AnsweredAssumed Answered

How to keep a residual influence on the prediction

Question asked by geonetadmin on Dec 31, 2010
Original User: FMIN

I have some problems with the residual values on the ordinary kriging prediction with the 100% error setting. The nugget is very large more than 70% of the variance. that because i have two scale. One very large with low variance and the other to smal for my sampling but with a great variance. The residual values are not autocorrelated, sure... but i need to evaluate what is the influence of these residual values on the total prediction when these values are over a threshold.

Also,  if i use a kriging indicator to map the probability over the threshold and i do a product between this probability map and the average of the value over the threshold and i add these map with predicted map, i imagine that i get a good estimation of how a i have a stoff. 

I d'ont have any abilities whith disjunctive kriging or conditional simulation.