GPS speeds (SkiTracks)

A snowHead

This came up elsewhere here recently (around here https://snowheads.com/ski-forum/viewtopic.php?t=151157&start=40 ) and the data I reported (a single point of 93km/h) was universally rubbished, as vastly excessive speeds.

I have no way to independently check it so I am no wiser, but in tests in a car I was not able to get the app to show incorrect speeds. Still, hard to believe one could do 58mph on skis, even momentarily.

So I contacted the app developer. I sent him the GPS track log and this is what he reported:

I do not think the phone has any issues - not from what I have seen from the analysis.
From my experience these are kinds of top speeds we see.
0kmh-50kmh - 40% of our users.
50kmh-70kmh is still quick but most intermediate skiers and snowboarders could reach this with the right conditions. 45%
80kmh-100kmh is pretty fast but can be obtained by good intermediate/advanced skiers and snowboarders - particularly on empty, smooth, 25-35deg long blues/reds 12%
However the problem is that the air resistance is the square of the speed so it gets increasingly difficult to go faster - as any road cyclist will know.
From 100kmh - 120kmh things start to get more difficult. You need steeper slope angles and more tucked position. <2.9%
From 120kmh - 140kmh you need speed suits, steeper slopes and pretty hard setup. <0.1%
Anything more than 140kmh is really beyond normal downhill skiing and requires a special course with downhill equipment.
Obviously these kinds of speeds should not be attempted on normal recreational slopes.
We probably see just a handful of people doing actually reaching 120kmh.
We have never seen an official 140+kmh speed ever recorded on Ski Tracks - mainly because we do not promote speed and hence we do not have speed leaderboard.

A snowHead

I (sadly) went back and re-read that thread. I didn't see you being "universally rubbished", I did see one poster DoM questioning whether a single point was accurate. You would have to plot the speeds on a graph and see if that single point was plausible or not.

To give you an example I recorded 110kph on Strava last week. Now at the time I was on Cross country skis and standing around watching a load of kids traverse a slope, my only movement was wagging a finger to indicate that the skier could start. If I plotted my speeds they would be 0,0,0....., 110kph, 0, 0, 0
It was probably a tree branch waggling that partially blocked the satellites, or a cloud, or something.

Now if you plot your speeds as 88,88,89,90,91,91,93,92,90... then that would be entirely plausible.

90 isn't crazy fast, you can touch a hundred kph on the Jeux at l'alpe d'Huez (it is one of the learner runs from the DMC) because you can straightline it from top to bottom and get into a mega tuck.

Feel free to send my your gpx and I will tell you if the 93kph is plausible but I've no reason to disbelieve it.

Well, the person's real but it's just a made up name, see?

You need to Login to know who's really who.

Usain Bolt’s 100m record is 9.58 seconds, average speed over the course of this race was 37.58 km/h or 10.43 m/s with his peak speed measured at 44.72 km/h or 12.42 m/s.

You skiing peak speed was 93 km/h which is 25.83 m/s. Your peak speed was ONLY double of someone running. Very Happy

Anyway, snowHeads is much more fun if you do.

I've no method of personal measurements except for using those speed trap setup by the ski area with display for having a go at.

As follows ;

96.88 kmh in Termingnon on Salomon 1080s hard piste, full tuck.
92.4 kmh in Les Arcs Vallandry on Kastlé XX90 fresh fallen snow, full tuck.
102.5 kmh in Verbier on Head Monster 88 hard pack, entrance barrier down so you could straight line it into speed area Very Happy

full tuck.

All in "holiday" attire. Tuck as taught to me by Italian ex racer skii instructor, chest on thighs with hands as far forward as possible, then raised to lift the air over your head like a canard wing.
Without special clothing it seems to me to top out around 100kmh ish.

You'll need to Register first of course.

Hmm. I haven’t used Ski Tracks for some years, but I did use it for a while so have some data on there still. I was more interested in distance than speed, but that is massively reduced when skiing off-piste so there was little point (for me anyway, I’m way more cautious) and I have a good feel for how far I can go in a day now.

You can see all the daily top speeds on the home screen, so it’s easy to get a feel for them. I haven’t done actual maths, but it looks like the mean top speed is about 75kph, though it regularly hit the high 80s. There are a couple in the 90s and then a couple over 100kph with the top speed of 115kph. I know that on at least one of those days I left it on in the car rolling eyes

and any over 90kph are probably anomalies too. Empty, smooth pistes would explain the rest though, as most of those were in the Dolomites.

So yeah, you maybe could get it to read 93kph, but I’d question the accuracy especially if you’re not regularly hitting similar speeds (that is NOT a challenge).

I’m more impressed with the day I skied 75km! I’m pretty sure that distance is accurate.

Then you can post your own questions or snow reports...

93 km/h seems quite plausible to me if it is transitory. Some years ago an instructor got me to ski tucked through a speed trap which recorded over 80 km/h and I am an average intermediate. (The speed trap was a sop to our daughter, for whom the adrenaline rush of speed is a big part of skiing).

But sustaining 93 km/h for any distance would be far harder except on a purpose prepared slope. Or maybe by a professional downhill racer with all the gear.

After all it is free

Most days if I can get first tracks I can get over 90 consistently on a snowboard - Rossi Xv - it’s quick

That’s peak speed obviously. Same slope each time.

Very plausible to us too. We average between 40-55mph as top speed each day we ski in good piste conditions, though obviously when we step off the sides speeds drop quite a bit. Well serviced longer skis help if your looking for that kind of thing, as does a good tuck (better without the tv ariel)

*Ski* the Net with snowHeads

snowHeads are a friendly bunch.

@rob@rar, did you leave it on in the train on the way back from Bourg?

snow reports

@Scarlet,

It's not mine, haven't used SkiTracks in a very long time. It's from a fellow snowHead, recorded while skiing on the Pre-Season Bash. I've seen more than a few similarly ludicrous speeds recorded by SkiTracks.

I know that the Garmin watch I had was seriously lacking in accuracy at speed because I tested it at a track day. Speeds would've been between about 20 and 80mph I'd guess and I obviously followed the same route each lap but got this.

You know it makes sense.

GPS altitude calculation is highly problematic: you need to have a signal from a minimum of 4 satellites (which is difficult in mountains) and even then it can be off by as much as 100m or even more.

This results in drastic variations and nonsense results like posted above.

Otherwise you'll just go on seeing the one name:

I managed to break my all-time record in Fieberbrun on piste109 week before last...
It was first thing on perfect piste, we put spotters at a few points and I did the whole thing in a full tuck.
Garmin told me my 3 runs were max speed 88kmh, 99kmh and the best 116.8kmh. On that run i think i left a long brown streak down most of the run - I'd definitely never been anywhere near that fast on skis before!
It may not have been absolutely accurate, but I left it on on the bus which didn't exceed 71kmh!
I did compare the gps track and it corresponded very nicely with the piste map and road...

A snowHead

@rob@rar, I'm not sure i could manage that even in freefall. And no, i won't give it a go. Laughing

A snowHead

Well, the person's real but it's just a made up name, see?

Very skeptical of the accuracy of these things. I recorded 61.4mph twice in Zermat at new year. The first one I can believe as it was a smooth, freshly posted steep section from Furgsattel to Trockener Steg, just off to the right of the main runs that had seemingly been designed for speed and I was kacking it. The second time was very closely geo located on the run from Theodulpass down to Trockener Steg and I’m not having for a second I reached the same speed and that it happens to be exactly the same recorded speed as the prior day.

You need to Login to know who's really who.

I’m sure it’s not that accurate but it’s a lot of fun trying to crack a hundy.

Generally I measure it on the same slope at the same time of day, and I usually get relative speeds that are consistent - ie no outliers and if it feels quick it usually is a bit quicker. The gps tracks are accurate as well. I normally ride with my daughter and she will be quicker than me and again the measurements reflect that.

Anyway, snowHeads is much more fun if you do.

Sweamrs has hit over 100kph that was verified by speed trap. Before the finger wagging starts that was on a fully closed course with full netting on either side and she was actually racing (i.e as safe as doing motorway speeds will ever be).

When she's "just cruising" with me the Garmin Fenix 5 records speeds in the 60 to 70 kph range which doesn't seem unreasonable.

You'll need to Register first of course.

@Oleski, @philwig, +1

And @sweaman22, a Garmin Fenix 5 is going to be far far more accurate than any SmartPhone so that makes sense!

For road cycling, good GPS units tend to cost circa €400 and there's a reason for that, even then you can get the occasional blip.

On Strava anyone that records their data with a Smartphone can expect some major inaccuracies on a ride.

On the mountain, I use a Suunto Peak watch and again whilst way more accurate than a smartphone, which is invariably stuffed in a pocket, that still is not as good as a dedicated GPS unit that you use on a bike.

Anyway I'm not interested in speed achieved but sadly height actually climbed physically rolling eyes

Then you can post your own questions or snow reports...

There can be absolutely no doubting the accuracy of Ski Tracks data.

I don’t know how to post the screenshot but last year I recorded

2695.9 kph

It’s on the screen so it must be true

After all it is free

"There's a huge technical difference between estimating the speed of a vehicle on a road for which you have geographical information and a skier on a slope."

Could you explain please?

"but with altitude which the GPS device needs to calculate the vertical component of your speed"

I saw consistent numbers for more leisurely speeds through the day, during which time the constellation would have changed totally.

The geoid error is a big problem but is consistent for a given location.

*Ski* the Net with snowHeads

The issue with using this data is when you use it as a group and it’s generates dangerous competition Confused

A number of years ago a group of us were skiing at Stubai. We all had ski tracks and it degenerated into who could ski the fastest. Between the 4 of us on the same run (hard packed but smooth starting at 3225m) we all achieved over 90kph with my peak of 105.6kph Shocked

Ended up with the nickname “Straight Line Steve” spoo

snowHeads are a friendly bunch.

Anyone done a measured speed run with their phone/GPS watch for comparison? Did you get the same readings?

snow reports

@Scarlet, haven’t but can. Will repot back next week.

My argument that 93kph is really based on ... I know when I have gone faster than that on a short straight race course on SG skis and in a catsuit and it was both hard to get to that speed in the available distance and quite fast enough thank you to understand that getting to that speed in normal ski clothes on a normal piste on punter skis would be difficult and unpleasant.

And truly irresponsible on a public slope. Even if likely. Which it isn’t.

My Polar typically peaks around 70kph. In Monterosa last year, on empty pistes, my Bro’s Garmin and my Polar were in general agreement in the high 70s.

Back home on Monday, poss not skiing as have guests for the week. Will test once back up hill.

@under a new name, You should make it a proper experiment, as far as you can. Do several of the same run in a catsuit and in normal ski gear. We need pics (of the gear) for proper evaluation, obvs Toofy Grin

You know it makes sense.

My wife and I both ran skitracks app on our phones for a recent trip with mixed results. At the start of the week both seemed very varied, diffent lift counts, piste distance and the speeds were a joke, 145kpm for me which was wildly inaccurate. Towards the end of the week we more or less had the same readings and our top speeds(not chased for just enjoyed) were more or less the same (68/71 ave). Seems the app settles down with repetition. Cross ref with skilines the distances and vertical feet skied were much of a muchness.
I must stress no scientists were hurt during these wildly vague none accurate and cheap app experiments of vanity based around a winter sport!

Otherwise you'll just go on seeing the one name:

Speed recorded seems to depend on both phone and app being used, as well as the odd anomaly popping up from time to time.

I use Maprika generally (not so much for speed, but more because I like that it shows your day's track mapped onto the piste map). When run side by side with SkiTracks, it shows consistently lower speeds than SkiTracks - I assume the sampling is over slightly longer periods, so more averaged.

Apple phones also seem to consistently show faster speeds than Samsung. We noticed this when people who were skiing slower seemed to show faster speeds. Out of curiosity we tried with one person with both phones - up to a 10kph difference was recorded using the same software. No idea why.

Individual peaks should be treated with caution - this seems to be where the anomalies creep in. You can also see surprising variation in speed on the Maprika trace when sitting on a lift, which are usually pretty constant speed.

I am doubtful about some of the really high speeds recorded in normal recreational skiing, and the "I went fastest" element is clearly dangerous - but inevitable....

A snowHead

I compared Skitracks with some satnav apps in a car, driven carefully to achieve a specific peak speed and not go past it again. I could not get any significant errors.

It is a good argument about the slope of the slope, so to speak. GPS uses doppler shift to get a "true 3D" velocity, but the number presented in the API is likely to be 2D. However, this argument works the opposite way: the real velocity is even higher than the displayed value.

A snowHead

GPS make spurious mistakes, apps interpret the data wrong. Among other things in the last year my Garmin has variously recorded me doing 58mph cycling UP one of the hills the Tour of Britain regularly uses as a king of the mountains climb, 40mph while out hiking and 30mph canoing on a canal! All single data points and all obviously wrong. Also between them and using the same data from my Quatix 5 watch Garmin Connect, Strava and Mapmyrun agreed on the location of my max speed on the Ride London last year but variously reported it as 49mph (which feels about right and looks consistent with the speeds either side of it), 62mph and 106mph (which are both single data points with speeds in the mid 40s either side of them). Don't rely on a single peak value, look at the profile either side of it to see if it seems reasonable.

Well, the person's real but it's just a made up name, see?

Warning: unless you are experiencing problems sleeping, then I don't advise reading any further ...
There is an outline explanation of GPS derived altitude here https://eos-gnss.com/elevation-for-beginners, but even with WAAS (higher accuracy receivers from additional ground stations widely used in aircraft), aircraft are generally not allowed to use the altitude calculation below 300ft (they switch to local radio beam based landing assist systems for approaches in poor visibility) - because there are potentially hundreds of feet difference between the actual surface of the earth and any (necessarily approximate) mathematical model of that surface. I don't know of any phones using WAAS GPS - so the vertical accuracy is far, far less. The horizontal x, y accuracy however is fairly good using one of the mathematical models for the approximate surface for even non WAAS GPS - which is why you can overlay it on a map and use it for lat/long navigation.

To give an idea why we use various mathematical models rather than a stored map of the surface height - the earth has a surface area of about 510 million square kilometres, with about 150 million square kilometres above the sea. If for example you wanted to store the "actual" (above mean sea level) height as a 1m grid then you would need 150million x 1 million = 150 x 10 (power of 14) which would take about 450,000 Gigabytes of storage (with the largest iPhone offering only 512 Gigabytes for example). Fundamentally, this is difficult to store on portable devices, so an approximate model is used!

You need to Login to know who's really who.

That's all true (except the bit about 300ft which is nonsense; it may be a reference to cat3 autoland which uses a radalt for the last 150ft or so; anyway aviation uses barometric altitude, not GPS altitude, in all phases of flight) for positioning, but GPS doesn't get the velocity that way.

It doesn't take consecutive position points and divide the distance moved by the time. That would produce huge jitter in the velocity.

Velocity comes out separately, from doppler shift.

WAAS is something else, but it doesn't affect the velocity calculation. The altitude calculation is only slightly more accurate.

Anyway, snowHeads is much more fun if you do.

You'll need to Register first of course.

The app should use the doppler shift derived number for velocity.

The Q here is whether it uses the 2D or the 3D velocity. I will do some digging...

Then you can post your own questions or snow reports...

@Peter Stevens, Doppler shift calculations are not always available end even when they are, they are not that accurate either when it comes to mountain profiles, as you need the visible satellites to be sufficiently apart to get accurate speed components .

from the same APi documentation https://developer.android.com/reference/android/location/Location.html#getSpeedAccuracyMetersPerSecond()

they give 68% probability to have the correct speed within +-accuracy interval with Doppler factored in. In other words, little better than hit or miss. What that accuracy will be in in mountain conditions - hard to say. Does SkiTracks display it anywhere?

After all it is free

@oui4ski, ...or download the answer from a public database:

https://snowheads.com/ski-forum/viewtopic.php?p=3284641

*Ski* the Net with snowHeads

I am sure the terrain mapping factor is totally irrelevant to getting velocity out of GPS.

IOW, the speed accuracy of a car satnav app does not depend on it knowing that the road is a road, etc. You can test this by taking it up in a plane.

Here is some reading on the various methods
https://insidegnss.com/how-does-a-gnss-receiver-estimate-velocity/
but one is none the wiser on (a) what smartphones actually do and (b) whether the velocity is the 2D or the 3D value.

The above 68% comment suggests that doppler velocity is NOT being used, which would explain why phones might be poor at calculating a maximum velocity if the velocity is widely varying and/or the heading is varying rapidly also. In the satnav context none of these are happening.

Help!!

GPS speeds (SkiTracks)

snowHeads Forum Index >>> The Piste