Ken is totally correct. You need to convert the CTPP commuter matrix
into something like an "observed home-based work" trip table using trip
rates per employed residents from your local travel survey, the NCHRP
#365 (1.50 trips/worker), or the NHTS. Remember that the census only
provides a count of workers, and is not a count of work trips, per se.
We (SF Bay Area) did *not* use the CTPP2000 Part #3 commuter matrices
in our Year 2000 model validation project. Our model validation project
ended in April 2004, and the CTPP Part 3 data was available May 2004.
Such is life. On the other hand, we did use the 5% PUMS data (available
August 2003) to construct what we needed: county-to-county commuters by
means of transportation by household income quartile. We then converted
these commuter matrices into "observed home-based work" trips. For our
validation, we sequentially validated our auto ownership (to Census PUMS
data), work trip generation, work trip distribution, and work trip mode
choice models to these census-based observed databases. We then went
through some fairly exhaustive "network validation" stages where we
compared the model simulated traffic (daily and peak) and transit
(daily) to observed traffic and transit counts. At this network
assignment validation stage we diverged from our "observed HBW" trips in
order to better match observed traffic/transit databases. The basic
problem we were having with the PUMS-based observed work trips was that
it was too low (transit, especially) in our Transbay corridor (SF
to/from the East Bay). This is a problem with either too few commuters
in the decennial census, or perhaps our trip rate for this distance of
work trip was too low.
So, my recommendation is to use the commuter matrices from the CTPP,
convert them into appropriate "observed home-based work" trips, then do
initial stages of validation against these observed databases, BUT be
prepared to diverge from these CTPP-based observed trip tables if and
when your transit/traffic assignment validation is inadequate. OR, you
can always "hold true" to the CTPP-based observed data, and focus your
attention on calibration / hammering away at your non-work trip models.
Chuck Purvis, MTC-Oakland
>> Ken Cervenka <kcervenka(a)nctcog.org>
08/16/04 10:46AM >>>
On your question #1: One thing to keep in mind
relates to the fact
"HBW trip" for modeling purposes represents a trip end, e.g., a Home to
trip and a Work to Home trip would actually be counted as two separate
trips, i.e., HBW productions and two HBW attractions.
So, using your NCHRP example, 1,000 employees would generate 1,500 HBW
trips--but only half of these (750) would be comparable to a
In terms of why travel survey data never supports 2.0 HBW trips per
employee: It is partly because of absenteeism on an "average"
partly because some jobs are simply not five-day-a-week jobs, and
because some (or even many) people don't actually go straight from home
work (or from work to home) on their surveyed day (e.g., they engage in
trip chaining, which turns some Home-to-Work trips into HNW and NHB
and turns some Work-to-Home trips into NHB and HNW trips).
Hope this helps!
[mailto:email@example.com]On Behalf Of Jiji Kottommannil
Sent: Monday, August 16, 2004 12:09 PM
Subject: [CTPP] CTPP Part 3
I am a new user of the CTPP data and am in the process of comparing
part 3 tract-tract worker flow data to the Home-Based Work trip table
from the travel demand model I am currently working on. I came across
few questions/issues for which I would welcome your
1) Do the CTPP part 3 flows account for all the home-based work trips?
noticed that the commonly accepted home-based work attraction rate is
about 1.5 trips per total employment (NCHRP 365) and this creates more
HBW trips than there are in the CTPP part 3. Let me know if I there is
something wrong in my interpretation of the data.
2) How much success have the MPO's had in calibrating their trip
distribution models based on CTPP Part 3? What level of accuracy is
Jiji V. Kottommannil
Transportation Modeling Specialist,E.I.T.
Crawford Bunte Brammeier
Phone: 314-878-6644, Ext. 38