HELP! - Height difference vs height climbed & different sources!

[Grumpyoldbiker]

I'm doing some route planning for a holiday in the Picos and I have been using a variety of aids, Google maps, Cicerone guide, Strava route planner, and MyGPSFiles.

I am getting thoroughly confused by the differences between the different sources. There are huge differences between them for the height differences between locations. I realise that the height difference between two locations is different from the height you will have to climb, but even so, the differences I am getting seem huge. I am now so confused that I am not even sure what each different source is showing me, as some of them don't explain what the figures they give actually mean. It seems to me that the height climbed is the most important figure, along with gradient data.

Can anyone shed a bit of light on this and suggest which source of data I should use as best for height climbed, and what it means exactly?

Sorry if I have missed a relevant thread, I did search, but couldn't see anything that answered my query.

Thanks

Peter

[Cyril Haearn]

There was a thread about exactly this recently, someone will find it for you and post a link

[Audax67]

They're all wrong. Most sites calculate metres climbed using the SRTM tables released by NASA around Y2K. These divide the world up into 90-metre squares and give the average height above sea level for each. This gives an extremely jagged profile for a ride, that the sites then apply smoothing algorithms to. Depending on which algorithm and the choice of parameters, the results can vary enormously.

More recently, NASA re-released the tables* with an accuracy of 30 metres instead of 90. Some sites have updated, others haven't (it's quite a bore to do). This causes even more disparity.

Take an average of the lot is my advice. Or just believe the worst and hope for the best.

*well, they always existed, but 30 metres was reserved for the US and the military.

[simonhill]

You could do it the old fashioned way and count contours.

I don't know if they do for that area, but google maps usually have contours in 'terrain' mode of viewing. Work out which contours are up and count them then multiply by contour interval. You may need to interpolate a bit, but this should give a fairly accurate method of height gain measurement.

[HarryD]

Even different GPS devices from the same manufacturer will give widely differing numbers. Probably the best you can hope for is consistency from a given source rather than between them.

[andrew_s]

There are two questions:
a) What constitutes a climb?
b) How do you measure it?

For (a), it's really about the smallest change in elevation that you want to count as a climb. You might decide to only add up "proper" climbs, and ignore everything less than 100 feet, to ignore everything that you can freewheel over (5 m or so), or to count everything. In the latter case, do you include 10 cm for a sleeping policeman, 2 cm for a pothole or sunken roadworks trench, or 0.2 mm for every piece of chipping on a surface dressed road?
The smaller the "climbs" you include, the larger your claimed climbing will be. In the case of chippings, that could be 100 m per km, if only you could measure it.

For (b), the realistic options are contour counting, DTM grids (eg the SRTM data that @Audax67 mentioned), GPS height data, or data from a barometric altimeter. All have their problems.
Contour counting is tedious, and requires access to decent maps (OS 1:50000 and OS 1:25000 are available on Bing maps, but foreign can be difficult).
DTM is limited by the cell size, and nearby higher ground affecting your track height (@CJ found somewhere in Spain where you can apparently go over a 100 m col whilst following a riverside road).
Raw GPS heights are subject to random errors - typically about 90% of the heights ( taken every second) will be within 10 m of the true height. Your GPS will smooth the errors, but how it does so is generally not known. I've sat at a cafe table trying to calibrate my barometric altimeter, and watched the GPS height drift up and down repeatedly over a 40 m range, giving about 200 m of climbing at the table.
Barometric altimeters are quite stable over moderate periods, and will correctly report height differences of 1 or 2 m, but the cailbration changes with the air pressure, so they need calibrating on a daily basis, and will quite commonly report the start and end of a circular ride as having elevations 50 m or so different.

FWIW, I regard contour counting as giving the best "true climb" figure. This ignores climbs smaller than the contour interval (5 or 10 m, depending on map base), and has the advantage that if I count contours, and someone else counts them for the same route, we get the same answer.
If I can't be bothered, I use barometric values from my Etrex 30. The error due to air pressure variation is generally limited to the 20 to 50 m drift during the ride.

I believe that Strava uses heights from your GPS if it's a model that uses a barometric altimeter, but DTM/map heights if it's not got a barometer, even if heights are recorded in the track data.

P.S. I did notice a month or so back that HM Gov was proposing to release a 5 m National Lidar dataset under an open data licence. Developers of UK mapping sites may be interested.

[Warin61]

Raw GPS heights are subject to random errors - typically about 90% of the heights ( taken every second) will be within 10 m of the true height.

10m ... that would be for horizontal location. And probably at a 1 sigma level.

For height I would think 3 times the horizontal accuracy would be a fair estimation, so 30 meters for height (again 1 sigma).