|
|
|
GIS approach to statistical modelling for mineral deposits in the Singhbhum copper belt, Bihar, India, using geological and geophysical parameters
Host Rock
Lithology:
The copper sulphides associated with other minerals occur
mainly in the Dhanjori metavolcanics and their derivatives, and the feldspathic
schist/soda granite and metabasics, chlorite schist, sericite schist, mica
schist and quartzites of Singhbhum Group (Chaudhuri, et.al., 1998).
Statistical Model:
Two statistical techniques were employed
for the study. The first technique was a "Characteristic Analysis" developed by
Botbol (1971) and the second technique was an application of the multiple
regression analysis as described by Agterberg (1972). At various stages of the
work, different standard softwares for computer processing of the data, were
utilized and attempt was made to automate the whole process. The idea was to
develop a software which could be routinely used by any geoscientist, for any
area having known mineral deposits, to predict probability of occurrence of
deposits in the area of interest. For the present study a database was developed
using dBase V software.
An important aspect of the database is that any
number of groups of variables can be added in the database simply by
incorporating the variables with their code in the form of a table e.g.,
Geophysical and Geochemical variables can be incorporated (which are not at
present considered) by simply adding Geophysical and Geochemcial tables with
their coded variables.
Methodology:
Broadly the variables
selected were grouped into seven classes (Annexure-I)
- Lithology consisting of 18 variables like soda granite, chlorite-quartz
schist etc.
- Surface indication consisting of 4 variables like gossan, old workings etc.
- Structure consisting of 5 variables like foliation, lineation etc.
- Ore mineralogy consisting of 22 variables like chalcopyrite, pyrite etc.
- Host rock consisting of 41 variables like biotite schist, sheared
conglomerate etc.
- Mode of occurrence consisting of 13 variables like stringers, specks etc.
- Control of mineralization consisting of 19 variables like lithological and
structural, localized along axial plane of fold etc.
So a total of 122
variables were considered. It is assumed that all these variables are present
throughout the area and for the time being, are of equal importance. After
scanning the literature on Singhbhum Copper Belt, 27 locations were selected
from where the variables were considered for statistical treatment. The
following two statistical techniques have been adapted for the present work.
Characteristic Analysis:
In order to find out those variables most
representative of a particular mineral in a particular area or in other words
those variables which were always or nearly always found to be associated with
copper in the study area and to reduce the huge number of variables
characteristic analysis was carried out. The total number of variables were
brought down to 20 from the original 122.
The whole process, right from
creation of the database to ranking the variables was automated with the help of
two softwares, dBase V for Windows and SPSS statistical package. Programme was
written in dBase V for Windows which manipulated the data from the database
created earlier and transformed the data into binary form. The data matrix M1
thus created was called the presence-absence matrix. This matrix was then
automatically transferred to SPSS, where the matrix was manipulated to generate
the ranked variables. The 20 variables so produced are shown in Table-7.
The geology of the area selected for the study was scanned from the
geological map of Singhbhum Copper Belt, Fig.2. It was found that out of the 20
geological variables some were not mapped separately. For example
sericite-chlorite-quartz schist and quartz-chlorite-biotite schist were not
mapped as separate units but as one unit and so could not be differentiated. It
was mapped as quartz-biotite-chlorite-schist and this variable was considered.
Too the final analysis the variables were brought down to only eight (8).
- Quartz chlorite schist.
- Quartz biotite-chlorite schist.
- Gossan.
- Quartzite.
- Mica schist/phyllite.
- Foliation(S1), cleavage(S).
- Pucker lineation/Plunge of fold axis.
- Faults.
New variables were formed by combining the above 8
variables (Botbol, 1971; Vyshemirshy et al., 1971) into 28 such new variables by
taking 2 variables at a time. So the total variables used for the analysis was
36. The reason for combining the geological variables is to allow for
interaction between different variables which is so common in geology.
|
|
|