Cation exchange capacity is often regarded as the single best summary statistic for soil fertility. Years ago, I learned how to measure it myself, in a soil lab. I used compulsive barium exchange, whereby I knocked off all of the cations using barium, and then I somehow (I forget) got all the barium off, before measuring how much barium is in solution. It makes sense as a means of measuring CEC.
I'm now in a situation where I need hundreds of soil samples analyzed, primarily for CEC. The lab with the best price does the CEC measurements by IR spectroscopy.
My question: how can you measure the potential of a soil to exchange cations using IR spectroscopy? It's based on correlations, right? Under what conditions does this method break down? I assume the prediction error is non-constant over the multivariate space comprising soil properties. In what regions of this space should I expect greater bias and/or variance in the prediction, and what would the sign of such bias be?
Or, in simpler terms, when do compulsive barium exchange and IR spectroscopy agree and when do they disagree?