[Freegis-list] time based maps

Ian Turton ian@geography.leeds.ac.uk
Wed, 25 Sep 2002 16:47:22 +0100


At 20:59 23/09/02, Christian Kaiser wrote:
>Time based maps are an old concept, that's right. Sometimes, you can find 
>them still today. But there are a couple of problems with those maps.
>Imagine two towns A and B. They are at a distance of 10 km. But there is a 
>mountain between them. So you have to drive 20 km from A to B. And you 
>will be slower because it is a pass road. So you will have 30 minutes for 
>the 10 km. When you like to make a time map, you have to "flaten" the 
>moutain. These means you have to make a sort of projection, or a 
>distorsion. This is possible in some cases, but not in other cases. 
>Normally, you will have a map which is not very exact. So far for the 
>problem of mapping time.
>There is an algorithm who creates a sort of map from a matrix of distances 
>which is called "multidimensional scaling". The algorithm is quite similar 
>to the one for a Factor Analysis. Some statistic programs (like SPSS) can 
>do the multidimensional scaling and produce simple time maps.

One of the problems with any travel time based map is that you have to 
define where the "start" of the map is. This again leads to interactive 
solutions being necessary. There are lots of interesting ideas to be 
investigated here, when I'm next in the office I'll see if I can find some 
references.

>What actually would be more interesting is to produce isolines of travel 
>time that you plot upon a normal (projected) map. Perhaps somebody knows 
>how to do it...

That's fairly commonly done - usually you define a travel time per km for 
different types of roads and then calculate the time taken for all possible 
routes between any location and the "start" point (or group of points e.g. 
hospitals) and the color a grid based on the shortest time to the "start" 
point. This gives a raster that can be displayed with different colors to 
show different times to a point or points.

>Christian Kaiser

Ian