Andrew,
this is an interesting idea and a nice thread - Stephen's results on area
averages coincide exactly with ours for central London areas - the main
difference being that the London areas we looked at are predominantly mixed
residential, retail and office areas.
I think Juan puts the case well for the need to take account of density and
land use, and again our results on London show that these are key factors in
pedestrrian movement area averages. I can see that attractions of a
configurational model that incorporates density (we have tried in the past
weighting axial lines according to the height of buildings by adding lines),
however my approach in this case would probably be slightly simpler than
Juan's.
I would suggest that all that is needed is a regression model in which the
effects of land use and density are incorporated as potential determining
variables. I would be confident (on the basis of the previous studies that
we have done of London areas using this method) that this regression model
would be good enough for one to make estimates with some confidence of the
pedestrian flows in areas that had not been observed, and so arrive at a
reasonable figure for total pedestrian km walked in Central London. The
things I would think likely to be problematic would be in getting access to
good land use and development density data at the axial line scale
(valuation office data might be good enough) and then in getting updates to
these data with the frequency needed if one were to try and track changes.
Ultimately, what you really need if I understand it right, is not just an
estimate of the pedestrian walking volume, but an error estimate as well.
This actually requires a thorough piece of research as error estimates for
the pedestrian mode have never to my knowledge been developed, and certainly
not at the level of the whole central area of a city as big as London. This
could form the basis for a nice responsive mode research application to the
EPSRC if anyone (Juan, Andrew?) were interested in collaborating... :-)
Alan
> -----Original Message-----
> From: [log in to unmask] [mailto:[log in to unmask]]On
> Behalf Of Andrew Smith
> Sent: 04 December 2001 10:37
> To: [log in to unmask]
> Subject: scaling observed pedestrian flows to an area total
>
>
> Dear colleagues,
>
> I'm considering using a space syntax model to estimate the distribution of
> pedestrian flows across London. The intention is that if the distribution
> is representative, then by calibrating the flows based on observations on
> some sampled links, I can then scale up to calculate the total amount of
> walking on the network. Then, by continuously monitoring some
> links, I can
> estimate total annual person-km walked in London.
>
> Are any of you acquainted with a previous application of this kind? Any
> pointers to potential hazards and ways to improve the robustness of the
> methodology will be very welcome.
>
> Andrew Smith
>
|