A simple method for the evaluation of the uncertainty in the predictions of a Lagrangian marine radionuclide transport model

Abstract

A method is proposed to assign an error bar to the concentrations predicted in water and seabed sediments by a Lagrangian radionuclide transport model for the marine environment. The method is based upon an analogy with radioactive counting statistics in a radiation detector, due to the stochastic nature of radioactive decay and turbulent mixing. However, it cannot be used to assess deterministic errors of the models, as those related to model parameters for instance. The method has been illustrated with a transport model of the northern Atlantic Ocean, previously tested by comparing model outputs with radionuclide measurements, released from European nuclear fuel reprocessing plants, in water and sediments at different locations and times. Time-series of calculated radionuclide concentrations in water and sediments at several locations, with the corresponding error bars, are provided as examples of the application of the method. In addition, some spatial distributions of errors are also shown. The methodology is simple, seems to be robust and can also be used to evaluate the number of particles required in a Lagrangian simulation to have a given precision level in the results.