• I have provided geostatistical consulting services to the mining industry since 1991 as a sole proprietor under the business license, Isaaks & Co. I am a member of AusIMM and qualify as a competent person for the purposes of ore resource/reserve audits and due diligence studies. For a list of previous project work, click here. For more information, please contact me at click here.
• I've been working on updating SAGE2001. The new version is called SAGE2005. One of the additions will be 3d graphics. One of the most difficult things users seem to have is to understand the orientation of dipping and plunging anisotropies. So SAGE2005 will provide 3d graphics of nested ellipsoids. You will be able to rotate and slice the ellipsoids through the axes. This will really help the user understand the orientation of the anisotropies relative to the sampling grid.
• Some users prefer copy protection via a sentinel or dongle rather than the current software and code numbers currently used with SAGE2001. If you would like a dongle version of SAGE2001, please let me know as it is available now.
• SAGE2005 will also have a new robust fitter for fitting a model to your sample variogram data. For example, SAGE2005 have an option to fit the model using robust techniques as follows:
  • Fit a variogram model using least squares. (This is what SAGE2001 does).
  • Calculate all of the residuals (Model values - sample variogram values).
  • Down weight the largest residuals using the Tukey Biweight Algorithm.
  • Refit the model. Extreme sample variogram values are downweighted and will have very little if any influence on the fit.
  • Repeat the previous 4 steps until a predefined convergence criterion is met.

This approach solves the problem of having a few extreme (probably outliers) sample variogram values over influence the fit.

• I have been successfull in developing a prototype of SAGE2001 for modeling the linear model of co-regionalization. I call it SAGE3X. It's quite amazing. You can feed SAGE3X as many as 37 directional sample variograms for a primary variable, 37 for the secondary variable, and 37 cross variograms, press a button, and in a few seconds, SAGE3X will provide you with the complete linear model of coregionalization. Let me know if you are interested in SAGE3X. I haven't decided whether to add it to SAGE2005 --- it all depends on your interest level.

 

Edward Isaaks