Pointers for the Future - Environmental Modelling: Finding Simplicity in Complexity - page 442

storage - has dramatically changed the role of modelling
in science since the 1970s. The technology enables more
and more calculations to be made for less and less money.
This rapid technological development is expected to
continue for some time to come with clear impacts on
our capacity to build ever more sophisticated models.
In addition to these changes in computer hardware,
significant developments in software and connectivity
have also created an enabling environment for modelling.
Improvements in hardware can only be of use if the
software environments keep up with these changes. Most
environmental modellers are not computer scientists
by training, so the development of powerful and easy
to use software environments, programming languages
and tools for modelling are critical to progress. Recent
software

availability of open source software libraries, which
mean that modellers can increasing focus on coding at
high levels of abstraction with the basic routines and
numerical recipes available for reuse.

The availability of increasingly sophisticated visualiza-
tion tools and libraries for the parsing of model input
(such as, for Python, pyHDF, 10 GDAL 11 and, for Java,
GDAL 12 ) and the display of model output (such as, for
Python, Matplotlib 13

•

and, for Java, JFreeChart 14 ).

The increasing sophistication and availability of
commercial - but especially of free and open-source -
remote sensing and GIS tools, which link with
programming languages and permit spatial modelling
(examples include PCRaster-Python, ESRI ARCGIS-
Python, 15

•

GRASS-GIS-Python, 16 SAGAGIS-Java, 17

R-spatial 18 ). developments of particular importance in

modelling include: The availability of sophisticated mark-up languages

(such as GML 19 and KML 20 ) for the display of geo-
graphic content in time and space using GIS, or
so-called Geobrowsers such as Google Earth, Google
Maps, Microsoft Bing maps, OpenStreetMap and NASA
WorldWind. This development has enabled much more
sophisticated communication of model outputs, espe-
cially to a lay audience.

•

The continued development of graphical user interface
(GUI)-based, so-called 'drag-and-drop' or graphical
coding modelling tools that enable the development of
mathematical models by non-programmers from sim-
ple conceptual models. Examples include VENSIM, 1
STELLA, 2 SIMILE, 3 MATLAB 4 and NetLogo. 5 These
tools have opened up modelling to a wide audience but
there are dangers in their use as 'black boxes' in which
different systems can give different results for the same
model because their numerical methods differ (Seppelt
and Richter, 2002).

•

In general, the recent trend away from proprietary
software development and towards free and open-source
software (FOSS) has much accelerated the rate of progress
since data formats have become more interchangeable
between software (facilitating collaboration). Moreover,
code reuse and linkage has given modellers access to a
range of sophisticated modelling software that can inter-
operate. For example, the Community Surface Dynamics

The development of easier to use and higher-level
programming languages that have components that
are specialized for environmental modelling, such as
PCRaster 6 for Python: PCRaster Python, 7 SciPy, 8 and,
for Java, the Eclipse Modeling Framework. 9 Python, in
particular, is considered a 'glue' language that brings
together a wide range of other computing resources.

•

10 See http://pysclint.sourceforge.net/pyhdf/index.html (accessed
6 April 2012).
11 See http://trac.osgeo.org/gdal/wiki/GdalOgrInPython (accessed
6 April 2012).
12 See

The development of increasingly sophisticated forms
of code modularity and reuse through, for example,
object-orientated

•

programming and the increasing

http://trac.osgeo.org/gdal/wiki/GdalOgrInJava (accessed

6 April 2012).
13 See http://matplotlib.sourceforge.net/ (accessed 6 April 2012).
14 See www.jfree.org/jfreechart/ (accessed 6 April 2012).
15 See www.esri.com/software/arcgis/index.html (accessed 6 April
2012).
16 See http://grass.osgeo.org/wiki/GRASS and Python (accessed
6 April 2012).
17 See www.saga-gis.org/ (accessed 6 April 2012).
18 See http://r-spatial.sourceforge.net/.
19 See www.opengeospatial.org/standards/gml (accessed 6 April
2012).
20 See http://code.google.com/apis/kml/documentation/ (accessed
6 April 2012).

1 See www.vensim.com/ (accessed 6 April 2012).
2 See www.iseesystems.com/softwares/Education/StellaSoftware
.aspx (accessed 6 April 2012).
3 See www.simulistics.com/ (accessed 6 April 2012).
4 See www.mathworks.com (accessed 6 April 2012).
5 See http://ccl.northwestern.edu/netlogo/ (accessed 6 April 2012).
6 See http://pcraster.geo.uu.nl/ (accessed 6 April 2012).
7 See http://pcraster.geo.uu.nl/support/courses/pcraster-python/
(accessed 6 April 2012).
8 See www.scipy.org/ (accessed 6 April 2012).
9 See http://eclipse.org/modeling/emf/ (accessed 6 April 2012).