Content uploaded by Vojtech Janousek
Author content
All content in this area was uploaded by Vojtech Janousek on Jul 02, 2015
Content may be subject to copyright.
TECHNICAL NOTE
Interpretation of Whole-rock Geochemical
Data in Igneous Geochemistry: Introducing
Geochemical Data Toolkit (GCDkit)
VOJTE
ˇCH JANOUS
ˇEK
1,2
*, C. M. FARROW
3
AND
VOJTE
ˇCH ERBAN
1
1
CZECH GEOLOGICAL SURVEY, KLA
´ROV 3/131, 118 21 PRAGUE 1, CZECH REPUBLIC
2
INSTITUTE OF PETROLOGY AND STRUCTURAL GEOLOGY, CHARLES UNIVERSITY, ALBERTOV 6,
128 43 PRAGUE 2, CZECH REPUBLIC
3
UNIVERSITY OF GLASGOW COMPUTING SERVICE, GLASGOW G12 8QQ, UK
RECEIVED NOVEMBER 3, 2004; ACCEPTED FEBRUARY 16, 2006
ADVANCE ACCESS PUBLICATION MARCH 22, 2006
Geochemical Data Toolkit (GCDkit) is a program for handling
and recalculation of geochemical data from igneous and metamorphic
rocks. It is built using the Windows version of R, which provides
a flexible and comprehensive language and environment for data
analysis and graphics. GCDkit was designed to eliminate routine
and tedious operations involving large collections of whole-rock data
and, at the same time, provide access to the wealth of statistical
functions built into R. Data management tools include import and
export of data files in a number of formats, data editing, searching,
grouping and generation of subsets. Included are a variety of calcu-
lation and normative schemes, for instance CIPW and Mesonorm,
as are the common geochemical graphs (e.g. binary and ternary
graphs, Harker plots, spider plots, and several dozens of classifica-
tion and geotectonic discrimination diagrams). The graphical output
is publication ready but can be further retouched if required. The
system can be further expanded by means of plug-in modules that
provide specialist applications. GCDkit is available as Free
Software under the terms of the Free Software Foundation’s GNU
General Public License and can be downloaded from http://www.
gla.ac.uk/gcdkit. The product is actively maintained and updated to
provide additional functionality; Unix/Linux and Mac OS versions
are being developed.
KEY WORDS: igneous rocks; geochemistry; norms; graphs; software; MS
Windows; R language
INTRODUCTION
The challenge: geochemical recalculations
and plotting
The need to quantify aspects of the world in which we live
is inherent in contemporary scientific thinking. Such an
approach leads inevitably to a flood of numerical
data, which have to be analysed by an appropriate soft-
ware tool.
The interpretation of whole-rock geochemical data is
not different, as it requires complex and time-consuming
calculations, which on PC compatible computers are
mainly performed using dedicated, stand-alone programs
(e.g. NewPet—Clarke et al., 1994; IgPet—Carr, 1995;
MinPet—Richard, 1995; PetroGraph—Petrelli et al., 2005).
An alternative preferred by many researchers are spread-
sheets, sometimes aided by special macros (Sidder, 1994;
Su et al., 2003). Most of these otherwise useful programs
are, however, not well suited to management and inter-
pretation of larger datasets. In addition, the usability
of many of these tools is hindered by the fact that it
is often difficult to determine exactly which algorithm
has been employed, because documentation is not
detailed enough and it is not a common practice to
make the source code available. Even if it is, any
modifications or additions to the original program can
*Corresponding author. Telephone þ420 251085308. Fax: þ420
251818748. E-mail: janousek@cgu.cz
The Author 2006. Published by Oxford University Press. All
rights reserved. For Permissions, please e-mail: journals.permissions@
oxfordjournals.org
JOURNAL OF PETROLOGY VOLUME 47 NUMBER 6 PAGES 1255–1259 2006 doi:10.1093/petrology/egl013
be tedious, or impossible for legal reasons. Concerning
spreadsheets, they have low efficiency for repeated
tasks, poor quality of graphic output, as well as
limited protection of the primary data. For more com-
plicated calculations the worksheet soon becomes too
complex and prone to errors.
The solution: Geochemical Data
Toolkit (GCDkit)
To address the major drawbacks of the existing geochem-
ical software, we have developed a new package called
Geochemical Data Toolkit for R (GCDkit). GCDkit is a program
for management and recalculation of whole-rock
geochemical data for igneous and metamorphic rocks.
It is built using the open-source (freeware) R language,
version for Windows, which in itself provides a flexible
and comprehensive environment for efficient data pro-
cessing, visualization and statistical analysis (Ihaka &
Gentleman, 1996; Grunsky, 2002; R Development
Core Team, 2005).
GCDkit provides:
(1) a graphical user interface aiding the user with pull-
down menus and dialogue boxes (Fig. 1);
(2) user-friendly front-end to the powerful statistical and
graphical functions of R;
(3) core routines for effortless import, modification,
searching, subsetting, classification and output of the
geochemical data;
(4) calculation and plotting tools designed specifically for
igneous geochemists;
(5) flexible and high-level graphical functions with an
output into widely used graphical formats;
(6) no licensing problems, as it is distributed free via the
WWW.
PROGRAM DESCRIPTION
Data management
The main area of application for GCDkit is the interpreta-
tion of larger datasets. Accordingly, great emphasis has
Fig. 1. Typical screen display of R with embedded GCDkit.
1256
JOURNAL OF PETROLOGY VOLUME 47 NUMBER 6 JUNE 2006
been put on data management tools (loading and saving
data files, data editing, together with searching and
generation of subsets).
GCDkit can read plain text files, including the variants
represented by the NewPet (*.ROC) and PetroGraph
(*.PEG) files as well as the outputs from WWW-based
databases such as GEOROC (http://georoc.mpch-mainz.
gwdg.de/georoc) and PETDB (http://www.petdb.org).
Data can be imported from Microsoft Excel (*.XLS),
Access (*.MDB), or dBase (*.DBF) formats, the latter
being the native format for IgPet (Carr, 1995) or MinPet
(Richard, 1995) packages. As an alternative, the data can
be pasted, via the clipboard, from any Windows-based
software, including popular spreadsheets. In any case
the data are freeform, which means that the columns
can be given in an arbitrary order and missing values
are allowed.
Grouping of data is a key part of data management,
controlling the output of calculations and the graphical
routines. Analyses may be grouped on the basis of various
criteria including sample attributes, so-called labels
(locality, rock type, etc.), numerical variable, cluster
analysis or selected classification diagram (e.g. total
alkali–silica). For greatest flexibility queries may be logic-
ally combined, allowing, for example, selection based on
rock type and contents of particular elements. Interactive
selection from classification diagrams is also supported.
Calculations
New variables can be calculated by mathematical expres-
sions including constants, brackets, arithmetic operators
and the whole spectrum of R functions. Statistical func-
tions include descriptive statistics, box-and-whiskers and
correlation plots or multivariate methods. Also imple-
mented are a variety of calculation schemes used in igne-
ous geochemistry, for instance CIPW norm, Catanorm,
Niggli’s cationic values, multicationic parameters of
De La Roche et al. (1980) and Debon & Le Fort (1983),
and improved Mesonorm for granitoid rocks (Mielke &
Winkler, 1979). Given major-element analyses of a rock
and its main mineral constituents, the best approximation
of the modal composition, using either unconstrained or
constrained least-squares methods (Albare
`de, 1995), may
be calculated.
The results of all calculations can be appended to the
current dataset for further processing (e.g. CIPW-
normative albite, quartz and orthoclase to be plotted on
a Ab–Qz–Or ternary diagram) or copied to a clipboard,
saved or exported to HTML, DBF, XLS, MDB or text files.
Plotting
The main strength of GCDkit is arguably the wealth of
built-in publication-quality plots that can be exported
into a number of data formats (including PostScript,
WMF, JPG, PNG and BMP). The available graphs
involve user-defined binary, ternary and multiple binary
diagrams (such as Harker plots) as well as a wide palette
of classification and geotectonic discrimination diagrams.
Three variables can be displayed in pseudo three-
dimensional plots, or on binary diagrams in which the
size and colour of the plotted symbols correspond to the
third variable (Fig. 2a).
All plots are publication ready but most can be
customized with the included plot editing facilities. The
system permits zooming and scaling of the diagrams,
editing of the text, font, size and colour of the titles and
axis labels, and of the colour, size and symbol for the data
points, together with colour, type and width of the lines.
Moreover, there is a function for interactive labelling of
individual analyses (typically by sample names but other
labels can be specified). The graph templates can be used
also as a basis for classification. Put simply, the classifica-
tion algorithm looks for the name of the polygon within
the diagram, into which the rock analysis falls according
to its x–ycoordinates.
GCDkit offers normalization of trace-element data by
any chosen standard and the ability to generate normal-
ized multielement diagrams, otherwise called ‘spider
plots’. One special feature are the ‘spider boxplots’, in
which no individual patterns are drawn. Instead, the
statistical distribution of each element is portrayed by a
box-and-whiskers plot of normalized concentrations
(Fig. 2b). The standard compositions used for normaliza-
tion can be readily edited and new schemes added. For
visualization of larger datasets there are contour and
frequency plots (Fig. 2c).
Expandability
The open architecture and simplicity mean that users will
readily be able to further enhance the capabilities of the
system, by modifying existing functions or creating new
ones. The system incorporates a plug-in mechanism
whereby new calculation or plotting code is automatically
loaded and made accessible through the menu system.
Included with the current version are plug-in modules
to calculate saturation temperatures of accessory minerals
( Janous
ˇek, 2006) as well as for interpretation of Sr–Nd
isotopic data (Fig. 2d).
CONCLUDING REMARKS
GCDkit will run on any IBM PC compatible computer
with Windows 98/ME/2000/NT/XP (but latest versions
are strongly recommended). The program with cor-
responding documentation is available free of charge at
http://www.gla.ac.uk/gcdkit and is easy to install.
1257
JANOUS
ˇEK et al. TECHNICAL NOTE
Our mission is to develop and release a platform-
independent system, for Windows, Unix/Linux and
Macintosh (System X). In the shorter-term perspective,
new plug-ins for direct and inverse modelling of major
petrogenetic processes in igneous geochemistry are under
development.
The whole system, which is modular and straight-
forward to modify, provides a practical solution to
the needs of petrologists and geochemists. If current
modules do not meet requirements, or new techniques
evolve, it is hoped that the open source availability of
the code will encourage its onward development by
users. In this context any feedback (bug reports,
suggestions for further development, and pieces of
contributed code or even ready plug-ins) would be
highly appreciated.
Fig. 2. Examples of diagrams generated by GCDkit. (a) SiO
2
–MgO plot for granitoids of the Variscan Central Bohemian Pluton, Czech Republic,
with size and grey shade of the plotting symbols corresponding to the K
2
O contents of individual samples. The data are from Janous
ˇek et al. (2000,
and unpublished). (b) Average upper crust normalized (Taylor & McLennan, 1985) ‘spider boxplots’ for felsic Variscan granulites from the
Moldanubian Unit, Bohemian Massif ( Janous
ˇek et al., 2004). (See text for explanation.) (c) SiO
2
–K
2
O plot with the discrimination boundaries
between the tholeiitic, calc-alkaline, high-K calc-alkaline and shoshonitic rocks of Peccerillo & Taylor (1976). Contoured grey fields are for all
available analyses from Moldanubian granulites and data points are for the Lis
ˇov Massif, Southern Bohemia ( Janous
ˇek et al., 2006). (d) Two-stage
Nd isotope development diagram for felsic Moldanubian granulites. DM, Depleted Mantle evolution lines after Goldstein et al. (1984) and Liew &
Hofmann (1988); CHUR, Chondritic Uniform Reservoir. The extra tick marks on axes show two-stage Nd model ages (x-axis) and initial
e
Nd
values ( y-axis).
1258
JOURNAL OF PETROLOGY VOLUME 47 NUMBER 6 JUNE 2006
ACKNOWLEDGEMENTS
We are indebted to Jakub S
ˇmı
´d for much of the ternary
plotting function, the Core Team and the whole com-
munity for their sterling efforts in continuous developing
of R—GCDkit is often merely an interface to their func-
tions. Thanks are specifically due to Friedrich Leisch
(Technische Universita
¨t Wien) and Duncan Murdoch
(University of Western Ontario) for their invaluable
advice. Last but not least, we thank beta testers (in
particular Pavel Hanz
ˇl and Bernhard Humer) for their
courage and patience, and the students in Prague, Brno
and Salzburg for asking nasty questions. This paper and
the program itself has benefited from extensive testing,
detailed comments and words of encouragement of
J.-F. Moyen. Also, the reviews by R. Giere, D. Geist
and T. Barry have helped us to improve aspects of the
software. Most of the coding of GCDkit originated during
the research stay of V.J. at the Institute of Mineralogy,
University of Salzburg, in the framework of the FWF
Project 15133-GEO (to Fritz Finger). Support from the
Czech Grant Agency (Postdoctoral Grant 205/97/P113)
and the Czech Geological Survey (Internal Project 3314)
is gratefully acknowledged.
REFERENCES
Albare
`de, F. (1995). Introduction to Geochemical Modeling. Cambridge:
Cambridge University Press, 543 pp.
Carr, M. (1995). Program IgPet. Somerset, NJ: Terra Softa.
Clarke, D., Mengel, F., Coish, R. A. & Kosinowski, M. H. F. (1994).
NewPet for DOS, version 94.01.07. St. John’s: Department of Earth
Sciences, Memorial University of Newfoundland.
De La Roche, H., Leterrier, J., Grandclaude, P. & Marchal, M. (1980).
A classification of volcanic and plutonic rocks using R
1
R
2
-
diagram and major element analyses—its relationships with current
nomenclature. Chemical Geology 29, 183–210.
Debon, F. & Le Fort, P. (1983). A chemical–mineralogical classification
of common plutonic rocks and associations. Transactions of the Royal
Society of Edinburgh, Earth Sciences 73, 135–149.
Goldstein, S. L., O’Nions, R. K. & Hamilton, P. J. (1984). A Sm–Nd
isotopic study of atmospheric dusts and particulates from major river
systems. Earth and Planetary Science Letters 70, 221–236.
Grunsky, E. C. (2002). R: a data analysis and statistical programming
environment—an emerging tool for the geosciences. Computers &
Geosciences 28, 1219–1222.
Ihaka, R. & Gentleman, R. (1996). R: a language for data analysis and
graphics. Journal of Computational and Graphical Statistics 5, 299–344.
Janous
ˇek, V. (2006). Saturnin, R language script for application of
accessory-mineral saturation models in igneous geochemistry.
Geologica Carpathica 57, (in press).
Janous
ˇek, V., Bowes, D. R., Rogers, G., Farrow, C. M. & Jelı
´nek, E.
(2000). Modelling diverse processes in the petrogenesis of a composite
batholith: the Central Bohemian Pluton, Central European
Hercynides. Journal of Petrology 41, 511–543.
Janous
ˇek, V., Finger, F., Roberts, M. P., Fry
´da, J., Pin, C. & Dolejs
ˇ,D.
(2004). Deciphering petrogenesis of deeply buried granites: whole-
rock geochemical constraints on the origin of largely undepleted
felsic granulites from the Moldanubian Zone of the Bohemian
Massif. Transactions of the Royal Society of Edinburgh, Earth Sciences
95, 141–159.
Janous
ˇek, V., Gerdes, A., Vra
´na, S., Finger, F., Erban, V., Friedl, G. &
Braithwaite, C. J. R. (2006). Low-pressure granulites of the Lis
ˇov
Massif, southern Bohemia: Vise
´an metamorphism of late Devonian
plutonic arc rocks. Journal of Petrology 47, 705–744.
Liew, T. C. & Hofmann, A. W. (1988). Precambrian crustal com-
ponents, plutonic associations, plate environment of the Hercynian
Fold Belt of Central Europe: indications from a Nd and Sr isotopic
study. Contributions to Mineralogy and Petrology 98, 129–138.
Mielke, P. & Winkler, H. G. F. (1979). Eine bessere Berechnung der
Mesonorm fu
¨r granitische Gesteine. Neues Jahrbuch fu
¨r Mineralogie,
Monatshefte 471–480.
Peccerillo, A. & Taylor, S. R. (1976). Geochemistry of Eocene
calc-alkaline volcanic rocks from the Kastamonu area, Northern
Turkey. Contributions to Mineralogy and Petrology 58, 63–81.
Petrelli, M., Poli, G., Perugini, D. & Peccerillo, A. (2005). PetroGraph:
a new software to visualize, model, and present geochemical data in
igneous petrology. Geochemistry, Geophysics, Geosystems 6, 1–15.
R Development Core Team (2005). R: a language and environment
for statistical computing. Vienna: R Foundation for Statistical
Computing. http://www.R-project.org.
Richard, L. R. (1995). MinPet: Mineralogical and Petrological Data Processing
System, Version 2.02. Que
´bec: MinPet Geological Software.
Sidder, G. B. (1994). Petro.calc.plot, Microsoft Excel macros to aid
petrologic interpretation. Computers & Geosciences 20, 1041–1061.
Su, Y. J., Langmuir, C. H. & Asimow, P. D. (2003). PetroPlot: a plotting
and data management tool set for Microsoft Excel. Geochemistry,
Geophysics, Geosystems 4, 1–14.
Taylor, S. R. & McLennan, S. M. (1985). The Continental Crust: its
Composition and Evolution. Blackwell: Oxford, 312 pp.
1259
JANOUS
ˇEK et al. TECHNICAL NOTE
