Importance and variability in processes relevant to environmental tritium ingestion dose models

The Aiken List was devised in 1990 to help decide which transport processes should be investigated experimentally so as to derive the greatest improvement in performance of environmental tritium assessment models. Each process was rated high, medium and low on each of two criteria. These were ‘Importance’, which rated processes by how much each contributed to ingestion doses, and ‘State of Modelling’, which rated the adequacy of the knowledge base on which models were built. Ratings, though unanimous, were, nevertheless, qualitative and subjective opinions. This paper describes how we have tried to quantify the ratings. To do this, we use, as measures of ‘Importance’, sensitivities of predicted ingestion doses to changes in values of parameters in mathematical descriptions of individual processes. Measures of ‘Modelling Status’ were taken from a recently completed BIOMOVS study of HTO transport model performance and based either on by how much predicted transport by individual processes differed amongst participating modellers or by the variety of different ways that modellers chose to describe individual processes. The tritium transport model UFOTRI was used, and because environmental transport of HTO varies according to the weather at and after release time, sensitivities were measured in a sample of all conditions likely to arise in central Europe. With HTO released from a point source and maximum ingestion dose to the most exposed individual as an end-point, sensitivities to selected processes were negligible. No single or small group of processes emerged as more important than any other so that the ‘Importance’ criterion had a limited use for deciding on research priorities. Those would be determined largely by the ‘State of the Art’ results from BIOMOVS. These conclusions are linked to the source and end point defined. They would not apply to any other combination. A different ‘Importance’ rating in the Aiken List is appropriate for each model application.