[OSM-dev] SoC Project: Preprocessor to add altitude information to OSM data

Hermann Kraus herm at scribus.info
Thu Apr 9 15:00:29 BST 2009


On Thu, 09 Apr 2009 08:43:40 +0200, <marcus.wolschon at googlemail.com> wrote:
> With the relations it would be possible to publish only these new  
> relations
> (so anybody can merge them with the corresponding planet-extract
>  while also merging other data.) as no existing entities are modified.

This sounds like a brilliant idea. The files with relations only should be  
small compared to the full planet dumps, so it would be easier to  
distribute them.


> I suggest changing the name to make it clear
> what point alt_total refers to and if distance is the 2d or 3d distance
> in meters.

Names are open to discussion. I'm no native english speaker, so I'd like  
to hear some suggestions.


> Also note that not all nodes with a degree >2 are intersections of 2  
> roads.
> It may be something else intersecting (e.g. a node with the sign of a  
> city
> may be shared by a road and a polygon surrounding the city or it may be
> 2 power-lines intersecting).

These could easily be skipped. I'm planning to add a filter function that  
only selects ways that have certain tags. For example it would only  
process ways which have some kind of "highway" tag set.


> I would be willing to add support for your 3d-distance to the metrics
> for ShortestWay and StaticFastestWay and for the 3d-distance and and
> alt_difference
> to my incomplete metric that involves real physics (car accelerating,
> maximum speeds in sharp curves, ...). However for applications like this
> later metric alt_difference may be misleading as there may be 2 inclines
> and 1 decline between the 2 intersections.

As I've written in my other mail I would compute two values:

The difference in altitude between just then endpoints and a second and  
more important value which counts all the contour lines crossed.

The second value will be the thing you use for routing. If required I  
could also add two more values one counting all inclines and one counting  
all declines.


Regards,

Hermann




More information about the dev mailing list