[Imports] [Cat2Osm2] Tool for exporting Spanish Cadastre data in OSM suitable format

[Paul Norman]

> From: Cruz Enrique Borges Hernandez [mailto:cruz.borges at deusto.es]
> Subject: Re: [Imports] [Cat2Osm2] Tool for exporting Spanish Cadastre
> data in OSM suitable format
> 
> > There are some duplicate nodes. An example can be found at 40.8253351,
> > -5.4437406. I'll be talking more about duplicate nodes later.
> 
> I am not sure about this (most probably is a duplicate node in the
> original data) but in any case the JOSM validator should catch these. We
> have tried very hard to reuse as much nodes and ways as the data allow
> us.

My guess is that it's something to do with combining the layers. Not a big
point if you're using the separate layers.

> > Tagging:
> >
> > You appear to be using landuse=reservoir for unknown water. Most of
> > the ones I looked at would be better suited as natural=water.
> > natural=water is also better suited for unknown bodies of water.
> 
> Geometries with water tags are one of the most difficult tags to
> translate.

But in this case would not natural=water be better suited?

> The problem comes from the original data: it does not maintain internal
> consistency between different towns. Others difficult tags are parking
> spaces and greenfield for the same reason. In the end, all of them (and
> probably others ones that will appear in other regions) should be
> revised manually by the local community. We have already planned to have
> wiki pages with regional information about the import process given tips
> about this things.
> 


> > Are there any tags that will be used other than those in this sample
> file?
> 
> This is the full list (sorry about the spanish):
> http://wiki.openstreetmap.org/wiki/Traduccion_metadatos_catastro_a_map_f
> eatures
> 
> Please, if you found any mistake, let us know before changing the table.

Well, it's not possible to review them without also looking at the cadastre
data. For example, how could I verify that what is in that table as "Jardín
que se valora" should really be tagged as leisure=garden? It's impossible to
do so without looking at the data from the catastro.

> > How were the tag translations arrived at?
> 
> The translation tags was made from:
> - What catastro documentation said:
> 
> http://www.catastro.meh.es/pdf/formatos_intercambio/catastro_fin_cat_200
> 6.pdf.
> - What experts mappers have said. In fact, the pages have been almost
> completely filled by expert mappers from the Spanish community.
> - What persons close to the catastro office have said to us.
> - Obviously, our own experience using this data in other projects.

It sounds like you are relying on the catastro documentation without
checking each feature type with another source. This is an extremely bad
idea for two major reasons:

1. It relies on being able to accurately translate the catastro
documentation into OSM tags. Documentation is frequently unclear or
ambiguous and words may have multiple definitions.

2. It relies on the catastro use reflecting their documentation. I haven't
seen a large GIS data source yet where there weren't some differences
between the data source and its documentation.

My preferred approach is to pick a feature type, find multiple examples of
it in the data, compare them with imagery or other sources, then use that
information combined with the documentation to determine what to tag it as.

> > Taking the example of
> > landuse=greenfield, was this decided on based on descriptions from the
> > catastro, or was it based on inspection of the areas that it applies
> > to with aerial imagery, surveys or other resources?
> >
> > The areas tagged with landuser=greenfield aren't areas with no past or
> > present buildings scheduled for development (i.e. virgin ground that
> > will be turned into a construction site)
> 
> Our experience over that specific tag is that catastro have overused it:
> there are places tagged (in catastro) as a greenfield that should be
> another thing (normally landuse=village_green or recreation_ground).
> Probably this is due to differences in the definitions of the two
> ontologies.

It's wrong to say that the catastro have overused it. They do not use OSM
tags. They use an attribute that according to the table is defined to mean
"Suelo vacante, sin construer," or "Vacant land without building." This is
different than landuse=greenfield.