Model behaviour should be analysed to understand their dependence on inputs, and the confidence in their outputs. Sensitivity analysis focuses on understanding the dependence of the model outputs on the inputs, including the interdependence between model inputs. This is an important part of model development as well as testing existing model structures on new datasets. Uncertainty analysis extends this to consider the propagation of input uncertainty through the model, as well as the impact of model structure on confidence in the predictions. A key part of such methods is efficient sampling of parameter space and measuring convergence to ensure an adequate sample size is used. In addition, surrogate modelling approaches offer a means to analyse complex models for which runtime prevents effective analysis of the full model. Papers are invited that explore the development, testing and/or application of tools for analysing model behaviour.
Key topics: Sensitivity analysis, Uncertainty, Surrogate models