AN OPTIMIZATION TECHNOLOGY FOR MORE PRECISE DETERMINATION OF A SOURCEFOR MODELING RADIOACTIVE RELEASES TO THE ATMOSPHERE

This paper deals with refining meteorological parameters and source te rm characteristics required for models of atmospheric dispersion and d eposition of radioactivity in the area adjacent to facilities affected by a nuclear accident. The proposed methodology involves the minimisa tion of an objective function that is the sum of weighted norms of dev iations between measured and model values. The measured values arc spa tially interpolated and the weights determined by the type of measurem ent and the precision of interpolation. It is demonstrated that the mi nimisation process can be improved if the extremum of the partial deri vative of the objective is first located. Consideration is given to a technique of initial approximation based on calculating regression fun ctions for the parameters to be refined using measured radioactivity. Coefficients for these functions are calculated by analysis of a multi tude of simulated situations. The minimisation process is adjusted to the given model and objective function. Also discussed is a heuristic method, using multiple random perturbations on the data. for assessing the validity of the solution. Based on these calculations confidence intervals are determined for the refined model parameters.