Some of you have read with wonder (and wondered) about the Inner Modal Subtractive gravity model formulation used in TMODEL2. We use it daily and find that it can control distributions very well. We have, however, been wrestling with the meaning of its components to better understand why they seem to work.

First, why wasn't this an issue with UTPS models of the '60s and '70s? Why was this not confronted before? The answer is that they were generally regional models and did not have zones very close together.

Our original notion was (and hence the name) that the purpose of the extra term, k^*t, was to suppress travel between close zones due to the competition of walking, taxis, busses or other inner or short-haul modes.

While this is reasonable, there was still a problem. Suppression of short trips, even in areas where busses and taxis don't exist, seemed to work well even beyond the distances which we Americans (and others) are willing to walk. So, we have been searching for a better explanation and feel that we have one. We would, of course, like to hear your ideas concerning both its utility and why it seems to work.

We now theorize that the suppression of close trips is needed because of typical land-use distribution. While there may be plenty of urban land-use in reasonable proximity, the particular land-use required to satisfy the trip purpose may not be nearby. There are residences in many areas, grocery stores in fewer, large department stores in even fewer, and professional sports arenas in fewer yet. For most individual trip purposes, a satisfactory destination may not be available in the few nearest zones.

You may reply that short trips are already somewhat suppressed simply by the fact that there aren't as many destinations available within a short distance. That is true and that, in fact, accounts for much of the characteristic shape of the inner portion of the trip length frequency distribution curve.

However, traveler behavior must be accounted for by the gravity model and its parameters. The results of origin-destination surveys show that people do NOT tend to fulfill their trip purposes at the "attractions" nearest them. Whether this is because the available attractions aren't the right attractions or because people choose to live comfortably distant from employment and shopping areas, the typical behavior is to not take many short trips.

To model this behavior, we must use formulations which somewhat suppress the possible short trips. Hence, the I Must Suppress gravity model.