about the issue of altitude gain I'd say the number is a bit meaningless. Or ill-defined if you like. I think what matters is the comparison of various routes evaluated with the same methodology.

Originally Posted by CAMPYBOB .
You might want to use toilet paper. Just saying.


I'll give that some thought too!

 

[Volnix]

I am using Strava.... Its fun to gaze at the data after a ride whilst having a cold one... but I think the whole KOM thing its meaningless. Not that this stopped me from creating a few segments on my regular route and comparing efforts.

Unless the various efforts are made on the same exact moment, even a slight change in wind can mean 30seconds more on a segment. If not more. I wonder how they keep these world records on these big sporting events like the summer and winter games etc...

 

[CAMPYBOB]

"However having thought about the issue of altitude gain I'd say the number is a bit meaningless."

Yeah. I agree with you on the value of the actual mapped out numbers. Maybe I need to upgrade to military spec GPS! I paid money to get a bit of accuracy in those numbers and a comparative use is probably what the data is reserved for. I have zero confidence in any of the websites.

So far, MMR falls in the middle of the various values and just going by 'seat of the pants' tracking of climbs it 'might' be somewhere in the realm of reality.

I've climbed one hill on a ride that was more altitude gain than Cateye Atlas calculated over a ride with three times the elevation gain that one hill amounted to. And according to STRAVA I'm some kind of alpine climbing god. I think that may be one reason why it is so popular...that and the social aspect of it.

I'll pick up a copy of the toilet paper...er...magazine and read what they have to say about STRAVA.

And yeah, the downhill KOM thing just makes no sense me.

 

[maydog]

I think that STRAVA uses the barometric altimeter data from the GARMIN units over the USGS data over the same route. I use a GPS watch with no altimiter, so STRAVA defaults to USGS data. A a result, my rides have considerably less altitude gain than my colleagues who have altimiters.

Also, STRAVA sometimes creates bogus "climbs" due to erroneous altitude data from some riders GPS unit. There is a "category 4" climb on my way to work that is no more than 10 actual feet of elevation change.

 

[alienator]

Also note that whatever smoothing routine and settings you have on Strave, MapMyRide, your computer, and etc. can make for significant differences in altitude readings. All the different devices and programs will also have a setting for the minimum altitude change it will record, which will add yet more difference between the "altitude ascended" values. From best to worst, here is how the different altitude measuring routines rate:

1. Calibrated barometric altimetry
2. USPS coordinates
3. GPS measurements

 

[CAMPYBOB]

"Also, STRAVA sometimes creates bogus "climbs" due to erroneous altitude data from some riders GPS unit. There is a "category 4" climb on my way to work that is no more than 10 actual feet of elevation change."

Yeah, there has to be fake climbs in the mix. I'm seeing a pretty much standard spread of 3:1 between what four websites report as climbed based on the same ride data file (GPX format). Tonight's short, flat ride had one climb of 75' or so times both directions plus a few really small rollers. Altitude climbed was either 419', 797', 233' or 164'. Take your pick or throw a dart. The longer rides and rides with more climbing (say 60+ miles and a WAG of 2000'-3000' of climbing yield data spreads that are laughable. I just don't GAF at this point as the data are hardly even valid for comparative purposes.

I have...Zero Confidence in it. Zero.

So now I need to buy a watch with a barometer. Great. And calibrate the sumbitch. And throw in correction factors for rising/falling barometer and/or temp correction? Garmin theorizes a max Barometric Sensor accuracy of 10' (but will not publish an accuracy). I thought I remember reading the cheapo sports GPS was good to 5 meters (let's call it 15' for ***** and giggles). Given baro pressure fluctuations/temp changes over time there might even be an equalization between the two.

The minor speed, time and distance errors I can live with. I guess I'm just glad my old heart can still rev right up there with Collin's.

"USPS coordinates"

****...the Postal Service can't even find my road most of the time, damned lazy, over paid unionized goobermint bastards. No wonder they're going broke. Spending stamp money on mil-spec GPS gear.

Damnit! Just give me the accurate new laser wave length beer googles!

 

[bartsie]

Quote: Originally Posted by CAMPYBOB .
I have...Zero Confidence in it. Zero.


Apart from implementation bugs, it's a slippery concept at the fundamental level. For example, if you go over a speed bump, should you add its 4 inches to the altitude gain? What about the rougher pavement when you feel the bars go up and down - should those 'up' bits be added?

While at the same time some people just like riding their bike, regardless of the metrics! Lucky ba....ds.

 

[alienator]

Apart from implementation bugs, it's a slippery concept at the fundamental level. For example, if you go over a speed bump, should you add its 4 inches to the altitude gain? What about the rougher pavement when you feel the bars go up and down - should those 'up' bits be added?

No slippery slope. It's solely a matter of determining what what you need out of an altimeter and then selecting accordingly.

 

[steve]

Strava can be a bit of fun, heaps take it to serious IMO. I'm not fit enough to even think about getting a KOM

though fat enough still to have a chance at getting some KOD's

Tailwind dopers ruin the fun for us honest strava users.

 

[maydog]

The barometric altimeters give the most optimistic elevation gain results. Their results are about 1.5 to 2 times that of the USGS data. Things get really screwy when a front passes through during your ride.

 

[alienator]

The barometric altimeters give the most optimistic elevation gain results. Their results are about 1.5 to 2 times that of the USGS data. Things get really screwy when a front passes through during your ride. 

That's the rare occasion. The problem with USGS data is the spatial frequency of altitude measurements is sparse enough that a lot interpolation has to be done between measurement points. It's in the interpolation--along with data smoothing and minimum altitude change recorded--that error creeps in. When I've passed elevation signs, my Garmin's barometric readings have been a close match. For GPS altitude, the greatest error in GPS coordinates is in the elevation component.

 

[Subliminal-SS]

I Found interestingly yesterday on the subject of Strava vs others.

Strava always added about 2k feet elevation gain on average. I have the Garmin 810 and certain formats are written to without any smoothing whereas the others like the 800 do have smoothing in the raw data.

If I upload my ride to Garmin Connect it looks good and correct in comparison to my Device ride summary and my knowledge of the area. If I then export a GPX and upload to Trainngpeaks I get one elevation set and then Strava another WILD elevation set.

However if I upload all directly from the device all the data is correct and good for me personally and its all pretty damn accurate. my TP, Garmin Connect and Strava all finally read the same elevation gain.

I would say the .fit files in Garmin are a good standard to work with however the different applications interpretation of GPX files leaves a little to be desired.

 

[CAMPYBOB]

"However if I upload all directly from the device all the data is correct and good for me personally and its all pretty damn accurate. my TP, Garmin Connect and Strava all finally read the same elevation gain."

I upload direct from the Cateye Stealth 50 to Training Peaks, STRAVA and Cateye Atlas. The three results vary wildly. Map My Ride gets a GPX file that I upload from Stealth to my desktop to the website. Oddly, MMR seems to have the most accurate results for climbing just doing quick climb height additions on the road and after the ride. MMR is generally neither the highest reading or the lowest.

"I would say the .fit files in Garmin are a good standard to work with however the different applications interpretation of GPX files leaves a little to be desired."

Thanks. I will try exporting the ride data from the Stealth in FIT format and uploading to the various websites to see if correlation improves.

 

[CAMPYBOB]

"Apart from implementation bugs, it's a slippery concept at the fundamental level. For example, if you go over a speed bump, should you add its 4 inches to the altitude gain? What about the rougher pavement when you feel the bars go up and down - should those 'up' bits be added?"

I'm not so much interested in GPS resolution and/or wringing every inch out of my climbing...trust me, we do enough climbing around here that we don't need to lie about it on STRAVA...as I am just getting a number that I have even a small bit of confidence in.

Did I climb 2600' or 3000'? Who cares? What I DO care about is wondering if it was actually the 1200' Cateye Atlas tracked, the 842' Training Peaks graphed out like it was chiseled in granite or the 2937.43' STRAVA crowns me with or the middle-of-the-road reading MMR spits out? Like I said, a 2:1 to even 3:1 data swing is just insane.

You would think that the various websites, getting the same data, would at least be 'close'.


"While at the same time some people just like riding their bike, regardless of the metrics! Lucky ba....ds."

I hear ya, bartsie!!! I left the fork sensor for my Cateye Strada Wireless on the bike, the computer mount is interchangeable between the Stealth 50 and the Wireless and I'm THIS close to removing the ANT+ rear speed/cadence sensor, stashing the heart rate strap back in the box with the Stealth 50 and going back to the Wireless. Speed & Distance worked for me for decades...and no, I was never interested enough in absolute accuracy to do a circumference rollout (with rider on board)! Oddly, among 5 or more odometer readings at the end of group rides they were all 'close' to agreement.

 

[CAMPYBOB]

"The barometric altimeters give the most optimistic elevation gain results. Their results are about 1.5 to 2 times that of the USGS data. Things get really screwy when a front passes through during your ride."

That's what the owners tell me. When I hear runners *****ing...

I'll get a barometer based watch. I've already been comparing ride data with a training partner's Garmin. Not enough data yet to establish a 'who is right and who is wrong' opinion, let alone fact.

 

[bartsie]

Quote: Originally Posted by CAMPYBOB .Oddly, among 5 or more odometer readings at the end of group rides they were all 'close' to agreement.

There was this incredibly amazing, er... person of the opposite gender on the last group ride and thanks to Strava's 'rode with...' I know her name.

Downhill KOMs, especially on segments that go right through stop signs? Thanks but no thanks.

 

[CAMPYBOB]

"...and thanks to Strava's 'rode with...' I know her name."



Dude...can you...uh...post pics of her...er....VAM numbers?

 

[CAMPYBOB]

A .FIT vs. .GPX test uploaded to Map My Ride yielded the same altitude/vertical climbing numbers.

STRAVA also gave me the exact same feet of climbing numbers. In this case, a direct from Cateye Sync software out of the Stealth 50 to STRAVA and a .FIT file uploaded from one of my desktops.

The difference in climbing stats was 1459' for STRAVA vs. 440' for Cateye Atlas over a short, rolling 35-mile ride. That is more than a 3:1 spread in the climbing measurements as graphed by the websites.

Confidence rating is still zero in any of the altitude numbers I'm seeing.