[OSM-talk] re contours

[elvin ibbotson]

On 20 Mar 2008, at 10:05, Steve Hill wrote:

> On Thu, 20 Mar 2008, elvin ibbotson wrote:
>
>> Treating contours as shape files seems to me to be heavy on  
>> storage, downloads and processing. I have made a proposal in the  
>> wiki at http://wiki.openstreetmap.org/index.php/ 
>> Relief_maps#a_proposal to use relief shading as a background to  
>> mapnik tiles. I'm sure there must be good reasons not to do this  
>> and look forward to hearing them.
>
> I hadn't come across that proposal before, but my initial thoughts  
> are:
>
> Coloured relief as described is good for an at-a-glance idea of the  
> terrain, but (IMHO) are less useful when you want to look at the  
> map in more detail.  It could be sensible to use this system on the  
> low-zoom tiles and the switch to contour lines on the more detailed  
> high-zoom ones.
>
> The proposed doubling of the intervals leaves them far too widely  
> spaced at high altitudes which would render it more or less useless  
> in mountainous terrain.  For example, a ski resort may have the  
> town centre at 1100m and the top of the mountain at 3300m - on that  
> map the only colours you will see are the 1024-2048m and 2048-4096m  
> bands - 2 bands to cover up to 3000m of altitude difference is  
> nowhere near enough to be useful.  On the whole I'm not convinced  
> about reducing the band frequency with altitude anyway - if you're  
> cycling (for example) at an altitude of 600m, a 100m high hill is  
> just as significant to you as it would be if you were cycling at  
> sea level, but in the former case it wouldn't show up on the map at  
> all whilst in the latter it would be very obvious.

I think, on balance, band width proportional to altitude makes a lot  
of sense. A rise of a few metres makes a lot of difference  if you  
live in a flood plain but is less significant when you're halfway up  
a mountain. But, the proposal was just kite flying and if people  
think the colour band approach is worth pursuing the banding and  
colours wold need more thought. 1m band width near sea level is  
perhaps too small while doubling each time is perhaps too exponential  
(though very easy to code). More bands would help but would be  
counter-effective if the colours became difficult to distinguish.

elvin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20080320/d320d648/attachment.html>