[Imports-us] NJ Landuse import (NJ2002LULC)

Serge Wroclawski emacsen at gmail.com
Wed Aug 21 18:15:35 UTC 2013


On Wed, Aug 21, 2013 at 10:32 AM, Reiser, John J. <Reiser at rowan.edu> wrote:

> I'm the original importer of the NJ 2002 Land Use data. First, I would
> like to say that I am against removing the data outright without a plan to
> replace it.

John,

Thanks for chiming in. Part of the reason I made this proposal was to
light a small fire under your butt- so at the very least we can get
this actively discussed and fixed (though my preference is still to
toss it).

> I would argue that any "simply wrong" areas are ones that have
> significantly changed since March 2002.

Here's where the cemetary/school was:

http://www.openstreetmap.org/browse/changeset/17435012

Here's the cemetary itself:
http://www.openstreetmap.org/browse/way/40165192/history

Obviously I can't go through the whole state and find all the wrong
instances, but the errors are this serious, I get concerned about data
quality.

Since this school was in GNIS, it's unlikely to be a new school (and
who puts a school on a graveyard?).

I also just don't believe that all the scrub is scrub. I've lived in
Jersey, and I checked places where I grew up- places I'd call "light
forest" (small undeveloped land with trees but not a complete forest).
These are not theoretical places that I've only seen in Bing, these
are places I've lived in.

I'm wholeheartedly against any plan that reads as "Remove the existing
data and replace it with new data".

The difficulty is that we have two separate main categories of problem:

1. Import issues

2. Data quality issues

Even if the new import addresses the second, with the sheer quantity
of data we're talking about, I have concerns about the first.

Better would be a plan to revise the data, rather than to remove and replace.

Additionally, such a plan would be usable to handle how the data would
be handled in the future.

> While the data was 7 years old at the time of import, it was
> the most current, comprehensive land use data available for New Jersey.

To be clear, this is land-use, rather than land zoning. Land zoning
data has /no/ reason to be in OSM, where as land use is more of a gray
area in my mind.

> My  main reservation is that almost all of the original changesets have since
> been touched and many of the source tags have been removed.

Changesets or objects?

I don't think it's true that most of the objects have been modified
since the upload (though I could run some analysis and see if that's
the case).

But for the objects which have had their geometries modified, wouldn't
this be the effect of OSM users attempting to refine the data?
Conflation needs to be part of any plan regarding external datasets.

> I would be more than happy to work with other interested mappers on the
> 2007 land use data and develop a plan to replace the existing 2002 data
> with the 2012 once it is available.

Why not develop the plan now (with accompany software to do the
complex geometry modification and conflation steps)? Then you can use
whatever new dataset is available....

- Serge



More information about the Imports-us mailing list