Sensitivity analysis in model calibration: GSA-GLUE approach

A new approach is presented applicable in framework of model calibration to observed data. The approach consists of a combination of the Generalized L ikelihood Uncertainty Estimation technique (GLUE) and Global Sensitivity An alysis (GSA). The method is based on multiple model evaluations. The GSA is a quantitative, model independent approach and is based on estimating the fractional contribution of each input factor to the variance of the model o utput, also accounting for interaction terms. In GLUE, the model runs are c lassified according to a likelihood measure, conditioning each run to obser vations. In calibration procedures, strong interaction is observed between model parameters, due to model over-parameterization. The use of likelihood measures allows an estimate of the posterior joint pdf of parameters. By p erforming a GSA to the likelihood measure, input factors mainly driving mod el runs with good fit to data are identified. Moreover GSA allows highlight ing the basic features of the interaction structure. Any other tool subsequ ently adopted to represent in more detail the interaction structure, from c orrelation coefficients to principal Component Analysis to Bayesian network s to tree-structured density estimation, confirms the general features iden tified by GSA. (C) 2001 Elsevier Science B.V. All rights reserved.