Abstract
Several iterative algorithms are proposed to improve the histogram and variogram inference in the framework of the multigaussian model. The starting point is the variogram obtained after a traditional normal score transform. The subsequent step consists in simulating many sets of gaussian values with this variogram at the data locations, so that the ranking of the original values is honored. The expected gaussian transformation and the expected variogram are computed by an averaging operation over the simulated datasets. The variogram model is then updated and the procedure is repeated until convergence. Such an iterative algorithm can adapt to the case of tied data and despike the histogram. Two additional issues are also examined, referred to the modeling of the empirical transformation function and to the optimal pair weighting when computing the sample variogram.
Similar content being viewed by others
Acknowledgments.
The authors would like to acknowledge the Chair in Ore Reserve Evaluation at the University of Chile and the sponsoring by Codelco-Chile for supporting this research.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Emery, X., Ortiz, J. Histogram and variogram inference in the multigaussian model. Stoch Environ Res Ris Assess 19, 48–58 (2005). https://doi.org/10.1007/s00477-004-0205-5
Issue Date:
DOI: https://doi.org/10.1007/s00477-004-0205-5