• covariogram
• Kriging

The sensitivity of the Kriging-based response surface (surrogate model) to co-variogram law characteristics is investigated. The reference data are obtained through RANS simulations with two turbulence models. The Ordinary Kriging (OK), which is probably the most widely used form of Kriging, where the anisotropy it is not considered, is used. The impact of the Covariogram model on the results of the surrogate model (Kriging) is evaluated for two models of covariogram (Von Karman model, cubic spline model).
The Kriging-response surface pdfs for the two covariogram models are com- pared with gPC-response surface pdfs, using the same number of interpolations points in the first time and with number of interpolations points reduced for Kriging in the second time.The Kullback-Leiber divergence, the relative error, the most probable error, and the maximum error values are estimated to study the accuracy of the method.