Methods for Uncertainty and Sensitivity Analysis. Review and recommendations for implementation in Ecolego

In this paper we review several uncertainty and sensitivity methods available in the literature and provide recommendations for implementing some of these methods in Ecolego. Predictive models used in the assessment of protection of the environment from ionising radiation present some extreme characteristics that make uncertainty and sensitivity analysis a non-trivial task. Those characteristics constrain the number and type of methods that are useful. The uncertainty methods selected in this work are based in Monte Carlo techniques or in other words they are probabilistic methods. These are found superior to deterministic methods in the context of radiological risk assessments and environmental protection. The Simple Monte Carlo sampling and the Latin Hypercube sampling were the sampling approaches selected to make uncertainty analysis. In respect to sensitivity analysis, the global methods were preferred over the deterministic or local methods. Global sensitivity analyses (GSA) are sampled-based analyses and again the Simple Monte Carlo and the Latin Hypercube techniques are necessary to allow their implementation in Ecolego. MatLab codes of the recommended methods were delivered 2004, in a CD to the Swedish Radiation Protection Authority (SSI).