I was reading some literature dealing with least-cost path calculation and accumulate cost-surface generation using the Tobler's hiking function. I was reading this literature since I am trying to implement in R the generation of isochrones around a starting location (Integrating use of surface distance when calculating cumulative cost-surface based on walking pace?).

I can't understand what seems to me quite odd: some sources (e.g., Wikipedia) indicate that the speed predicted by the function is in KmH, while the author of the gdistance R package (see this PDF at page 14) indicates that the off-path speed would be in meter-per-second (he actually indicates m/s).

I am having hard time in understanding which is correct. This is not a trivial issue since different results are bound to crop out using Kmh vs. m/s.


The formulae in wikipedia and the R doc are identical, lets see what they say walking speed is on flat ground:

> tobr = function(m){6 * exp(-3.5*abs(m+0.05))}
> tobr(0)
[1] 5.036742

About 5 units. Is that km/h or m/s?

If it was 5 m/s that would be 5 * 3600/1000 = 18 km/h. That's a pretty quick running pace, never mind walking.

So I think gdistance is wrong, and you should perhaps contact the maintainer. I don't think it affects the construction of the path in the R package vignette, and I don't think the travel time is quoted in that example so the units of speed don't make a difference.

  • Hello. Thanks for your reply, which clarifies things. Yes, I believe that in the 'gdistance' PDF there is a typo: m/s would indeed be kmh. I agree with you that that does not make any difference for least-cost path calculation, but I think it would indeed change things for the calculation of isochrones (walking-time areas) around a start location. Thanks again for replying and for the elucidation. – NewAtGis Jun 19 '18 at 6:51
  • 1
    I've just posted a friendly email to the maintainer, and pointed him here. – Spacedman Jun 19 '18 at 12:39

This is indeed incorrect. The original article by Tobler has km/h. Note, however, that Tobler's Hiking Function is just there for illustration and should probably not used at all in research about human mobility.

  • 1
    Being a hiker myself, I can attest to the function's overall robustness. There are variations between individuals (my cruising speed on flat open ground is about 6km/h and I can maintain 4km/h on a +15° slope if the surface is regular), but it's a very good model. I review my hiking data after each outing, which is close to 600 tracks now, and the numbers are pretty close. – Gabriel C. Jun 19 '18 at 12:54
  • @Jacob van Etten: thank you for your feedback. I did not mean to criticize, just trying to wrap my head around that discrepancy. – NewAtGis Jun 19 '18 at 14:20
  • @Gabriel C.: yes, indeed I came across some literature in which (more recent) empirical measurements seem to confirm the broad validity of the Tobler's equation. – NewAtGis Jun 19 '18 at 14:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.