Surrogate modeling for sensitivity analysis of models with high-dimensional outputs

Abstract

Sensitivity analysis provides important information on how the input uncertainty impacts the system output uncertainty. Typically, sensitivity analysis entails large number of system evaluations. For expensive system models with high-dimensional outputs, direct adoption of such models for sensitivity analysis poses significant computational challenges. To address these challenges, an efficient dimension reduction and surrogate based approach is proposed for efficient sensitivity analysis of expensive system models with high-dimensional outputs. As an example, the proposed approach is applied to investigate the sensitivity of peak water level over large coastal regions in San Francisco Bay with respect to the construction of levees at different counties under projected sea level rise.