[Gdal-dev] Problems with ogr2ogr and TIGER line data - California is a bit off

[Martin, Daniel A]

> As Ed mentions, you should check into whether there is 
> anything odd about the lineage of your California TIGER data. 
>  Are you working from raw TIGER data from the Census bureau?  
> All from the same year?

Yes, I create the entire country out of the same year at one time.  I
have run this against new TIGER data as it comes out every year.  I have
independent layers for each year.  I checked and all years exhibit the
same problem I'm describing.  So unless raw TIGER data has been wrong in
California for years...

> The TIGER data is all distributed in NAD83 decimal degrees.  
> There isn't much that can go wrong with from a projection 
> point of view.  
> 
> What coordinate system are you writing out your tab files in? 
>  Also just the original NAD83 or are you transforming to 
> state plane?  There might be ways the state plane definitions 
> could be messed up. 

I guess I don't know.  I'm not doing anything special.  I download the
files from the Census TIGER website, and then run:

ogr2ogr -f "MapInfo File" mydatadir

I'm doing absolutely nothing else to the data.
 
> Also, TAB format scales locations to integers based on a 
> min/max range that is supposed to be computed based on the 
> coordinate system.  The default for geographic coordinates 
> systems should be a -180 to 180 plane and should be fine.  
> But if you accidentally got a projected bounding box the 
> resolution would likely be really crappy.  This shouldn't be 
> a systematic offset though, just sort of random perturbation 
> so I doubt this is the issue. 

I'm not reprojecting.  I'm pulling the output of ogr2ogr straight into
MapInfo, and also straight into MapServer without touching it.

> 
> My suggestion is to use ogrinfo on a known feature from TIGER 
> and in your TAB file to see if the coordinate match.  If they 
> do, then it is likely the problem is the original tiger data. 
>  If they don't match it is something about the translation, 
> perhaps in the TAB code.  

I'm not sure I have any idea how to do that.  I will try.

> 
> If you are reprojecting the problem might be the coordinate system. 
> 
> You might want to try converting one to mid/mif and see if 
> you get the same issues. 

When I view any state other than California, it is located in MapInfo
identical to how it is located in MapServer.  However, when I view
California in MapServer, California alone is in a different location
than the same layer in MapServer where OGR is used to retrieve the data.

My point being, now we have two distinct situations where OGR is
producing inconsistencies in California with TAB files.  Once in the
ogr2ogr conversion, and then on the rendering into MapServer.

Thanks for your time.

-Dan