I'm not sure if this is the right community to ask this, so let me know if it's not.

I'm wondering if there's been some empirical research as to how villages and cities are distributed around other cities.

If one looks at a map, usually there's some clustering around big cities visible. I'm wondering if the distance from that central, big city to villages or cities around is follows some kind of probability distribution.

Indeed interesting, and probably one cannot find a distribution that fits cities all around the world, but that doesn't matter so much.

I'm looking to find some kind of pattern in city distribution, here's why:

I want to create realistic data set of point in the plane that represent places (e.g.: cities, villages, ...). Uniformly generating these won't be a good idea, since this is not realistic in my opinion.

  • It's an interesting question. Also interesting would be looking at that spread and distribution in "older" areas compared to more newly settled ones -- while that's probably less apparent in Europe, for example, in North America there are distinctly different city layouts in East Coast (old) cities compared to midwestern or western ones (newer), and the spread of suburbs may be heavily affected by when cars were introduced (early or late in the growth of the city and surrounding areas)
    – Erica
    Commented Apr 24, 2015 at 11:12
  • I suggest you to post it to stats.stackexchange.com too. Commented Apr 24, 2015 at 12:19
  • I'm voting to close this question as off-topic because it is geographical analysis, not geo data processing.
    – Martin F
    Commented Apr 24, 2015 at 15:07
  • You mean as in Central place theory?
    – whuber
    Commented Apr 24, 2015 at 15:29
  • 1
    @MartinF Our GIS site is about more than just geographic data processing! On our help center pages, at the very outset, we state that we welcome "questions concerning geographic information systems and science." Please do not forget that last word.
    – whuber
    Commented Apr 24, 2015 at 18:31

2 Answers 2


Normally they follow a parabolic fractal distribution (like towns, oil fields, coasts etc...) cf. Benoit Mandelbrot's original work

That should give you a starting point:


Now in terms of implementation could you tell more about your fluency in Postgis for PostgreSQL which seems to me the way to code the problem.

  • I see some aspects about size or frequency of cities, but nothing really about their locations? Commented Apr 28, 2015 at 7:44

When I were a lad it was all about Christaller's hierarchy. Not sure if that is still a valid theory, but worth looking into I would think. See: http://en.wikipedia.org/wiki/Central_place_theory

  • That's probably easy to implement and accurate enough. I'll let you know when I succeeded. Commented Apr 28, 2015 at 7:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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