Geoscience Reference
In-Depth Information
14.1 INTRODUCTION
In much the same way that Bivand and Lucas (2000), a chapter in the irst edition of this topic on
the integration of models and geographical information systems (GIS), was a review of literature,
this chapter will consider relationships between GeoComputation (GC) and open-source software.
Some of the insights from our earlier work in fact fed directly into the development of interfaces
between the open-source GRASS (Geographic Resources Analysis Support System) GIS and the
R statistical language and environment, as initially described by Bivand and Neteler (2000). This
positive feedback between GC and software development has been important both for GC and for
software development. Software development enables GC to be applied to wider ranges of problems
across increasing numbers of fields of study. Sustained contact and interaction with the GC com-
munity enriches and strengthens software development, by ensuring that progress in the field is
acknowledged and used in practice. The structuring of relationships between software components
(defined in Section 14.2.2), with ensuing workflow challenges and opportunities, has matured over
time, informing GC communities using either open-source or proprietary software, or both together.
An aspect of the progress made in software development communities has been the ratio of noise
to signal in information diffusion. Books such as Mitchell (2005), Erle et al. (2005) and Gibson and
Erle (2006) gave rich insight into myriad possibilities for committed customisers and consultants,
but at a distance from what might be termed “mainstream” GIScience; perhaps “hacking” and
GIScience are more comfortable at a distance? Applied research often, however, lives between these
two places and needs to find practical solutions to real problems within the constraints of available
hardware, software and programming and scripting competence. It is perhaps a paradox that very
little software used to tackle real scientific problems is written by programmers with a background
in computer science nowadays; much is written by domain scientists with deadlines to meet.
As many, including recently Rey (2009), have pointed out, the involvement of domain scientists
in coding has effectively “included” the code in their research output, making its openness for
scrutiny important for the verification of project results and methodologies. Different disciplines
approach this question in different ways, with some journals still unwilling to allow software to
be cited in references and unhappy about fully documented software footnotes; others require the
submission of supplementary materials including code for the convenience of referees and readers.
Access to code to permit research to be reproduced is becoming important in many disciplines, as
Leisch and Rossini (2003) show with respect to statistics.
Voices of free and open-source software insiders like Ramsey (2007) are important, because
they suggest the apparent level of reflection available to those developers closest to the bug track-
ers. More reflection is perhaps shown in contributions such as Câmara et al. (2012), but in Ramsey
(2007), we are reading a narrative written by a developer with commit rights to major open-source
geospatial software projects. His distinction between the 'C', the 'Java' and the '.Net' tribes seems
well taken, fairly reflecting the ways in which developer communities have evolved; we will return
to these communities later in the chapter.
The field of geospatial open-source software projects was surveyed in detail by its participants
in Hall and Leahy (2008a), and their descriptions constitute a clear picture of the ways in which
they see their contributions. Some of the chapters have no references and are obviously statements
by developers with practical rather than academic goals. Other chapters are more similar in char-
acter to two other topics published in the same year, Neteler and Mitasova (2008) and Bivand et al.
(2008), both of which aim to provide applied researchers with guides to the software tools they may
find useful in carrying out their work.
This practical approach to the conduct of research is noted by Sui and DeLyser (2011) in the con-
text of academic geography, which one might hope will make helpful contributions in the future after
a period of discriminating against quantitative methods even where they were appropriate. Recent
years have seen surveys of the potential of open-source geospatial software in areas as diverse as
health geographics and spatial epidemiology (Fisher and Myers 2011, Vanmeulebrouk et al. 2008,
Search WWH ::




Custom Search