Geoscience Reference
In-Depth Information
The
R
interface with SAGA has been used by Brenning (2009) for integrating terrain analysis
and multispectral remote sensing in automatic rock glacier detection, using modern regression tech-
niques - the availability of many varied techniques in
R
permitted them to be evaluated rapidly.
Goetz et al. (2011) follow this up in integrating physical and empirical landslide models. In a paper
on geostatistical modelling of topography, Hengl et al. (2008) uses the interface between
R
and
SAGA to benefit from the strengths of both software components. Hengl et al. (2010) address the
associated problem of stream network uncertainty, when the stream networks are derived from inter-
polated elevation data, again using the interface between SAGA and
R
;
R
is also used extensively for
scripting SAGA. Tomislav Hengl is also very active in organising courses for field scientists using
open-source geospatial software, especially the GEOSTAT series,* run in the open-source spirit,
and now with a useful collection of video recordings.
Finally, it is worth noting that TerraLib is linked to
R
using
sp
classes in the
aRT
package, which
uses
R
as the computational front-end and TerraLib as a data store and middleware component.
†
14.5 FUTURE PROSPECTS
The relationships between GC and open-source software development reviewed earlier show prom-
ise for the future. GC continues to provide new impulses for innovative applications and deeper
insights into the nature of spatial data. Both the availability of spatial data and the pervasiveness
of geospatial applications, not least in location services, are increasing. Existing software compo-
nents are being used to support these advances and are being challenged to adapt to new platforms,
especially mobile platforms. Choices are being made between server and client-side functionalities,
often in settings in which developer communities and software companies find themselves over-
taken by the opportunities offered by technology and desired by users.
As an example, we have seen earlier that TerraLib has offered broad functionality. The current
version, 4.2.2, is well supported, but the TerraLib developers are moving to embrace much more of
the OSGeo community than in the past. It seems that the envisioned developments in TerraLib 5
will combine its unique software components with more OSGeo components as a stack.
Figure 14.7 is taken from a presentation
‡
by Karine Reis Ferreira and Pedro Ribeiro Andrade
and reflects some of the prospects being considered by the development team. These design choices
engage many more existing projects, leveraging the user and developer communities of these projects
and increasing opportunities for shared technology exploration and development. The push towards
representing, querying and analysing spatio-temporal data is of particular importance, given the
publication of Cressie and Wikle (2011). It will be of great interest to see how the broader TerraLib
community develops in coming years, also in its interactions with other open-source geospatial
communities, and how the
aRT
interface with
R
progresses.
Experience from
R
spatial has shown that the nurturing of communities of interest and intention
is of fundamental importance. One unresolved issue concerns channels for information exchange, in
which the aging mailing list technology is straining to keep afloat as newer users prefer hosted fora
to search for answers to what they understand to be their questions. I believe that mailing lists have a
hidden bonus, that is, that readers, if they are willing to do so, can read threads that are not relevant
to their current concerns, but which may offer insight that will increase productivity later on.
§
*
http://geostat-course.org/.
†
http://www.leg.ufpr.br/aRT.
‡
http://giv-wikis.uni-muenster.de/agp/pub/Main/SpatioTemporalDataInRWorkshop2011/ST-Data-in-R-2011_Reis-
Ferreira_STData_2011.pdf.
§
I read the development lists of
R
, GRASS, GDAL, GEOS and PROJ.4, among others, and many of the insights presented
in this chapter have matured from exchanges on these lists; administering the
R
-sig-geo list has been a most valuable
source of information.
