LETTER TO THE EDITOR — Geostats provide one piece of estimation puzzle

This discussion is written as a continuation of the ongoing debate of resource estimation and, in particular, geostatistics.

I take issue with the endless commentary, arguments and opinion about the validity of kriging. If done correctly, kriging is a valuable tool that, when integrated with a multi-disciplined approach to resource estimation, can produce a sound estimate. However, geostatistics alone is only part of the process of resource estimation. A good resource estimation requires more than the mathematical manipulations and “wizardry” associated with kriging; it also requires common sense, logic and experienced professionals from several disciplines. Unfortunately, because of the endless arguments about geostatistics, a single aspect of a process, the beneficial contributions to the estimate from other disciplines are often ignored.

The numerous authors who have written letters about geostatistics to The Northern Miner make valid points, but they disregard three fundamental aspects of resource estimation, which, when ignored or done incorrectly, can never compensate for, or be corrected by, geostatistics or any other estimation technique. These fundamental aspects of resource estimation are: proper sampling of the deposit; detailed understanding of the deposit’s geology (and appropriate use of geology in resource modeling procedures); and the integrity of the database used to calculate the resource.

Careful and accurate sampling is critical. Nothing will compensate for incorrect samples. An infinite number of conditions exist which can influence sample results, including poor rock quality, mislocation, mislabeling, excessive water, weather conditions, geologically inappropriate samples, incorrect sample size, incorrect sample reduction procedures, improper analytical tests and improper handling. Often, the most critical components of an exploration program — core splitting, sample bucking and splitting — are performed by low-paid, inexperienced and poorly trained personnel. It seems incongruous that cutting-edge modeling technology, intricate mathematics and high-precision calculations are based on data that originate from such highly variable conditions with such potential for error. A large effort must be made to verify underlying sample data, which, if shown to be inaccurate, must be corrected or eliminated.

The understanding of geology alone is not sufficient, because this information must be used properly. Geology is used to define mineral domains, which may or may not be coincidental to geologic zones. Definition of mineral domains, a concept used by few in the industry, is key not only in predicting proper locations of mineralization, but in predicting the amount and tenor of mineralized material. Mineral domains are areas of mineral or metal concentrations that have uniform geologic statistical and geostatisitical conditions.

Without a good understanding of the geology on which to base interpreted mineral domains, grade estimation, through geostatistics or any other method, will rarely be successful, and then only under circumstances in which the geology is uniform or uncomplicated, or in which there is abundant data. Using the samples out of context can lead to poor conclusions. If a valid mineral model is defined and sufficient data exist, then the estimation method chosen, and many parameters used in estimation become less important. However, without a good mineral model and little data, the estimation method and parameters are important and, in some cases, critical.

Data integrity — the accuracy of information that describes a deposit — is critical. The construction of an accurate database is not common. Although it is a difficult and time-consuming project, the negative consequences associated with an inaccurate or inadequate database can be substantial.

Data sources can range from a disorganized pile of paper to electronic media, but all sources must be verified and evaluated.

An anecdote was told to me by a respected geostatictician. A company was performing a resource-reserve study and invited 11 well-known and capable independent geostaticticians to perform estimates for a well-drilled deposit. The highest estimate concluded that the deposit contained 50% more metal than calculated in the lowest estimate. There were differences in grade and tonnage, and on the location of the mineralization. All of the estimates were geostatistically justified by the data, yet only one estimate, if any, could have been accurate. Add to these variances potential error caused by a lack of sample integrity and use of geology, and it becomes painfully obvious that the risk tolerance of potential investors is exceeded in many instances.

Many of the problems of resource estimation have been blamed on geostatistics. It is an easy target. If not treated with respect or controlled in some manner, estimates can be erroneous. In fact, poor sample and data handling, a lack of geologic understanding and using analytic data in a vacuum are responsible for most of those cases of mis-estimation blamed on geostatistics.

Resource estimation is an art because subjectivity and opinion are part of the exercise. It is a science because it incorporates geologic principles, physics and chemistry. It is a mathematical exercise because statistics are needed for estimation, verification and elimination of as much subjectivity and opinion as possible.

Common sense and a firm handle on an integrated approach are critical. The point is, don’t rely entirely on a geologist, an engineer or a geostatictician for a resource estimate. Take an integrated approach to all these disciplines. Only a specialist should krige a deposit, but a general practitioner should integrate the elements of resource estimation into a process.

Steve Ristorcelli

Mine Development Associates

Reno, Nev.