Geoscience Reference
In-Depth Information
contemporary computational infrastructures and explain how exciting new approaches to GC are
beginning to emerge at this new research frontier.
E-Science will always be an evocative term for computer scientists in the United Kingdom and
elsewhere, not least for its ability to unlock the vaults of research funding to an unprecedented
degree. The major push to development was in the years 2001-2004, in which UK Research Councils
provided no less than £340 million of e-Science funding to universities and research institutes.
On the back of this funding, a number of impressive applications were produced, most prominently
in the physical sciences. AstroGrid, for example, is a multi-institutional project led by researchers
at the University of Edinburgh aiming to create a 'Virtual Observatory… (for)… the world's astro-
nomical data' (astrogrid.org) which embraces one of the key themes for e-Science of data-driven,
inductive research (Bell et al., 2009). Traditional research is hypothesis-driven, so the argument
goes. Thus, if a researcher wishes to find, say, a black hole in space, then some reasoned argument
for its location is formulated, a telescope is pointed in that direction, and tests are undertaken to
confirm the presence of that object. According to the latest paradigm of data-intensive research
(Bell et al., 2009), it is assumed that if enough data of the right kind can be provided, then the right
hypotheses will follow later. Thus, the challenge now is to map out the whole sky with telescopes
and then to trawl through the data searching for interesting things. It is an argument that is mirrored
in fields such as bioinformatics, where it is argued that computational infrastructures for designing,
sharing and reproducing numerical experiments with bioinformatic data are now the key facilitator
for new knowledge generation and are making classical experiments in wet laboratories increas-
ingly redundant (Stevens et al., 2003; see also Laffan, 2014).
It is useful to note that although geography and spatial analysis have come relatively late to
the e-Science party, nevertheless, a data-intensive approach to spatial analysis has been strongly
advocated by Stan Openshaw (one of the editors of the irst edition of this topic) in a series of
papers in the late 1980s and 1990s referring to concepts such as the geographical analysis machine
(Openshaw et al., 1988) and geographical explanation machine (Openshaw and Turton, 2001). In the
third part of this chapter, we will discuss examples of the e-Science concept in relation to human
systems and regional science, the environment and physical geography, using examples from Leeds
and elsewhere. The implications of the latest developments, current directions and future possibili-
ties will all be considered in the concluding section.
10.2 E-RESEARCH CAPABILITIES AND INFRASTRUCTURE
A simple view of grid computing is that by linking together processing resources on a huge scale,
it is possible to perform complex transformations on data sets of an ever-increasing size and vari-
ety. The benefits of this can be seen in the cyberspace applications which are now a commonplace
of everyday life - for example, video-sharing sites (YouTube), search engines (Google) and social
media (Facebook) are all reliant on massive processor farms to handle and distribute data at diz-
zying speed (e.g. http://highscalability.com/google-architecture). In this view, the
grid
concept is
analogous to a power grid in which, whenever an application is
plugged in
to the grid, the compu-
tational power required to execute the application is automatically and seamlessly available - just
as one would switch on a light or boil a kettle. Although this vision has not yet been fully realised,
some technologies and services are emerging under the name
cloud computing
that come close to
the pervasive, seamless and invisible computing that grid technologies aspire to (which are dis-
cussed in more detail in Section 10.6). To facilitate the flexible deployment of application codes
across a variety of hardware platforms,
grid middleware
technologies are being developed. There
is no doubt that this computational resource perspective is important - not least for spatial analysis
and applications of GC including simulation modelling and geovisualisation - as we will show later.
However, to present
the grid
in such a limited way is largely to misunderstand the significance
of these developments in hardware and software engineering. A far more relevant view is to under-
stand grid computing as the provider of services which underpin a virtual organisation. As the
