Dr Isobel Clark FSAIMM FIMMM CEng MMSA(QP)
My major area of investigative study has been in the application of the "Theory of Regionalised Variables" in ore reserve estimation and other appropriate fields. This subject, now more commonly known as Geostatistics, is still a relatively new one. In earlier days all papers were written in French or in such highly mathematical language that the potential user was unable to benefit from them. I have endeavoured to develop basic courses and to publish papers which were comprehensible to mining engineers and geologists. Constant contact with the Minerals Industry has enabled me to include a fair amount of reality in both teaching and publications.
Attempts to put geostatistics to work in practical situations have raised many problems which are not covered in the literature. I have, therefore, spent most of my 'research' time in studying methods of applying geostatistics in: non-ideal situations; extending applications to three dimensional deposits; evaluating the errors inherent in approximations at various stages of an analysis; reducing computation times; investigating problems directly related to production mining such as the so-called 'conditional bias'; and so on.
Some of my research has been oriented towards adapting and/or applying existing statistical techniques to mining problems. Investigations here have been in two main directions. Distribution theory, including the analysis of mixtures of distributions, can be used to produce estimates of the global characteristics of ore grade distributions. Mixtures of 'populations' usually signal mixtures of geological environments. Combinations of mixed population analysis and indicator geostatistics have produced successful estimation methods which combine both geology and geostatistics. I have also made some contribution in the updating of the application of the three parameter lognormal, sparking further research by interested workers on generalisations of the lognormal distribution. Regression techniques have been used in two major areas: trend surface analysis and the investigation of the regression effect or 'conditional bias' in production mining.
In the 1990s, together with Gavin Lind of Wits University, we studied the uses of indicator kriging in identifying voids and pillars in abandoned coal mines. With Daniel Limpitlaw (also Wits), we have been looking at an effective marriage of GIS and geostatistical techniques.
Non-mining applications of geostatistical methods have also been studied. For a four year period in the late 1980's, I was involved in characterisation of a potential site for a high-level nuclear waste repository in the Pala Duro Basin, Texas. The major orientation of this study was towards the quantification of the uncertainty associated with the likely travel paths of radionuclides should the repository be breached. The study site is underlain by two major aquifers, so research was mainly on the geostatistical estimation and simulation of the hydrogeological variables normally associated with groundwater flows. This type of study has, since, been extended to the investigation of toxic chemical pollutants in groundwater in New Jersey and other similar applications.
Another environmental application which has been studied in detail is that of atmospheric pollution from coal fired power stations in the Transvaal. Working together with ESKOM, we have investigated the possibility of reducing particulate and gaseous emissions from power stations by controlling the actual mining of the coal at the face. Geostatistical methods of estimation and simulation have been used to anticipate the likely contravention of government regulations with regard to particulate emissions. Ongoing research is being carried out into the less obvious impacts of the beneficiation of coal before delivery to the power station. For example, many of the power stations in the Transvaal rely on the inherent sulphur content of the coal to assist in the precipitation of ash and stone dust before venting through the stack. 'Washing' coal reduces the amount of sulphur and can --- in certain circumstances -- lead to significantly higher particulate emissions from the stack.
Back to Top of Page Back to Main Resume Page Back to Home Page