I sent the following to the TMIP (Travel Model Improvement Program) listserv earlier this morning, but since this is focused on CTPP, I would be happy to get comments from the CTPP group as well.
It is somewhat common practice to compare district-to-district flows consisting of aggregations of TAZ-to-TAZ travel model outputs for Home-to-Work person trips to similar home-to-work aggregations based on CTPP flows. Since most AM peak period auto travel consists of home-to-work movements, what successes or failures have forecasters had in the use of TAZ-to-TAZ CTPP data in an AM peak period UE traffic assignment? Or some variation that might include enhancements to the commuter flow data to represent the (relatively small) non-commuter AM Peak vehicle travel and/or other adjustments to the CTPP’s representation of home-to-work travel? I am aware of one test that notes the “lumpiness” of working directly with the TAZ-to-TAZ level CTPP data (with the end result being the %RMSE statistics do not look stellar), but am also aware of many forecasters who manage to squeeze good value out of whatever ground truth-based data is available. I am reluctant to bring up the “origin-destination matrix estimation” approach that can start with even a seed matrix of 1’s in every TAZ to TAZ cell, but would be happy to hear of any CTPP-based efforts where the study has summarized how the “raw” CTPP trip table matrix compares to the final adjusted “seems to assign well” matrix.
I am not expecting to see many responses that focus on use of CTPP in this type of exercise, and suggest that you reply directly to this listerv rather than just to me, since whatever info you might offer might inspire others to comment. But if there is some reason why that may be a problem, I am happy to get your comments directly to the email noted below (and can send to the full listserv without attribution to you).
Thanks in advance,