Altitude 'noise'

I was interested in Steve Rock’s use of a Naismith formula to predict his riding time when ‘Touring with Technology’ in the April-May issue. He’d allowed only half as much time for climbing as I would, so I wondered if that explained his late arrival some days. It turned out to be more complicated.

Steve planned his routes in Memory-Map, which generates climb and descent totals from contour maps. But roadbuilders use bridges and cuttings to ease the gradient. And except on very large-scale maps, road widths, bends and junctions are exaggerated for clarity. So trackpoints placed onto the depiction of a road may be further up or down a slope than the road really is. All these discrepancies superimpose a random pattern of ‘altitude noise’ upon the route profile, which can greatly exaggerate climb and descent totals – sometimes overestimating the climb by a factor of two!

This explains why Steve allowed just half a minute per 10m rise, according to which a 1200m ascent would take one hour longer than a flat road, rather than the two extra hours most cycle-tourists should allow for such a major climb. Memory-Map’s 1200m was in fact only 600m.

Memory-Map isn’t alone: it seems that all modern methods for computing route profiles are prone to altitude noise. A popular method is to plot your route on a website such as Google Maps or Bikehike etc. What these sites gain by more accurately following roads, they lose in sketchy contours. These usually come from a survey of the whole planet in 90m squares. Narrow features are missed by such a coarse grid, so that rivers in gorges seem to flow uphill as well as down and likewise the roads following them!

You might think that a recorded GPS track should be more accurate than anything taken off a map. Unfortunately GPS does not record altitude with the same precision as location. So GPS also produces altitude noise, as its height reading fidgets up and down by several metres. A GPS which uses a barometric altimeter will keep a better record, but will still add some altitude noise.

Some route-planning websites do attempt to filter out the altitude noise, and I find that yields fairly realistic data, good enough to use with Naismith’s rule. The calculation of riding time is never an exact science, but as one who typically tours at 20kmph on the flat, I add another hour per 600m up and subtract an hour per 2000m down. The formula is simple: time in hours = km/20 + UPm/600 – DOWNm/2000.

After loading Steve’s tracks into Bikeroutetoaster to get more realistic totals, I found that 19km and 500m per hour yielded times which corresponded most closely to those he recorded. This lower flat-speed and climb rate are reasonable given his camping load. For an unladen day ride, I might suggest 21km and 700m. Your mileage may vary, as they say.

Chris Juden


This was first published in the August / September 2014 edition of Cycle magazine.