Applied Mineral Exploration
Discussion and research relevant to mineral exploration.

Home

Rapid fluid inclusion data for exploration (decrepitation)

Newest Topics

Viewpoints

Bicycles

Software used in Baro-acoustic decrepitation presentation and interpretation.

• Gnuplot - A very capable plotting program. Although it is driven from the command line, this makes it very easy to generate plots using a scripting language rather than having to manually generate each of many plots. In addition to direct display on the monitor, Gnuplot can generate its output in many different image formats and png format files are used for all the plots in the web pages here. (jpg format images are a very poor choice for graph plots as its compression algorithm generates artifacts,  seen as random spots on the image.)

• Perl - This scripting language is very comprehensive and is used to automate the generation of gnuplot control files from the analytical data files.

• Gimp - The Gnu image manipulation program is used to add additional annotation to the gnuplot image plots.

• Scidavis - A very complete data plotting and presentation program with comprehensive statistical functions. This is a GUI driven program which could be used in place of gnuplot. (QTIplot is a very similar program. Both grew from the same origin and diverged under different development teams.) This program can perform de-convolution of the decrepitation graphs back into a group of individual populations which sum to the observed data curve. Various population types are built in, but gaussian distributions are not. Because the decrepitation data is best fitted by multiple overlapping skewed gaussian distributions, mathematics to add this capability was added to Scidavis. This can be done because Scidavis has a python scripting interface, and the gaussian function was written in python and added as a module. The maths for the gaussian distribution requires the error function, and a python module called erfmodule.py, originally from SunPro, was used.

With the added python module, Scidavis was used for the first of the deconvolution plots. It does work, but can be tricky to drive and is critically dependent on the start parameters. The default start parameters fail to lead to convergence, so fitting must be done manually to allow more appropriate starting parameters to be set. In addition, it will by default allow the component populations to go negative, which is  unrealistic. This behavior has to be stopped with appropriate choice of settings before the curve fit is started. Normally, the inbuilt "automatic-fit" option automatically plots the individual component curves and the fitted curve as well as the original data curve. But when using manual fitting, the system no longer plots the component curves. Additional python code was written to read in the curve parameters and add these component curves to the final plot. This scripting also allows automated title and legend addition so the plots are of presentation standard. Although the software does eventually perform reliable and realistic curve fits, it is hard to drive and sometimes requires numerous attempts.  Scidavis was used for the curve fitting interpretation of the Brusson mine, Italy and the Malanjkhand mine, India. A much more convenient curve fitting program was subsequently found - fityk (see below).

• Python - An object oriented programming language required to build additional modules for Scidavis.

• Fityk - A very capable curve fitting program, frequently used in all sorts of scientific spectral work, to derive the constituent peaks/populations of complex spectra. It has an astonishing range of statistical populations inbuilt (not including skewed gaussian, although this can be added easily) and is driven from a very good GUI interface. It easily performs curve fitting to the decrepitation data, with only rare intervention required for particularly unusual data plots. This program has superseded the use of Scidavis and its Python scripts, although it does require slightly more manual image annotation to tidy up the overall plots to presentation standard.
  

• Grace plot - A GUI driven graphing and data presentation package with very comprehensive capabilities. The user interface is slightly unusual, but very functional. This is one of the few programs which can easily make X-Y plots with Z information represented by a feature of the plotted point, such as size and colour. This program was used to generate the X-Y plot for the data from Brusson, Italy.