Geoscience Reference
In-Depth Information
to explore alternative
what if
scenarios relating to the extent and intensity of the congestion charging
proposals (Figure 10.7d). Further discussion of the simulation exercise is provided by Birkin et al.
(2011) in which links to data and model code are also provided - a tangible example of the publica-
tion and reproducibility of geosimulation experiments.
10.10 FROM THE STRATEGIC TO THE IMMEDIATE
In the past, the focus of GC has been on problems which are essentially static in their nature. The
objective has been to find patterns and statistical relationships in the data or to calibrate models as
a basis for predictive forecasting or scenario planning. Increasingly, however, the availability of
distributed computational architectures is placing the emphasis on real-time data generation and
decision-making.
Many of the most interesting developments in this domain are associated with transportation. For
some time now, sensors within cities have been able to detect the speed and movement of traffic.
It is then logical to ask the question of whether such data cannot be used as the basis of real-time
models which are capable of regulating and improving the traffic flow. Indeed, agent-based simula-
tion has been proposed as a means for the
self-organisation
of traffic lights which 'could improve
traffic flow by up to 95%' (Lämmer and Helbing, 2008), not only reducing congestion and unneces-
sary expenditure of time but improving efficiency and reducing the environmental impacts of road
traffic. Real-time data processing and modelling technologies are now increasingly promoted as
practical tools for managing the sustainability of urban traffic networks (The Climate Group, 2008).
In the future, it seems likely that similar technologies will extend to individual vehicles. Is there
any reason, for example, why individual vehicles might not calibrate their own routes based on
previous experience, the current conditions on the road and the objectives of the driver (e.g. to
minimise time or maximise fuel efficiency)? At the present time, this task is beyond the process-
ing capacity of on-board devices, but this will surely change. As an example, it is already the case
that the time it takes to drive between pairs of locations can be monitored and captured by satellite
navigation devices such as
TomTom
. These drive times are widely used in applications like retail
location planning, and so it is the case that where accessibility has been previously estimated on the
basis of rather unsophisticated models of road linkage and average speed, these estimates can now
be calibrated much more closely to real data.
Real-time data feeds - in association with simulation modelling and scenario planning - will be
more generally applicable to a much broader class of problems involving distributions in space and
time. For example, the air quality in cities is something else that can be remotely sensed, interpo-
lated and displayed, at the very least providing useful information for pedestrians or cyclists in plan-
ning routes or recreational activities. The
mappiness
project uses mobile telephones and a simple
survey form to capture mood patterns (or levels of
happiness
) and then attempts to relate this to
the environment and activities in which the respondent is engaged (http://www.mappiness.org.uk/).
This idea therefore starts to link crowdsourcing with real-time data analysis in a rather interesting
way. Mappiness has a user base of
only
some 50,000 individuals, but given that this is essentially a
survey that has been created for a PhD project, exponential growth in crowd participation in proj-
ects of this type can be anticipated. Another example is CASA's CityDB project, which has begun
to assemble a real-time dashboard of urban life in six UK cities (http://citydashboard.org/london/).
Again, it seems only a matter of time until this kind of intelligence is augmented with model-based
reasoning for decision-making and scenario planning.
Crowdsourced data are also likely to become an important means of conducting
real-time
simu-
lation modelling. The variety and scope of volunteered data from the crowd are continuing to esca-
late at an incredible rate. For example, social media applications such as Twitter and Foursquare
are providing information about the movement and activity patterns of individuals on a constant
and global basis. In the city of Leeds alone (with 750,000 people), an experiment to monitor indi-
vidual behaviours through the twitter messaging service has captured over 1 million individual data
