Geoscience Reference
In-Depth Information
products by the Ordnance Survey (OS) to the general public, as well as the widespread availabil-
ity of products such as OS MasterMap to the UK academic community, is an excellent illustra-
tion of this trend. In part, this may be stimulated by the appearance of high-quality competing
products on a free-to-use basis, such as Google Maps and Google Earth. A third dimension
to this trend is the advance of mapping products from the bottom up: a couple of years ago,
OpenStreetMap was a rather sporadic patchwork of local networks varying in quality and cover-
age and contributed by a relatively small number of enthusiasts. Now it is possible to go almost
anywhere and find good coverage.
Socio-demographic data have also been impacted by the outburst of new data sources. For many
years, the census has been considered the gold standard of small area demographic data. Now the
census is complemented by sources such as hospital and patient records (particularly valuable as
a source of migration data), pupil records, housing and electoral data and even commercial and
market research sources (e.g. Acxiom's research opinion poll - see Thompson et al., 2010). So pro-
nounced is this trend that it now seems a distinct possibility that 2011 could see the last traditional
census of the United Kingdom, to be replaced by some more imaginative combination of alternative
sources, an idea that is already established to some degree in countries such as the Netherlands and
Sweden (Ralphs and Tutton, 2011).
In a recent project, we set out to demonstrate the benefits of integrating a wide variety of
data sources for a crime analysis problem. Spatial analysis is commonly used by police forces
around the world to support the understanding of crime patterns and allocation of resources (see
Chainey and Ratcliffe, 2005). There is no doubt that geodemographic classifications can contrib-
ute something to this understanding (e.g. Ashby and Longley, 2005). However, geodemographics
fails to recognise many factors, including local patterns of exposure relating to quality of lighting
or road access, vacancies and derelictions, the layout of the transport infrastructure and the ebb
and flow of daily movements around the city region. In the Geospatial Data for Crime Analysis
(GeoCrimeData, http://geocrimedata.blogspot.co.uk/) project, we combined geodemographics and
census data with land use and building data (OS MasterMap), workplace data (from directory
sources), road network data (from OpenStreetMap) and others, in order to create an extended pack-
age of analytics for the investigation of crime patterns. It was possible to demonstrate the value of
data integration (and the associated analytic developments) to an academic audience (e.g. Birkin
et al., 2011) and also to a community of practitioners where a log of the meeting can be found
at http://geocrimedata.blogspot.co.uk/2011/09/producers-and-consumers-brainstorm.html. For
example, Figure 10.6 presents a selection of secondary data (generated as part of the project) that
illustrates some important geographical features of the physical environment which can have an
influence on the vulnerability of houses to residential burglary. Traditionally, practitioners would
have to estimate these features by hand - see, for example, the manual classification of cul-de-sacs
performed by Johnson and Bowers (2009). These risk factors can be combined to form an overall
risk profile which could be useful for practitioners working at a local level, that is, that of the indi-
vidual road or neighbourhood.
However, genuinely crowdsourced data, in which new information is generated from an online
audience, open perhaps the greatest possibilities for the future generation and analysis of original
information sources. In a recent experiment, we demonstrated the value of the NeISS e-Research
framework which was introduced in Section 10.2. In 2008, following the release of contentious
proposals for a Greater Manchester congestion charge, the group at CASA was approached by BBC
North West to conduct a web survey of attitudes to the new charge. With the assistance of local
TV and radio, it was possible to obtain 16,000 responses to this survey within a 24 h period. The
responses detailed, by postcode of residence, how people felt their choice of travel mode would be
impacted by the changes that were proposed. Online mapping of these data made for an informative
and newsworthy piece. However, by the integration of spatial modelling workflows, , it was possible
to go much further. Mode preferences were obtained from the respondents which provided behav-
ioural data for calibrating the transport simulation (Figure 10.7a). We also used responses from the
Search WWH ::




Custom Search