[OSM-talk] HOT Community Webinar - Wed July 26th 3:00 PM – 4:00 PM UTC

Christoph Hormann osm at imagico.de
Wed Jul 26 11:41:31 UTC 2017

On Wednesday 26 July 2017, Frederik Ramm wrote:
> The practice of "draw some blobby residential landuse around a
> cluster of buildings" isn't totally unheard of elsewhere; I did that
> myself a lot in the early Landsat days, especially when the
> individual buildings couldn't be seen precisely.

Thanks for pointing this out since this provides a good starting point 
for me to illustrate the core of the problem.

In [1] you can find a typical German rural landscape as it shows up in 
Landsat imagery, not much different from many other parts of Europe.

What you probably remember doing is tracing the villages and hamlets 
here as landuse polygons.  This is perfectly fine because:

a) these are actually observable forms which are verifiable in the 
images and on the ground on the level of accuracy the data offers.  

b) these are indeed almost all residential areas.  In the German society 
like in all industrial/post-industrial societies the vast majority of 
people even in rural areas work in the secondary and tertiary sector, 
usually away from their homes and villages/hamlets are fairly uniform 
residential areas.  If you are familiar with the area you probably know 
that most of the smaller isolated groups of buildings are actually 
landuse=farmyard and not landuse=residential.  This can be reliably 
determined from high resolution images but not from Landsat.

In [2] you can find a typical area in South Kivu at the same scale in a 
Landsat image which likely also features a similar, maybe even higher 
population density.  The structure of the local society and of the 
settlements is totally different here.  The vast majority of people in 
rural areas work in agriculture, there is usually no separation between 
agricultural areas and residential areas.  Even if there are separate 
residential and farm utility building this separation ends at the 

There are larger villages here with distinct residential areas which can 
also be seen on the Landsat image (less easily though since the 
building styles and the unpaved streets provide much less contrast).  
But the majority of dwellings are scattered around the landscape.

What the HOT instructions do now is encouraging mappers to project their 
experiences and assumptions from their own industrial/post-industrial 
society onto parts of the world that are very different turning 
scattered huts of peasants and subsistence farmers into European 
residential quarters on the map using some formal geometry construction 
rules and at the same time devaluing any diligent local mapping of 
actual residential areas.

And as Frederik indicated this is not specifically a problem of the 
presentation linked to, this is something that can be observed in many 
HOT instructions.  It is fine to prioritize the important points in 
instructions but the more you do that the more you need to be careful 
to actually carefully select what is really important and to not 
pretend Earth is not a complex and diverse place.

For reference the areas shown in OSM can be found in [3] and [4].  Note 
the scale difference in Mercator is about half a zoom level.

[1] http://www.imagico.de/files/LC81940262016238LGN00_crop.jpg
[2] http://www.imagico.de/files/LC81730622017061LGN00_crop.jpg


Christoph Hormann

More information about the talk mailing list