Kriging In Mathematica

posted 2 Jun 2011, 01:47 by João Carneiro

Building your own code instead of just using some already tried and tested software sometimes leads o discover things that never crossed your mind: while rebuilding the Kriging workflow for Ordinary and Simple Kriging (and building hopefully more efficient and faster code) I stumbled upon a nasty fact of life: Negative Kriging Variances. These occur when kriging weights are very large and the kriging matrix is badly conditioned.

As such, there is a need to implement a kind of mathematical debugging report for the current kriging functions. I still hope to have these itches solved in the next few days and to finally release the first notebook.

Porosity kriging results (left) in Mathematica  end estimation variance (right). Estimation preformed on a 100x100 regular grid.