Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!bloom-beacon!mit-eddie!apollo!oj
From: oj@apollo.COM (Ellis Oliver Jones)
Newsgroups: comp.graphics
Subject: Re: Computerized maps
Message-ID: <43d87f13.d5b2@apollo.COM>
Date: 15 Jun 89 13:42:00 GMT
References: <2645@portia.Stanford.EDU> <11982@well.UUCP> <3867@eos.UUCP> <3895@amiga.UUCP> <1284@netxcom.UUCP>
Reply-To: oj@apollo.com (Ollie Jones)
Organization: Apollo Computer, Chelmsford, MA
Lines: 57

In article <1284@netxcom.UUCP> jallen@netxdev.UUCP (John Allen) writes:
>The data referred to is the Census Bureau DIME files.

About nine years ago, I had the great privilege ;-) of working with
the Census Bureau's GBF/DIME files.  They're pretty good for
what they are.  What they are is:

* 160 bytes per record (I think, this is hazy in my memory)
* One record per block per city street, and others for 
    borders, water features, railroad rights-of-way, 
    Interstate highways, etc.
* Each record contains such things as:
   - Street name
   - House numbers, left side and right side
   - Census tract/block numbers, left side and right side
   - Block endpoints in lat/long,  state plane coordinates, and
       (I think) Universal Transverse Mercator coordinates.
       No elevations are provided (you get x and y, but no z)

These files get pretty big.  There are a lot of blocks and streets in
places like Chicago.

They covered, in '79, only cities ("Standard Metropolitan Statistical 
Areas").  Not surprisingly, areas which have experienced massive growth
have less accurate GBF/DIME files.

The census bureau uses these tapes for such things as
geocoding (making maplike plots) of census data, and verification 
of address records (to make sure the census takers know where
every house is).

In 1979, the quality of these data bases (as measured by
completeness, internal consistency, and positional accuracy)
was widely varying.  Between censuses, each locality held a
large amount of responsibility for maintaining the data bases.
Some localities used the data for municipal planning;  these
files were often excellent.  Others didn't care at all, and the
files showed it.  I'm convinced that some of the position data
on some maps was digitized from USGS quadrangle maps with a mouse.

When everybody went into panic mode for the '80 census, the quality
improved dramatically.  This didn't just happen, but took the hard
work of large gangs of digitizers and cartographic quality
inspectors.

Even when they're at their best, the plotted output from these 
files isn't very pretty:  they're simple stick maps, the wireframe
of the mapping business.  If you want to make nice-looking maps,
you may want to try USGS data, or Defense Mapping Agency data
if you can get it.

GBF/DIME files are quite good for address matching
(e.g. "Estimate the latitude and longitude of 750 Fifth Avenue, New
York City") as long as you can tolerate spotty coverage and inaccurate
positioning.

/Ollie Jones (speaking for myself, not necessarily Apollo Division, H-P).