Calibration Consternation

I’ve been obsessed with distance measurements recently.

This paranoia started when I did the first recce of the regularities for the Cloverleaf Historic Rally. I’d already measured the distance digitally using the TrackLogs software I mentioned in the last issue, so I knew the public route totalled around 140 miles. During the recce I compared three distance measurements: 1) the TrackLogs reported distance, 2) the distance displayed on my Garmin GPS, and 3) my car’s odometer reading. Consistently 1) was less than 2) which was less than 3). The difference in results set me on an interesting trail to discover which was the correct distance … and the answer – well, none of them actually.

Odometer measurements can be dismissed immediately. They are usually to one decimal place so that doesn’t help in terms of precision, and accuracy is a major issue too. Odometers only have to be accurate within plus or minus 10 percent, although recent statistics from the RAC said that on average they were about two percent fast.

A discussion with the TrackLogs people revealed how inaccurate OS maps can be for measuring exact distance. You can see evidence of this on a combined road and aerial view on Google Maps – the drawn roads do not always follow the line of the photographed roads. You will also see road position variations by overlaying the digital versions of 1:25000 and 1:50000 maps of an area. The smaller the scale the greater the accuracy, but still the measurement of road distance is affected by exaggerated road widths (about four times the real width on 1:50000 maps) and smoothing of road contours, which will understate the true mileage. On a local nine-mile route I found that the shortfall was about two percent. For the same reason, Internet based measurements using the Google Earth or MapMyRun websites will only give you an approximate distance.

Of course trying to measure distance on a two-dimensional map will add to the error, especially in hilly areas. Going up and down 1:7 gradients you will travel one percent further between 2-D points than on a flat road – that’s an effect of almost two seconds a mile at 20 mph. TrackLogs does however compensate for this with a measurement toggle for road distance (vehicles) and aerial distance (aircraft).

GPS has its problems too. A clear path to the sky is required, since interruption to signals can occur around tall buildings, tree covered routes and in bad weather. More importantly, a typical GPS device is accurate to about 10 metres – for a single point and this can lead to two types of error. First, imagine travelling down the centre of a dead straight road 20 metres wide. At the extreme, successive GPS plots could alternate between the right and left hand edges of the road. Join these plots together and measure the distance along your zig-zagging route and it will clearly be longer than along the centre of the road. Second, if you go round, say, a 90 left and the GPS plots are taken before and after the bend then the distance between the points will be shorter than you have driven. Over a long distance these pluses and minuses should balance out, but on my standard nine-mile route I have seen intermediate differences which equate in regularity terms to one or two seconds. Organisers that use GPS distance as a basis for setting their regularity times be warned.

Until the US gives public access to their military version of GPS – which is supposedly accurate to one metre – we will have to make do with our regular analogue-driven trip meters for measuring exact distances. But where do you find an exact distance for initial calibration? As a competitor you’re probably not worried; crews set their trips according to a rally’s calibration route, so everyone uses the same base; but an organiser needs a standard to set their master trip. Some use police or highway authority measured miles, others rely on the (not consistent) 100 metre motorway markers. Given that I have to change the Porsche’s trip setting on every event, it seems a standard mile varies depending upon where you go rallying. As the designer of the Cloverleaf regularities I needed a distance I could trust, so here’s what I did to get my calibrated route.

First I used a trundle wheel to measure an exact mile (walking such a device for longer than that is a pain). I then configured my bicycle computer to mirror the trundle’s readings (don’t rely on the quoted circumference of your cycle wheel, measure it instead) and cycled my standard route several times to get consistent results. Finally, I configured the Brantz tripmeter in my car to mirror the bicycle readings. Fiddly, but other than dragging a very long tape measure around the Cloverleaf I’d never satisfy my distance pedantry.