Ensemble simulations are generated to estimate uncertainty and variability in their underlying numerical models. In this cumulative thesis, we present techniques to visualize, evaluate and handle sensitivities involved when clustering ensemble data. Correlation analysis is improved by basing it on statistical coherent regions, automation and summarizing overview plots. Relations to initial simulation parameters are extracted by comparing ensemble members based on distributions of their multi-parameter values.
«
Ensemble simulations are generated to estimate uncertainty and variability in their underlying numerical models. In this cumulative thesis, we present techniques to visualize, evaluate and handle sensitivities involved when clustering ensemble data. Correlation analysis is improved by basing it on statistical coherent regions, automation and summarizing overview plots. Relations to initial simulation parameters are extracted by comparing ensemble members based on distributions of their multi-par...
»