Distribution-based sensitivity analysis from a generic input-output sample

摘要: Abstract In a previous paper we introduced distribution-based method for Global Sensitivity Analysis (GSA), called PAWN, which uses cumulative distribution functions of model outputs to assess their sensitivity the model's uncertain input factors. Over last three years, PAWN has been employed in environmental modelling field as useful alternative or complement more established variance-based methods. However, major limitation up now was need tailored sampling strategy approximate indices. Furthermore, this required tuning parameters whose optimal choice rather unclear. paper, present an approximation procedure that tackles both issues and makes applicable generic sample inputs while requiring only one parameter. The new implementation therefore allows user estimate indices complementary metrics multi-method GSA applications without additional computational cost.