IM Louisville Climbing - Bike Course GPS Data

I’m going to just have to plain disagree with you here.

The vertical resolution of a GPS is FAR worse than the X-Y. Not sure what the factor is but it is worse. Considering that the GPS itself can be off 10-15ft in the XY the vertical will be off more than that. Now 10-15 ft is no biggy in the XY as most of our movement is in a realatively stright direction so 15 feet to the left and then 200ft later 15 ft to the right doesn’t significantly alter the total distance traveled. However in the Z axis that is a HUGE difference and over the course of 100 miles can make a major difference. Now factor in that the accuracy in the Z is worse than that in teh XY due to the “Depth perception” issue and it’s easy to see that vertical GPS data is pretty inaccurate.

Now the above is worst case scenario and certainly your not going to get that kind of deviation all the time. But the point is you never really know when it’s right and when it’s wrong.

For the most part you’ll probably get decent match up from a GPS and TOPO although I’ve seen 1500-2000 off from several people using GPS in a couple IM courses.

Also not saying that TOPO is without it’s flaws as it certainly isn’t perfect, just more accurate overall, IMHO.

~Matt

Just checked the Garmin EDGE 305 not the forerunner has altimeter none of the ones I have includeing the forerunner 305(not the edge) has altimeter.

Imusa 2004 I ended up with 7300 of climbing BUT I had to subtract all the spurious data for 2 and half hours while my bike was in transition I had to leave it on my bike and powered on not stopped while I was doing the swim… While it’s setting there the numbers will fluctuate wildly while setting still(problem with gps if your standing still and don’t autopause it goes all over the place but on average stays in same location. It showed 9800 of gain just setting there… SO AGREED if your standing still you will get that.
Once moving it’s at it’s best because can use averageing and filling in blanks and missing trackpoints along the way.

Once I subtract that(just looked at profile topusfusion I move the curser along it and the map changes so I moved until I actually started moving. Then at end of profile total climbing was 17100 feet subtract out the noise in beginning when wasn’t moving and I get 7300 feet…

Florida course I have 3447 feet with no filtering. .Even though it’s flat it does have some minor ups and downs and over 112 miles they add up.
But if you look at the profile see the highest points(sane points filtering out those wierd peaks) the highest point lowest points
will all be pretty close to what they actually are.

For florida here are multiple climbing reports using the tracks but different algorithms
TopoFusion: 3618 ft
Straight GPS: 3618 ft
Straight DEM: 1830 ft
Maptech: 482 ft
TopoUSA: 2000 ft
Custom Algorithm: 1830 ft

Ironman wisconsin in 2004 I have 6335 feet in climbing
using different algoritms in topofusion I get these.

TopoFusion: 6335 ft
Straight GPS: 9036 ft
Straight DEM: 7056 ft
Maptech: 4897 ft

Anyway the over all profiles match whats online and what others have now the actual climbing numbers again depend on how you calculate it…
I’ve wasted to much time on this gotta get out and bike.

gholmes
TopoUSA: 7143 ft
Custom Algorithm: 7056 ft

I’m going to just have to plain disagree with you here.

The vertical resolution of a GPS is FAR worse than the X-Y. Not sure what the factor is but it is worse. Considering that the GPS itself can be off 10-15ft in the XY the vertical will be off more than that. Now 10-15 ft is no biggy in the XY as most of our movement is in a realatively stright direction so 15 feet to the left and then 200ft later 15 ft to the right doesn’t significantly alter the total distance traveled. However in the Z axis that is a HUGE difference and over the course of 100 miles can make a major difference. Now factor in that the accuracy in the Z is worse than that in teh XY due to the “Depth perception” issue and it’s easy to see that vertical GPS data is pretty inaccurate.

Now the above is worst case scenario and certainly your not going to get that kind of deviation all the time. But the point is you never really know when it’s right and when it’s wrong.

For the most part you’ll probably get decent match up from a GPS and TOPO although I’ve seen 1500-2000 off from several people using GPS in a couple IM courses.

Also not saying that TOPO is without it’s flaws as it certainly isn’t perfect, just more accurate overall, IMHO.

~Matt

Well I’m gonna agree with you and disaggree maybe my point isn’t being made… Any instantaneous(say 1 second sample)
reading on the gps is just as you say it will be off… And if you use those values to get total climbing you will get some way off numbers… But filtering is the key as it is with any measurement with noise…averageing things out which the garmin will do (you don’t see it change every second as the 305 does or even 5 seconds as the others do depending on how you set it)
you will see the average… If you did you’d see it fluctuate all over the place especially in the z(I agree it’s the least accurate of all) But over a longer period and being filtered it will suffice… (WHO needs one or even 20 feet accuracy anyway except
when you are doing the overall climbing calc) I just know for the most part when doing long climbs or driving over varied terrain where they have elevation signs it’s been right on unless I stop and stand around a bit and it will fluctuate some.

MY Point is for getting an overall profile map of a course knowing the high spots low spots with some descent resolution
It is very useful… I’ve got 100’s of tracks like I said and over the same courses where there are some good hills and high and low points(not every foot like on the trails) my readings are consitant across all of them meaning the one from 1 year ago is very close to the same as the one last week on same route… it’s consistant… MY MAIN Point is it’s not USELESS
as Cousin keeps saying if it were then would be no reason to use them…

I agree altitude is the lesser of the readings in accuracy but overall it’s not useless…

I got to get off this computer and go ride guys… I’ll keep using my gps you use what you use doesn’t matter to me…

Later…

OK, let’s see if I can make this clear (and maybe we’re just speaking different languages). I’m talking about GPS course profiles,elevation gain, etc.
Here’a a track workout, where the elevation shouldn’t have deviated by more than a few inches and yet Garmin shows over 60 feet. This is a brand new artificial surface - rubberized trak and synthetic football turf) so it’s not like it might have settled and might be all out of wack) Additionally, this was 14 laps around the track (plus some drills on the football field) so why don’t we see a pattern?
http://i11.tinypic.com/5xi0mm0.jpg
And here’s the Lake Merrit run, which goes around the shore of a small lake in downtown Oakland, and I’d guess the elevation variance (highest to lowest points on the course) are within 10 feet.
http://i14.tinypic.com/5xpwfop.jpg

Now here’s the same course run in warmup just before the race? See the similarities? Good, 'cuz neither do I.
http://i12.tinypic.com/6747x41.jpg

Definetly on a track or mowing your yard(I’ve tried this) in a confined short space with turns your gonna get this…
I talking over a 112 miles with the data filtered properly… Also running seems show lot’s more jumping around of the data than
riding or driving not sure if it’s because of the distance covered per sample(part of it) or what…

But Agreed if you do a short track or confined area it’s not gonna look good plus garmin training center doesn’t do a good filter on the data… I just know when I’ve used topfusion(I’m not selling it don’t even know if it’s still available I downloaded it years ago so I could overlay aerial photos on my tracks) does good job filtering the data over a longer faster course.
I’ve got running routes though that seem to be consitant…
Other factors are tree coverage buildings… Depends on time of day where the sattelites you last aquired have moved to and where the new ones you pick up are. lot’s of factors…

when I get back from my ride will go find some of the courses I’ve done more than once like imf and hawaii with a gps and see how they look…

Anyway gotta go bike…

“Just checked the Garmin EDGE 305 not the forerunner has altimeter none of the ones I have includeing the forerunner 305(not the edge) has altimeter.”

    • That would explain why my Forerunner gives me such bad data and my Polar gives vastly superior info.

“It showed 9800 of gain just setting there…”

    • At which point I point to my earlier statement about GPS elevation data being worthless, and I buy you a beer…

“Once moving it’s at it’s best because can use averageing and filling in blanks and missing trackpoints along the way.”

    • When you find yourself in a hole, stop digging. Your 9800 feet of elevation gain sitting still for for an hour pretty well seals the deal.

“Once I subtract that(just looked at profile topusfusion I move the curser along it and the map changes so I moved until I actually started moving. Then at end of profile total climbing was 17100 feet subtract out the noise in beginning when wasn’t moving and I get 7300 feet…”

    • Which your saying agrees with Topo, which I pointed out in an earlier post is extremely inacurrate as well.

“Florida course I have 3447 feet with no filtering. .Even though it’s flat it does have some minor ups and downs and over 112 miles they add up.”

    • Not to 300 feet it won’t. Silverman is only 10,000, and if you think it’s only 3X Florida, then we don’t need to continue this discussion. I’ve done rides with Polar and Garmin (sorry not on the same day) and gotten 7800 feet on the Polar and +15,000 on the Garmin. There are probably some games you can play to make that work, but the data at any point is going to be pretty darned worthless. It just has to be.

“But if you look at the profile see the highest points(sane points filtering out those wierd peaks)” the highest point lowest points will all be pretty close to what they actually are.

    • No on the ones I posted just above from perfectly flat courses

"For florida here are multiple climbing reports using the tracks but different algorithms
TopoFusion: 3618 ft, Straight GPS: 3618 ft, Straight DEM: 1830 ft, Maptech: 482 ft, TopoUSA: 2000 ft, Custom Algorithm: 1830 ft

    • I’d say Maptech sounds like the only one in the ballpark.

“Anyway the over all profiles match whats online and what others have now the actual climbing numbers again depend on how you calculate it…”

    • And the course profiles on line are usually from Topo and again, far from accurate.

“Definetly on a track or mowing your yard(I’ve tried this) in a confined short space with turns your gonna get this…
I talking over a 112 miles with the data filtered properly… Also running seems show lot’s more jumping around of the data than riding or driving not sure if it’s because of the distance covered per sample(part of it) or what…”

    • Could be the motion and rotation of the unit while you run. I’ve noticed that running data is more hideous than biking, although both are still way too wrong to be used for anything.

“when I get back from my ride will go find some of the courses I’ve done more than once like imf and hawaii with a gps and see how they look…”

    • That would be interesting

"Anyway gotta go bike… "

    • Yeah, I’ve gotta do chores and clean out the garage. That’s why I’m still sitting here.

As a fact point: there is little to no problem is establishing the algorithm for giving “good” altitude gained, it is simply
that Avocet has the patent on the algorithm when they developed the Avocet 50, a number of years ago. With Avocet out
of business, nobody has really had a good altimeter with cum altitude gained.

While GPS has some problems with getting altitude changes very accurate (for instance, vertical height with WAAS is lucky to pick up 20-30 ft),
a barometric GPS (think Garmin **Edge **205, 305, or Vista) can pick up a 15 ft change nicely.) Sometimes the barometric method has its own set of
problems since it depends on a given air temperature for air density, thus very hot and very cool days can also screw you up.

A bigger problem with “flat” course is that they are not.

Most people simply cannot see low degree inclines, however, without them we’d all be facing very unpleasant pooling of water in our local
neighborhoods and lots more mosquitoes. Town are built to make sure roads slope and lead to collection sites. So, you are gaining and
losing elevation and not knowing it even on flat courses.

However, low degree inclines aren’t really going to be seen by the body either. The additional input of going up a 1 degree incline will be approximately
1.7% of your weight. Thus a 200 lb rider will feel 3.5 lbs pulling back on them. The uphills won’t slow you down (much) and the downhills don’t
speed it up (much).

Avocet recognized this, and as Jobst Brandt likes to point out, Avocet said "well we just won’t register a climb if it is less than 10M or roughly 30 ft.
Thus using the Avocet 50, you could ride that pancake flat course, which may never vary more than 20 ft up and down, and you would see
zero feet gained. The Avocet, while maybe a bit conservative, felt right to many of us old timers.

Jobst had claimed that nobody wanted to violate their patent, so nobody has this algorithm in their altimeter. Topfusion, however, allows you to
“dial in” a number from zero to “what-ever” like the old Avocet, and thus you can recreate an “on board” Avocet with their “off board” software.

See picture below for a comparison of the raw data, GPS map data, and a filter set to 10M like the Avocet.

Topofusion probably has the best assortment of ways to figure out altitude gained. The custom Algorithm is the Avocet algorithm.

http://img484.imageshack.us/img484/4450/climbko3.jpg

So, now the question “how do I get data into TopFusion.” You want to dump your data from your Garmin into a GPX file for pre-existing
rides. If you have something like Mapsource this is a “save as” feature to save as GPX.

TopoFusion will also import most Garmin type GPS through USB/Serial directly.

Theo

“Definetly on a track or mowing your yard(I’ve tried this) in a confined short space with turns your gonna get this…
I talking over a 112 miles with the data filtered properly… Also running seems show lot’s more jumping around of the data than riding or driving not sure if it’s because of the distance covered per sample(part of it) or what…”

    • Could be the motion and rotation of the unit while you run. I’ve noticed that running data is more hideous than biking, although both are still way too wrong to be used for anything.

“when I get back from my ride will go find some of the courses I’ve done more than once like imf and hawaii with a gps and see how they look…”

    • That would be interesting

"Anyway gotta go bike… "

    • Yeah, I’ve gotta do chores and clean out the garage. That’s why I’m still sitting here.
      Yep I’m back real quick I’ve heard discussion on the rotating and bouncing up and down maybe but still just noise and a good filter can filter that out.

When I get back I’ll get my run data from races(ironman) I’ve done where I wore my polar(with altitude) and my garmin which are most of my races over last 5 years or so and compare them with the online profiles… Would be interesting to see how close…

Most of my talk had to do with bikeing and bike profiles since the original topic was louisville and comparing climbing.
New post above me(I’ll read more when I get back) seems to explain bit more from someone else whos used topofusion.

outta here to bike(2 hours instead of my planned 5 hours so I’m using slowtwitch as my excuse…)

Again if the original data is non repeatable and teh noise ratio is to high no level of filtering will get you to a good answer. Sure you can run the numbers through any various numbers of filters and without a doubt one of them is going to come up with a good number…but you never know which one.

For instance from your post.

For florida here are multiple climbing reports using the tracks but different algorithms
TopoFusion: 3618 ft
Straight GPS: 3618 ft
Straight DEM: 1830 ft
Maptech: 482 ft
TopoUSA: 2000 ft
Custom Algorithm: 1830 ft

Ironman wisconsin in 2004 I have 6335 feet in climbing
using different algoritms in topofusion I get these.

TopoFusion: 6335 ft
Straight GPS: 9036 ft
Straight DEM: 7056 ft
Maptech: 4897 ft

In these too case you have numbers that are correct. However the close number that is correct for Florida is somewhere between the “Maptech” and “Straight DEM”. If memory server the total for IMFL is around 1200-1500. In the case of IMWI the closest match is TopoFusion number of 6335. Again I wouldn’t go so far as to say “Worthless” in the sense that it gives you an idea. I’d say it’s pretty much worthless for comparing various courses as the variability is greater than the differences between many course.

~Matt

30 posts later and I still can’t figure out if IMLOU course is hilly? I love Slowtwitch! (see you in KY!)

I posted a elevation with picture and states.

~Matt

As a fact point: there is little to no problem is establishing the algorithm for giving “good” altitude gained, it is simply
that Avocet has the patent on the algorithm when they developed the Avocet 50, a number of years ago. With Avocet out
of business, nobody has really had a good altimeter with cum altitude gained.

Theo
Thanks Theo well put much better than I tried to set it. Heres the patent. http://www.patentstorm.us/patents/5058427-fulltext.html

bit tough to follow and I’m and engineer how beit a very hap hazard non displined free spirit engineer.

Says was 1991… I still have problem with the fact(some dispute it I haven’t checked) you can patent a mathmatical algorithm
Which is the big argument over software saying it’s just another mathmatical algorithm but heck I’m no lawyer and even if you can’t you’d have to spend lot’s of money defending youself if you used someones so called patent.

anyway… It does seem to be pretty simple and probably something number crunchers and statisticians dealing with
different types of telemetry data probably do all the time anyway…
But your explanation was right on…
Thanks

I went and pulled up 2005 ironman canada polar and gps(forerunner 301)data if you put the elevation charts in line with each other from topofusion profile and polar they line up pretty well for the most part… Polar gets 5560 using barometric climbing
and garmin letting topo calculate it gets 6500 feet. no scientific test here just eyeballing makeing charts same size.

When I figure out how to post pics I’ll see if I can post them both…
Was going to do this for usa but 2004 my polar got in a wierd state where it was in some countdown timer alarm mode every 30 seconds so after about 5 miles on the bike I turned it off until I figure out what I’d done wrong about half way through the bike so I don’t have a good polar altitude for usa. I have one for wisconsin,florida canada and hawaii though.
Want to take the gps for two loop courses see how the second loop matches up to first loop as well when I got some time.

So my 5 hour bike which was turned into 2 hour ended up hour 15 oh well such as life. Well go run and hour instead.

gholmes

I just pulled up my old IMWI and IMLP maps. Here’s my take, map data only never have ridden IMLP or IMKY.

IMKY seems to lie somewhere between IMWI and IMLP. Overall I’d guess it’s a tad bit easier on the legs as by map appearances it doesn’t seem to have the amount of intense, short, climbs that IMWI has but also doesn’t have the longer grinds that IMLP has. IMKY appears to have 3 or 4, 7-10 grade hills although they all appear smaller than the similar hills on the IMWI course. The killer on the IMKy course will be that after you finish most of these steeper climbs you are faced with more, although significantly shallower, climbing rather than a downhill in IMWI.

Should be a very interesting course. I’d be interested in hearing from people that have ridden the IMKY course as to where the “Killer” hills are at. From the map it looks like there is a steep, although short, climb on river road right after it turns SE. The next climb comes a few miles after you get on 42. That seems to have a nice steep start, but short and tapers off to a semi steady climb. The real nasty climb appears to be on the out and back area although this may be incorrect as you are crossing over a creek and sometimes TOPO hoses bridges. Only other steep climb appears to be on 393 everything else looks like longer shallow grinds or steadily rising rollers.

~Matt