Determination of Uncertainty Model of a Particle-Reflection-Distribution

The modelling of plasma-wall interactions (PWIs) depends on distributions describing the angle and energy distribution of particles scattered at the first wall of fusion devices.Most PWI codes rely on extensive tables based on data from reflection simulations, employing a Monte Carlo method.At first glance, the uncertainty distribution of the data should be assumed Gaussian.However, in order to obtain the resulting particle distribution, the reflected ions are counted within angle sections of the upper hemisphere, which hints to a Poisson uncertainty distribution.In this paper, we let Bayesian model comparison decide which uncertainty model should be taken.