An interesting discussion - thank you all. Since I originally posted, I did a little investigating of my own. Without getting into detail, I've concluded that (1) the discrepancy between CalTopo and OS Maps is likely due to different sampling inervals. CalTopo seems to adjust the sampling interval to a point where the shorter the track, the more accurate the elevation data; (2) OS Maps does not seem to suffer nearly as much from the same sampling interval problem, for reasons not entirely clear (different algorithm), and (3), in conclusion I think the OS Maps numbers are closer to ground truth. As they say, all models lie...but some can be useful anyway.

Agree that track accuracy is important. And, yes, a single total ascent number hardly tells the story. In fact, I've never paid attention to a "total ascent" figure before now. I generally pay attention to the bigger picture of when and how much the ascents and descents occur and how steep they are, over what terrain etc.

Experience is the key. We know what a 230-mile +/- 50,000 ft trek and 26-mile +/- 5,000 ft day feels like, so as long as we can keep our mental metric conversion algorithms online, I suspect we'll be be ok. :wink:

I am getting a slight feeling of deja vu.

Indeed Emma - it's déja vu all over again! :upside_down_face:

But this is a favourite old chestnut, so let's not spoil the fun. For completeness I'll add to the posts above by repeating my own observation regarding graphic representations on maps with regard to ascent figures - from my blog way back in 2008!

The big problem seems to be with roads, tracks and paths. Any road on the map is merely a graphic, scaling at maybe 30 m wide on a 1:50,000, and, crucially, unconnected to the height data. This is why, on Memory Map, the occupants of the cottage near Braemar at NO127909 are apparently faced with the problem of the opposite side of the road being 16 metres (52 ft) higher than it is outside the house (must be a problem backing the car out of the garage!).

Thus any route plotted along this road, if it deviates slightly from one side to the other, will record 16 metres of additional ascent at this point alone - and along its length these errors will add up to a great deal.

Zoom in on Memory Map and pick any land rover track contouring around a hill, between contours. Using a paper map, we will say that the ascent here is zero, and so it is. But check the height on the left and right hand sides of the track, and you will find that it often apparently slopes by as much as 3m or 4m. The same applies to canals - sloping water!

When we plot a route we might stay within the confines of a road or track, but it is simply not possible to get an accurate ascent figure whilst the software is incapable of recognising that a track or road is (mostly) flat from side to side.

That said, our 'paper' method does not take account of undulations along the track that do not break the contour line. However, I reckon that any underestimation that this might produce is relatively insignificant.

One for the boffins to work on, but for now ..... count those contours!!!

Plus ça change and all that.

Happy New Year - although it's probably just a repeat of the old one. :wink:

Hi Phil,

I certainly have no wish to spoil such innocent fun – and with all the boys and girls playing together so nicely. I think I have seen that house outside Braemar. I wondered why they were parking on the road.

2018 does seem to have a lot in common with 2017 (it is raining in a very similar way), but I feel it is going to be much better. Happy New Year to all! :champagne: