Smoothing algorithm for calculating ascent, comparison between sites.

[Somnolent]

At various times I've used different sites to give me mtrs of ascent: bikeroutetoaster, ridewithgps, bikehike.
All gave different results, presumably because they used different "smoothing" algorithms to ignore or include individual small lumps & bumps.
The differences were fairly consistent though, bikeroutetoaster always the low number, bikehike always the highest (although often fairly consistent with those given by the AAA man) .

But now bikehike is giving lower numbers than rwgps.  I'm wondering if they changed their algorithm.

Anyone else noticed the change?

Anyone got a different mapping site they consider to give better results than the others?

[Ningishzidda]

What IS the recognised and accepted standard AAA method of determining climbing?

Contour count on the Landranger?

[Somnolent]

What IS the recognised and accepted standard AAA method of determining climbing?

Contour count on the Landranger?

http://myweb.tiscali.co.uk/audaxaltitudeaward/howtoassessclimbing.html

Accepted Methods of Working out Climbing

Contour Counting. This is the original AAA method, and still the standard by which other methods are judged. It is consistent and accessible, it does not need any computer software or electronic means of recording altitude.
Using a GPS tracklog. GPS tracklogs are now acceptable as evidence of the climbing in an AAA event. However, the tracklog has to be processed in a particular way in order to produce a result which is acceptable for AAA purposes.
Three independent altimeter readings. Some cycle computers and heart rate monitors give altimeter readings and record metres climbed. They require. calibration at the start of an event against a known height above sea level. Results can be inconsistent, which is why three independent readings are required, preferably from three different dates. This method is little used.
Mapping software or on mapping websites. These can be used to draw a route which can be exported as a tracklog and processed further to give an acceptable climbing figure. The climbing figure which the mapping software or website itself produces is usually not acceptable

[Hillbilly]

I believe both RWGPS and BH both changed in the past year.

I understand that if the AAA man is passed a gpx track that looks a bit iffy, he will run it through some correction algorithm.  I don't know which one he uses, but suspect it is something like gpsvisualizer.com which allows you to upload a track and it will replace elevations with satellite data.

If I were to be the AAA ubergruppenfuhrer, I would use something like gps visualizer for all submitted tracks, as in my personal anecdotal experience, this gives figures close to what a contour count gives (albeit I haven't done in each and every part of the country - it may be that some areas are poorly covered by the satellite data, which relies upon NASA topography data, and I can't imagine they need the same accuracy for every part of the world).

[Ningishzidda]

If I were the AAA Gaffer, I’d be asking for the OS Grid ref for the bottom and top of each Significant Climb.
Then I’d check each climb’s Category, and award 0.25 pts for a Cat 4, 0.5 pts for a Cat 3, 0.75 pts for a Cat 2, 1 pt for a Cat 1 and 2 pts for a HC.

[Somnolent]

I believe both RWGPS and BH both changed in the past year.

I understand that if the AAA man is passed a gpx track that looks a bit iffy, he will run it through some correction algorithm.  I don't know which one he uses, but suspect it is something like gpsvisualizer.com which allows you to upload a track and it will replace elevations with satellite data.

If I were to be the AAA ubergruppenfuhrer, I would use something like gps visualizer for all submitted tracks, as in my personal anecdotal experience, this gives figures close to what a contour count gives (albeit I haven't done in each and every part of the country - it may be that some areas are poorly covered by the satellite data, which relies upon NASA topography data, and I can't imagine they need the same accuracy for every part of the world).

I had a look at gpsvisualizer.com but I must be a numpty because whilst I could see how to get it to add the satellite contour data to track that lacked elevations, and draw a profile from that I couldnt get an ascent figure out of it.   Any ideas?

[Somnolent]

I believe both RWGPS and BH both changed in the past year.

Simon from Bikehike has responsed to confirm the change there was about 10 months ago.

[andrew_s]

All gave different results, presumably because they used different "smoothing" algorithms to ignore or include individual small lumps & bumps.

It's just as likely to be different sources of height data. There are at least two NASA data sets, and the OS supply one as "OpenData". As well as different data sets, there will be different methods of interpolation between the supplied heights (which will be on a regular grid spacing), and possibly also different treatments of where the track goes (as drawn, snapped to road & if so which road).
In some areas the NASA data can give decidedly dodgy data. Chris Juden posted a nice screen shot of a river flowing up and over a 180m high col.