With reference to “Kriged Variances Fatally Flawed (September, 1992), I want to draw the author’s attention to several facts of which he is evidently not aware:
1. Effectively, geostatistics originated in South Africa more than 40 years ago on the basis of applying classical statistical models and incorporating observed spatially related correlation levels.
2. There have been, over the years, many mathematical statisticians involved in geostatistics and particularly in recent years; these were or are quite capable of judging any “fatally flawed” departures in geostatistical models from statistical theory.
3. Throughout the development of geostatistical models for improved ore valuations and their error variances, the theory has been subjected to practical follow-up tests on many occasions. This has particularly been the case in South Africa where massive data bases are available and repeated borehole valuations can, for example, be simulated and compared with the grades indicated by the complete data bases. I can refer the author to two recent papers covering this aspect and which, by themselves, refute most, if not all, of his accusations.
First, there is Krige, D.G. and Assibey-Bonsu, W., “New Developments in Borehole Valuations of New Gold Mines and Undeveloped Sections of Existing Mines,” J.S. Afr. Inst. M.M., Vol. 92, No. 3, Mar. 1992, pp.71-77.
Second, he can read Krige, D.G. and Assibey-Bonsu, W., “The Relative Rating of Macro Kriging and Other Techniques as Applied to Borehole Valuations of New Gold Mines,” Ed. Y.C. Kim, Port City Press, Baltimore, Maryland, April, 1992.
D.G. Krige,
Honorary Professorial
Research Fellow,
University of Witwatersrand,
South Africa
I am amused by the new controversy (September, 1992) surrounding geostatistics and Merks’s claim to perfection. I don’t care how anyone estimates grades; all methods require some empiricism or “tweeking” to match reality. We have audited or used just about every known modelling technique (and a few unknowns) on dozens of operating mines. To reconcile the models with production, all required some adjustment. Champigny et al, and Merks are, I suggest, as fallible as the rest of us. However, I score Champigny 1, Merks 0.
Our industry is not so naive as to accept or reject theses based on magazine articles and arguments intelligible only to statisticians. In the end, the best performer will prevail. Roll the fractals.
Doug Stewart,
Mining Consultant,
Snowden Associates,
Australia
Be the first to comment on "PROFESSOR KRIGE RESPONDS"