Uncertainty Quantification Methodologies for Climate Model Data with Discontinuities

Abstract

Uncertainty quantification in climate models is challenged by the sparsity and bifurcative character of the available climate data. To circumvent these challenges we propose a methodology that employs Bayesian inference to locate discontinuities in the model output, followed by an efficient propagation of uncertain quantities using spectral expansions of random parameters/fields. Stochastic emulators are used to assess the performance of the proposed approach.

Date
Jul 14, 2010
Location
Pittsburgh, PA