TO: CTPP News mailing list & MPO staffs
FR: Chuck Purvis, MTC
We received our copy of TIGER'98 and the TAZ-UP software on April
7th. I already had ArcView 3.1 installed on my 233 MHz PC, actually
on a 1 gigabyte jaz cartridge. ArcView takes up 83 MB of space;
TAZ-UP an additional 61 MB. So far, so good.
Installing county TIGER files into TAZ-UP can be *very* slow for
large urban counties. I started installing Alameda County (CA) on
Friday, April 9 at 4:00 PM. It took 60 minutes to install! My largest
county, Santa Clara County, took 90 minutes to install files of 40.2
megabytes in size. My entire region (9 counties) takes 205 MB in
space. The "log" records that are written as part of this
installation process note that I have 104,600 polygons (this number
of polygons is the number of records that should be exported from
TAZ-UP once all of my edits are complete).
RECOMMENDATION: Use the fastest PC you can afford. I believe the
documentation recommends at least a 200 MHz machine, but I would
highly recommend something like a 450 MHz PC. And I think that a
gigabyte of free disk space should be adequate for most regions.
After installing these county files, my first step was to "validate"
the existing boundaries: "invalid boundaries"; "contiguity errors";
and "completeness errors". This is where I initially started to freak
out: we have 2,445 invalid boundaries in our region; 106 contiguity
problems; and 1 completeness problem. The contiguity and completeness
errors are manageable; the invalid boundary errors were highly
unexpected (and may be unmanageable?)
Many of my invalid boundary errors are "F84" errors which are
basically old 1990 block boundaries (be they tract boundaries or
block group boundaries) that are flagged by the Census Bureau as
invalid because they aren't associated with any physical feature
(e.g., roads, streams, power lines). My spot check of my invalid
boundary errors suggest that 90 percent of mine are "F84" errors. I
even got "F84" errors in a county (Napa County) where I was strictly
using 1990 Census block groups as my 1990 TAZes!
My desire is that my Year 2000 TAZ boundaries are nested on Year 2000
Census Tracts. I essentially want to create what I would call
"user-defined block groups." The big issue is "what are my year 2000
Census Tract boundaries?" This is where you (MPOs working on this
project) will need to contact your regional office census geographer
to check up on the status of your Year 2000 Tract Boundaries. I may
be wrong in my interpretation of this, but Year 2000 Tract boundaries
will be valid CTPP TAZ boundaries.
I had numerous phone conversations this past Friday with our USDOT
and Census folk back in DC, and they've expressed concerns about our
plight regarding these invalid boundary issues. I'm still fuzzy on
this, so I'm hoping that the Bureau and USDOT folks can clear up
these issues before we all get too hot under the collar with all of
these invalid boundary issues. (USDOT is also very concerned because
all of the TIGER files for the entire US aren't out yet :-(
The "contiguity errors" are very manageable. Some of these contiguity
errors are due to correspondence errors that are our fault, leading
to some really unintended results. The rest of these errors are due
to a misallocation of "water blocks" that separate real, physical
islands. An example in my region is Alcatraz Island and Treasure
Island in the middle of San Francisco Bay. They're part of the same
TAZ, but were separated by a "water block" that includes much of the
Bay, plus Fishermans' Wharf and Pier 39. So, a simple solution is to
include the Fishermans' Wharf "water block" with Alcatraz Island, as
one TAZ, and have Treasure Island as its own TAZ.
My point is that I believe that all of the contiguity errors in your
TAZ boundary files can (and probably should) be eliminated. Be sure
to have "water blocks" connect your sandbars and physical islands to
your "mainland" blocks.
In terms of the TAZ-UP software, I have no problems. It works as
advertised, though it's all (ArcView/TAZ-UP) very slow with large
urban counties on a 233 MHz PC. I wish there were "multiple un-dos"
since I tend to make mistakes in batches, but I'll learn to make
only one mistake at a time.
Well, those are my first impressions on TAZ-UP and this TAZ
definition program. Just take it easy and don't freak out (too much)
when all your invalid boundaries start to pile up!
MTC - Oakland/San Francisco/San Jose metro region.
Chuck Purvis, AICP
Senior Transportation Planner/Analyst, Planning Section
Metropolitan Transportation Commission
101 Eighth Street, Oakland, CA 94607-4700
(510) 464-7731 (voice) (510) 464-7848 (fax)
MTC DataMart & InfoMart:
MTC FTP Site: ftp://ftp.abag.ca.gov/pub/mtc/planning/
Personal WWW: http://home.earthlink.net/~clpurvis/